diff --git a/Makefile b/Makefile
index 35fa6bb6811f4588d283d36ee32a4e57b5d14d12..5493351ffd79472ee3536a8c0cca1118e8402ff3 100644
--- a/Makefile
+++ b/Makefile
@@ -61,6 +61,19 @@ UTIL_SUBDIRS = $(SRC_DIR)/utils/lineFileUtilities \
$(SRC_DIR)/utils/genomeFile
BUILT_OBJECTS = $(OBJ_DIR)/*.o
+# BUILT_OBJECTS = $(OBJ_DIR)/bedtools.o \
+# $(OBJ_DIR)/BamAncillary.o \
+# $(OBJ_DIR)/Fasta.o \
+# $(OBJ_DIR)/bedFile.o \
+# $(OBJ_DIR)/bedFilePE.o \
+# $(OBJ_DIR)/bedGraphFile.o \
+# $(OBJ_DIR)/chromsweep.o \
+# $(OBJ_DIR)/fileType.o \
+# $(OBJ_DIR)/gzstream.o \
+# $(OBJ_DIR)/sequenceUtils.o \
+# $(OBJ_DIR)/split.o \
+# $(OBJ_DIR)/intersectBed.o \
+# $(OBJ_DIR)/intersectMain.o \
all:
[ -d $(OBJ_DIR) ] || mkdir -p $(OBJ_DIR)
@@ -82,6 +95,7 @@ all:
done
@echo "- Building main bedtools binary."
+ gcc $(CXXFLAGS) -c src/bedtools.cpp -o obj/bedtools.o
@$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/bedtools $(BUILT_OBJECTS) -L$(UTIL_DIR)/BamTools/lib/ -lbamtools $(LIBS)
diff --git a/src/annotateBed/Makefile b/src/annotateBed/Makefile
index 6eb7f8cdc543c5ef5377031e5d3af1265593ed08..23bad93dc1a29a23adef5230d60aa257076bb3d3 100644
--- a/src/annotateBed/Makefile
+++ b/src/annotateBed/Makefile
@@ -18,14 +18,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= annotateBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/annotateBed/annotateMain.cpp b/src/annotateBed/annotateMain.cpp
index 07b2f198864ec187a16ffa35c1a7dc2967a69b0d..8ae9e98cdc2ece98023326a7319ecee5c2c1bea0 100644
--- a/src/annotateBed/annotateMain.cpp
+++ b/src/annotateBed/annotateMain.cpp
@@ -15,15 +15,15 @@
using namespace std;
// define the version
-#define PROGRAM_NAME "annotateBed"
+#define PROGRAM_NAME "bedtools annotate"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void annotate_help(void);
-int main(int argc, char* argv[]) {
+int annotate_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -56,7 +56,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) annotate_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -133,20 +133,18 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ annotate_help();
+ return 0;
}
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void annotate_help(void) {
- cerr << "Summary: Annotates the depth & breadth of coverage of features from multiple files" << endl;
+ cerr << "\nTool: bedtools annotate (aka annotateBed)" << endl;
+ cerr << "Summary: Annotates the depth & breadth of coverage of features from mult. files" << endl;
cerr << "\t on the intervals in -i." << endl << endl;
- cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -files FILE1 FILE2 .. FILEn" << endl << endl;
+ cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -files FILE1 FILE2..FILEn" << endl << endl;
cerr << "Options: " << endl;
diff --git a/src/bamToBed/Makefile b/src/bamToBed/Makefile
index 65741ae8f4e4e5bef22438b5b2f45746e22ab5fa..5cf20cd2433cd4388f15944feb5af55ab19f5224 100644
--- a/src/bamToBed/Makefile
+++ b/src/bamToBed/Makefile
@@ -25,14 +25,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= bamToBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/bamToBed/bamToBed.cpp b/src/bamToBed/bamToBed.cpp
index 4a4c56c662a5d953191698534e74185808326baf..bcae2c7a1eddac02d747a092f35a2dd2dcb1a35d 100644
--- a/src/bamToBed/bamToBed.cpp
+++ b/src/bamToBed/bamToBed.cpp
@@ -26,14 +26,14 @@ using namespace std;
// define our program name
-#define PROGRAM_NAME "bamToBed"
+#define PROGRAM_NAME "bedtools bamtobed"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void bamtobed_help(void);
void ConvertBamToBed(const string &bamFile, bool useEditDistance, const string &bamTag,
bool writeBed12, bool obeySplits, const string &color,
@@ -52,10 +52,10 @@ void ParseCigarBed12(const vector<CigarOp> &cigar, vector<int> &blockStarts,
vector<int> &blockEnds, int &alignmentEnd);
string BuildCigarString(const vector<CigarOp> &cigar);
-bool IsCorrectMappingForBEDPE (const BamAlignment &bam);
+bool bamtobed_IsCorrectMappingForBEDPE (const BamAlignment &bam);
-int main(int argc, char* argv[]) {
+int bamtobed_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -88,7 +88,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) bamtobed_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -178,16 +178,15 @@ int main(int argc, char* argv[]) {
ConvertBamToBedpe(bamFile, useEditDistance); // BEDPE
}
else {
- ShowHelp();
+ bamtobed_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void bamtobed_help(void) {
+
+ cerr << "\nTool: bedtools bamtobed (aka bamToBed)" << endl;
cerr << "Summary: Converts BAM alignments to BED6 or BEDPE format." << endl << endl;
@@ -605,7 +604,7 @@ void PrintBedPE(const BamAlignment &bam1, const BamAlignment &bam2, const RefVec
// deprecated.
-bool IsCorrectMappingForBEDPE (const BamAlignment &bam) {
+bool bamtobed_IsCorrectMappingForBEDPE (const BamAlignment &bam) {
if ( (bam.RefID == bam.MateRefID) && (bam.InsertSize > 0) ) {
return true;
diff --git a/src/bed12ToBed6/Makefile b/src/bed12ToBed6/Makefile
index e2e1288341db4dac6b027061d9815b3b35bf9b80..fd6fac5e51183c52f8a9d31406232042f8058fc6 100644
--- a/src/bed12ToBed6/Makefile
+++ b/src/bed12ToBed6/Makefile
@@ -18,15 +18,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= bed12ToBed6
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/bed12ToBed6/bed12ToBed6.cpp b/src/bed12ToBed6/bed12ToBed6.cpp
index d6aae22adb5e3166f5018036de178c4283fbb38e..fc1884e931cdaad92ed53ebf320b6f3ebae42298 100644
--- a/src/bed12ToBed6/bed12ToBed6.cpp
+++ b/src/bed12ToBed6/bed12ToBed6.cpp
@@ -22,21 +22,21 @@ using namespace std;
// define our program name
-#define PROGRAM_NAME "bed12ToBed6"
+#define PROGRAM_NAME "bedtools bed12tobed6"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void bed12tobed6_help(void);
void DetermineBedInput(BedFile *bed);
void ProcessBed(istream &bedInput, BedFile *bed);
bool addBlockNums = false;
-int main(int argc, char* argv[]) {
+int bed12tobed6_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -54,7 +54,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) bed12tobed6_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -88,17 +88,16 @@ int main(int argc, char* argv[]) {
DetermineBedInput(bed);
}
else {
- ShowHelp();
+ bed12tobed6_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void bed12tobed6_help(void) {
+ cerr << "\nTool: bedtools bed12tobed6 (aka bed12ToBed6)" << endl;
+
cerr << "Summary: Splits BED12 features into discrete BED6 features." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed12>" << endl << endl;
diff --git a/src/bedToBam/Makefile b/src/bedToBam/Makefile
index f6e18c0c1df051d8b33d214db1cbdf8d5ef389aa..0666e314475bd1a63060f704e3338932ac7e62f3 100644
--- a/src/bedToBam/Makefile
+++ b/src/bedToBam/Makefile
@@ -25,15 +25,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= bedToBam
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/bedToBam/bedToBam.cpp b/src/bedToBam/bedToBam.cpp
index a689533cb1322ed1d571243dde87e25c2f0e09f0..2334622656717cfefa9792cd4d2131ac2312aca0 100644
--- a/src/bedToBam/bedToBam.cpp
+++ b/src/bedToBam/bedToBam.cpp
@@ -29,22 +29,22 @@ using namespace std;
// define our program name
-#define PROGRAM_NAME "bedToBam"
+#define PROGRAM_NAME "bedtools bedtobam"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void bedtobam_help(void);
void ProcessBed(BedFile *bed, GenomeFile *genome, bool isBED12, int mapQual, bool uncompressedBam);
void ConvertBedToBam(const BED &bed, BamAlignment &bam, map<string, int> &chromToId, bool isBED12, int mapQual, int lineNum);
void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header, map<string, int> &chromToInt);
-int reg2bin(int beg, int end);
+int bedtobam_reg2bin(int beg, int end);
-int main(int argc, char* argv[]) {
+int bedtobam_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -70,7 +70,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) bedtobam_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -131,16 +131,15 @@ int main(int argc, char* argv[]) {
ProcessBed(bed, genome, isBED12, mapQual, uncompressedBam);
}
else {
- ShowHelp();
+ bedtobam_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void bedtobam_help(void) {
+
+ cerr << "\nTool: bedtools bedtobam (aka bedToBam)" << endl;
cerr << "Summary: Converts feature records to BAM format." << endl << endl;
@@ -154,10 +153,10 @@ void ShowHelp(void) {
cerr << "\t-bed12\t" << "The BED file is in BED12 format. The BAM CIGAR" << endl;
cerr << "\t\tstring will reflect BED \"blocks\"." << endl << endl;
- cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
+ cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default writes compressed BAM." << endl << endl;
cerr << "Notes: " << endl;
- cerr << "\t(1) BED files must be at least BED4 to be amenable to BAM (needs name field)." << endl << endl;
+ cerr << "\t(1) BED files must be at least BED4 to create BAM (needs name field)." << endl << endl;
// end the program here
@@ -211,7 +210,7 @@ void ConvertBedToBam(const BED &bed, BamAlignment &bam, map<string, int, std::le
bam.Name = bed.name;
bam.Position = bed.start;
- bam.Bin = reg2bin(bed.start, bed.end);
+ bam.Bin = bedtobam_reg2bin(bed.start, bed.end);
// hard-code the sequence and qualities.
int bedLength = bed.end - bed.start;
@@ -341,7 +340,7 @@ void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header,
/* Taken directly from the SAMTools spec
calculate bin given an alignment in [beg,end) (zero-based, half-close, half-open) */
-int reg2bin(int beg, int end) {
+int bedtobam_reg2bin(int beg, int end) {
--end;
if (beg>>14 == end>>14) return ((1<<15)-1)/7 + (beg>>14);
if (beg>>17 == end>>17) return ((1<<12)-1)/7 + (beg>>17);
diff --git a/src/bedToIgv/Makefile b/src/bedToIgv/Makefile
index 3ad616b9c5019e9de5699753067ef2b5105c8ea8..50ea2daebf58d5e0b6f7229cd2dc4071e5830941 100644
--- a/src/bedToIgv/Makefile
+++ b/src/bedToIgv/Makefile
@@ -24,15 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= bedToIgv
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/bedToIgv/bedToIgv.cpp b/src/bedToIgv/bedToIgv.cpp
index 9697d8085301235fb08c83e06a1fa7933787cff8..562062d31d41331987ffdaade75521ccae848c23 100644
--- a/src/bedToIgv/bedToIgv.cpp
+++ b/src/bedToIgv/bedToIgv.cpp
@@ -22,13 +22,13 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "bedToIgv"
+#define PROGRAM_NAME "bedtools igv"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void bedtoigv_help(void);
void DetermineBedInput(BedFile *bed, string path, string sortType, string session,
bool collapse, bool useNames, string imageType, int slop);
@@ -36,7 +36,7 @@ void ProcessBed(istream &bedInput, BedFile *bed, string path, string sortType, s
bool collapse, bool useNames, string imageType, int slop);
-int main(int argc, char* argv[]) {
+int bedtoigv_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -62,7 +62,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) bedtoigv_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -139,16 +139,15 @@ int main(int argc, char* argv[]) {
DetermineBedInput(bed, imagePath, sortType, session, collapse, useNames, imageType, slop);
}
else {
- ShowHelp();
+ bedtoigv_help();
}
+ return 0;
}
-void ShowHelp(void) {
+void bedtoigv_help(void) {
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+ cerr << "\nTool: bedtools igv (aka bedToIgv)" << endl;
cerr << "Summary: Creates a batch script to create IGV images " << endl;
cerr << " at each interval defined in a BED/GFF/VCF file." << endl << endl;
@@ -182,9 +181,9 @@ void ShowHelp(void) {
cerr << "\t\tDefault is png." << endl << endl;
cerr << "Notes: " << endl;
- cerr << "\t(1) The resulting script is meant to be run from within the IGV GUI version 1.5 or later." << endl;
- cerr << "\t(2) Unless you use the -sess option, it is assumed that prior to running the script, " << endl;
- cerr << "\t\tyou have loaded the proper genome, tracks and data files." << endl << endl;
+ cerr << "\t(1) The resulting script is meant to be run from within IGV." << endl;
+ cerr << "\t(2) Unless you use the -sess option, it is assumed that prior to " << endl;
+ cerr << "\t\trunning the script, you've loaded the proper genome and tracks." << endl << endl;
// end the program here
diff --git a/src/bedpeToBam/Makefile b/src/bedpeToBam/Makefile
index b2f7d128400b9f8354d2dc5b394c3ffe424cd5ff..592d68e6de7547d6d8fcfbde28abae08052ab4b3 100644
--- a/src/bedpeToBam/Makefile
+++ b/src/bedpeToBam/Makefile
@@ -26,15 +26,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= bedpeToBam
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/bedpeToBam/bedpeToBam.cpp b/src/bedpeToBam/bedpeToBam.cpp
index 06e9af88453664534b241347c5e8f39811bfe548..374e3fa91829245e9c9a4c9b7dd8efb70a2d04a8 100644
--- a/src/bedpeToBam/bedpeToBam.cpp
+++ b/src/bedpeToBam/bedpeToBam.cpp
@@ -29,27 +29,23 @@ using namespace std;
// define our program name
-#define PROGRAM_NAME "bedpeToBam"
+#define PROGRAM_NAME "bedpetobam"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
-//ROYDEN
-//NEED TO ADD A IS TYPE12+ check or fail
-//END ROYDEN
-
// function declarations
-void ShowHelp(void);
+void bedpetobam_help(void);
void ProcessBedPE(BedFilePE *bedpe, GenomeFile *genome, int mapQual, bool uncompressedBam);
void ConvertBedPEToBam(const BEDPE &bedpe, BamAlignment &bam1,BamAlignment &bam2, map<string, int> &chromToId, int mapQual, int lineNum);
-void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header, map<string, int> &chromToInt);
-int reg2bin(int beg, int end);
+void bedpetobam_MakeBamHeader(const string &genomeFile, RefVector &refs, string &header, map<string, int> &chromToInt);
+int bedpetobam_reg2bin(int beg, int end);
-int main(int argc, char* argv[]) {
+int bedpetobam_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -75,7 +71,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) bedpetobam_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -133,16 +129,15 @@ int main(int argc, char* argv[]) {
ProcessBedPE(bedpe, genome, mapQual, uncompressedBam);
}
else {
- ShowHelp();
+ bedpetobam_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void bedpetobam_help(void) {
+
+ cerr << "\nTool: bedtools bedpetobam (aka bedpeToBam)" << endl;
cerr << "Summary: Converts feature records to BAM format." << endl << endl;
@@ -153,10 +148,10 @@ void ShowHelp(void) {
cerr << "\t-mapq\t" << "Set the mappinq quality for the BAM records." << endl;
cerr << "\t\t(INT) Default: 255" << endl << endl;
- cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
+ cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default writes compressed BAM." << endl << endl;
cerr << "Notes: " << endl;
- cerr << "\t(1) BED files must be at least BED4 to be amenable to BAM (needs name field)." << endl << endl;
+ cerr << "\t(1) BED files must be at least BED4 to create BAM (needs name field)." << endl << endl;
// end the program here
@@ -171,7 +166,7 @@ void ProcessBedPE(BedFilePE *bedpe, GenomeFile *genome, int mapQual, bool uncom
RefVector refs;
string bamHeader;
map<string, int, std::less<string> > chromToId;
- MakeBamHeader(genome->getGenomeFileName(), refs, bamHeader, chromToId);
+ bedpetobam_MakeBamHeader(genome->getGenomeFileName(), refs, bamHeader, chromToId);
// set compression mode
BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
@@ -218,10 +213,10 @@ void ConvertBedPEToBam(const BEDPE &bedpe, BamAlignment &bam1,BamAlignment &bam2
bam1.Name = bedpe.name;
bam1.Position = bedpe.start1;
- bam1.Bin = reg2bin(bedpe.start1, bedpe.end1);
+ bam1.Bin = bedpetobam_reg2bin(bedpe.start1, bedpe.end1);
bam2.Name = bedpe.name;
bam2.Position = bedpe.start2;
- bam2.Bin = reg2bin(bedpe.start2, bedpe.end2);
+ bam2.Bin = bedpetobam_reg2bin(bedpe.start2, bedpe.end2);
// hard-code the sequence and qualities.
int bedpeLength1 = bedpe.end1 - bedpe.start1;
@@ -294,7 +289,7 @@ void ConvertBedPEToBam(const BEDPE &bedpe, BamAlignment &bam1,BamAlignment &bam2
}
-void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header,
+void bedpetobam_MakeBamHeader(const string &genomeFile, RefVector &refs, string &header,
map<string, int, std::less<string> > &chromToId) {
// make a genome map of the genome file.
@@ -333,7 +328,7 @@ void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header,
/* Taken directly from the SAMTools spec
calculate bin given an alignment in [beg,end) (zero-based, half-close, half-open) */
-int reg2bin(int beg, int end) {
+int bedpetobam_reg2bin(int beg, int end) {
--end;
if (beg>>14 == end>>14) return ((1<<15)-1)/7 + (beg>>14);
if (beg>>17 == end>>17) return ((1<<12)-1)/7 + (beg>>17);
diff --git a/src/bedpeToBam/notes.txt b/src/bedpeToBam/notes.txt
new file mode 100644
index 0000000000000000000000000000000000000000..b7c406467c80a19c87eb2d5f54bab48424b398cf
--- /dev/null
+++ b/src/bedpeToBam/notes.txt
@@ -0,0 +1,44 @@
+Public Member Functions BamAlignment (void) constructor BamAlignment (const BamAlignment &other) copy constructor ~BamAlignment (void) destructor
+bool IsDuplicate (void) const
+bool IsFailedQC (void) const
+bool IsFirstMate (void) const
+bool IsMapped (void) const
+bool IsMateMapped (void) const
+bool IsMateReverseStrand (void) const
+bool IsPaired (void) const
+bool IsPrimaryAlignment (void) const
+bool IsProperPair (void) const
+bool IsReverseStrand (void) const
+bool IsSecondMate (void) const
+void SetIsDuplicate (bool ok) Sets value of "PCR duplicate" flag to ok.
+void SetIsFailedQC (bool ok) Sets "failed quality control" flag to ok.
+void SetIsFirstMate (bool ok) Sets "alignment is first mate" flag to ok.
+void SetIsMapped (bool ok) Sets "alignment is mapped" flag to ok.
+void SetIsMateMapped (bool ok) Sets "alignment's mate is mapped" flag to ok.
+void SetIsMateReverseStrand (bool ok) Sets "alignment's mate mapped to reverse strand" flag to ok.
+void SetIsPaired (bool ok) Sets "alignment part of paired-end read" flag to ok.
+void SetIsPrimaryAlignment (bool ok) Sets "position is primary alignment" flag to ok.
+void SetIsProperPair (bool ok) Sets "alignment is part of read that satisfied paired-end resolution" flag to ok.
+void SetIsReverseStrand (bool ok) Sets "alignment mapped to reverse strand" flag to ok.
+void SetIsSecondMate (bool ok) Sets "alignment is second mate on read" flag to ok.
+void SetIsMateUnmapped (bool ok) Complement of using SetIsMateMapped().
+void SetIsSecondaryAlignment (bool ok) Complement of using SetIsPrimaryAlignment().
+void SetIsUnmapped (bool ok) Complement of using SetIsMapped().
+bool AddTag (const std::string &tag, const std::string &type, const std::string &value) Adds a field with string data to the BAM tags.
+bool AddTag (const std::string &tag, const std::string &type, const uint32_t &value) Adds a field with unsigned integer data to the BAM tags.
+bool AddTag (const std::string &tag, const std::string &type, const int32_t &value) Adds a field with signed integer data to the BAM tags.
+bool AddTag (const std::string &tag, const std::string &type, const float &value) Adds a field with floating-point data to the BAM tags.
+bool EditTag (const std::string &tag, const std::string &type, const std::string &value) Edits a BAM tag field containing string data.
+bool EditTag (const std::string &tag, const std::string &type, const uint32_t &value) Edits a BAM tag field containing unsigned integer data.
+bool EditTag (const std::string &tag, const std::string &type, const int32_t &value) Edits a BAM tag field containing signed integer data.
+bool EditTag (const std::string &tag, const std::string &type, const float &value) Edits a BAM tag field containing floating-point data.
+bool GetTag (const std::string &tag, std::string &destination) const Retrieves the string value associated with a BAM tag.
+bool GetTag (const std::string &tag, uint32_t &destination) const Retrieves the unsigned integer value associated with a BAM tag.
+bool GetTag (const std::string &tag, int32_t &destination) const Retrieves the signed integer value associated with a BAM tag.
+bool GetTag (const std::string &tag, float &destination) const Retrieves the floating-point value associated with a BAM tag.
+bool GetTagType (const std::string &tag, char &type) const Retrieves the BAM tag type-code associated with requested tag name.
+bool GetEditDistance (uint32_t &editDistance) const Retrieves value of edit distance tag ("NM").
+bool GetReadGroup (std::string &readGroup) const Retrieves value of read group tag ("RG").
+bool RemoveTag (const std::string &tag) Removes field from BAM tags.
+bool BuildCharData (void) Populates alignment string fields (read name, bases, qualities, tag data).
+int GetEndPosition (bool usePadded=false, bool zeroBased=true) const Calculates alignment end position, based on starting position and CIGAR data.
\ No newline at end of file
diff --git a/src/bedtools.cpp b/src/bedtools.cpp
index a76ed33cc75dfbcd52b4b8929de6dc6d6b5a64d0..965da00718b52c3203fac3af27cf7b061613d4b6 100644
--- a/src/bedtools.cpp
+++ b/src/bedtools.cpp
@@ -21,51 +21,158 @@ using namespace std;
// define our program name
#define PROGRAM_NAME "bedtools"
+// colors for the term's menu
+#define RESET "\e[m"
+#define GREEN "\e[1;32m"
+#define BLUE "\e[1;34m"
+#define RED "\e[1;31m"
+
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
+int annotate_main(int argc, char* argv[]);//
+int bamtobed_main(int argc, char* argv[]);//
+int bed12tobed6_main(int argc, char* argv[]); //
+int bedtobam_main(int argc, char* argv[]);//
+int bedtoigv_main(int argc, char* argv[]);//
+int bedpetobam_main(int argc, char* argv[]);//
+int closest_main(int argc, char* argv[]); //
+int complement_main(int argc, char* argv[]);//
+int coverage_main(int argc, char* argv[]); //
+int fastafrombed_main(int argc, char* argv[]);//
+int flank_main(int argc, char* argv[]); //
+int genomecoverage_main(int argc, char* argv[]);//
+int getoverlap_main(int argc, char* argv[]);//
+int intersect_main(int argc, char* argv[]); //
+int links_main(int argc, char* argv[]);//
+int maskfastafrombed_main(int argc, char* argv[]);//
+int merge_main(int argc, char* argv[]); //
+int multibamcov_main(int argc, char* argv[]);//
+int multiintersect_main(int argc, char* argv[]);//
+int nuc_main(int argc, char* argv[]);//
+int pairtobed_main(int argc, char* argv[]);//
+int pairtopair_main(int argc, char* argv[]);//
+int shuffle_main(int argc, char* argv[]); //
+int slop_main(int argc, char* argv[]); //
+int sort_main(int argc, char* argv[]); //
+int subtract_main(int argc, char* argv[]); //
+int tagbam_main(int argc, char* argv[]);//
+int unionbedgraphs_main(int argc, char* argv[]);//
+int window_main(int argc, char* argv[]); //
-int intersect_main(int argc, char* argv[]);
-int coverage_main(int argc, char* argv[]);
-int merge_main(int argc, char* argv[]);
-int substract_main(int argc, char* argv[]);
-
-static int help()
+int bedtools_help()
{
- cerr << "\n";
- cerr << "Program: " << PROGRAM_NAME << " (Tools for genome arithmetic.)\n";
- cerr << "Version: " << "2.15.5" << "\n";
- cerr << "Author: " << "Aaron Quinlan (aaronquinlan@gmail.com)" << "\n\n";
- cerr << "Usage: bedtools <tool> [options]\n";
-
- cerr << "\n-Genome arithmetic tools:\n";
- cerr << " intersect Find overlapping/intersecting intervals b/w two files.\n";
- cerr << " coverage Compute the coverage of one set of intervals over another.\n";
- cerr << " merge Combine overlapping or nearby intervals into a single interval.\n";
- cerr << " subtract Remove intervals based on overlaps b/w two files.\n";
-
- cerr << "\n-Format conversion tools:\n";
- cerr << " intersect Find overlapping/intersecting intervals b/w two files.\n";
-
- cerr << "\n-Fasta tools:\n";
- cerr << " intersect Find overlapping/intersecting intervals b/w two files.\n";
+ cout << "\n";
+ cout << PROGRAM_NAME << ": flexible tools for genome arithmetic and analysis.\n";
+ cout << "Version: " << "2.15.5" << "\n";
+ cout << "Authors: " << "Aaron Quinlan and others (see THANKS)" << "\n\n";
+ cout << "Usage: bedtools <tool> [options]\n";
+
+ cout << "\nGenome arithmetic:" << endl;
+ cout << RED << " intersect " << RESET << "Find exactly overlapping intervals.\n";
+ cout << RED << " window " << RESET << "Find overlapping intervals within a \"window\".\n";
+ cout << RED << " closest " << RESET << "Find overlapping or nearby intervals.\n";
+ cout << RED << " coverage " << RESET << "Compute the coverage over defined intervals.\n";
+ cout << RED << " genomecov " << RESET << "Compute the coverage over an entire genome.\n";
+ cout << RED << " merge " << RESET << "Combine overlapping/nearby intervals into a single interval.\n";
+ cout << RED << " complement " << RESET << "Extract intervals _not_ represented by an interval file.\n";
+ cout << RED << " subtract " << RESET << "Remove intervals based on overlaps b/w two files.\n";
+ cout << RED << " slop " << RESET << "Adjust the size of intervals.\n";
+ cout << RED << " flank " << RESET << "Create new intervals from the flanks of existing intervals.\n";
+ cout << RED << " sort " << RESET << "Order the intervals in a file.\n";
+ cout << RED << " shuffle " << RESET << "Randomly redistrubute intervals in a genome.\n";
+ cout << RED << " annotate " << RESET << "Annotate coverage of features from multiple files.\n";
- cerr << "\n";
+ cout << "\nMulti-way file comparisons:" << endl;
+ cout << RED << " multiinter " << RESET << "Identifies common intervals among multiple interval files.\n";
+ cout << RED << " unionbedg " << RESET << "Combines coverage intervals from multiple BEDGRAPH files.\n";
+
+ cout << "\nPaired-end manipulation:" << endl;
+ cout << RED << " pairtobed " << RESET << "Find pairs that overlap intervals in various ways.\n";
+ cout << RED << " pairtopair " << RESET << "Find pairs that overlap other pairs in various ways.\n";
+
+ cout << "\nFormat conversion:\n";
+ cout << RED << " bamtobed " << RESET << "Convert BAM alignments to BED (& other) formats.\n";
+ cout << RED << " bedtobam " << RESET << "Convert intervals to BAM records.\n";
+ cout << RED << " bedpetobam " << RESET << "Convert BEDPE intervals to BAM records.\n";
+ cout << RED << " bed12tobed6 " << RESET << "Breaks BED12 intervals into discrete BED6 intervals.\n";
+
+ cout << "\nFasta manipulation:\n";
+ cout << RED << " getfasta " << RESET << "Use intervals to extract sequences from a FASTA file.\n";
+ cout << RED << " maskfasta " << RESET << "Use intervals to mask sequences from a FASTA file.\n";
+ cout << RED << " nuc " << RESET << "Profile the nucleotide content of intervals in a FASTA file.\n";
+
+ cout << "\nBAM focused tools:\n";
+ cout << RED << " multicov " << RESET << "Counts coverage from multiple BAMs at specific intervals.\n";
+ cout << RED << " tag " << RESET << "Annotate BAM alignments based on overlaps with interval files.\n";
+
+ cout << "\nMiscellaneous tools:\n";
+ cout << RED << " overlap " << RESET << "Computes the amount of overlap from two intervals.\n";
+ cout << RED << " igv " << RESET << "Create an IGV snapshot batch script.\n";
+ cout << RED << " links " << RESET << "Create a HTML page of links to UCSC locations.\n";
+
+ cout << "\nGeneral help:\n";
+ cout << RED << " faq " << RESET << "Frequently asked questions.\n";
+
+ cout << "\n";
return 1;
}
int main(int argc, char *argv[])
{
// make sure the user at least entered a sub_command
- if (argc < 2) return help();
+ if (argc < 2) return bedtools_help();
+
+ // genome arithmetic tools
+ if (strcmp(argv[1], "intersect") == 0) return intersect_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "window") == 0) return window_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "closest") == 0) return closest_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "coverage") == 0) return coverage_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "genomecov") == 0) return genomecoverage_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "merge") == 0) return merge_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "complement") == 0) return complement_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "subtract") == 0) return subtract_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "slop") == 0) return slop_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "flank") == 0) return flank_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "sort") == 0) return sort_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "shuffle") == 0) return shuffle_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "annotate") == 0) return annotate_main(argc-1, argv+1);
+
+ // Multi-way file comparisonstools
+ else if (strcmp(argv[1], "multiinter") == 0) return multiintersect_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "unionbedg") == 0) return unionbedgraphs_main(argc-1, argv+1);
+
+ // paired-end conversion tools
+ else if (strcmp(argv[1], "pairtobed") == 0) return pairtobed_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "pairtopair") == 0) return pairtopair_main(argc-1, argv+1);
+
+ // format conversion tools
+ else if (strcmp(argv[1], "bamtobed") == 0) return bamtobed_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "bedtobam") == 0) return bedtobam_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "bedpetobam") == 0) return bedpetobam_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "bed12tobed6") == 0) return bed12tobed6_main(argc-1, argv+1);
+
+ // BAM-specific tools
+ else if (strcmp(argv[1], "multicov") == 0) return multibamcov_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "tag") == 0) return tagbam_main(argc-1, argv+1);
+
+ // fasta tools
+ else if (strcmp(argv[1], "getfasta") == 0) return fastafrombed_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "maskfasta") == 0) return maskfastafrombed_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "nuc") == 0) return nuc_main(argc-1, argv+1);
+
+ // misc. tools
+ else if (strcmp(argv[1], "overlap") == 0) return getoverlap_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "igv") == 0) return bedtoigv_main(argc-1, argv+1);
+ else if (strcmp(argv[1], "links") == 0) return links_main(argc-1, argv+1);
+
+ // help
+ else if (strcmp(argv[1], "-h") == 0) return bedtools_help();
+ else if (strcmp(argv[1], "--help") == 0) return bedtools_help();
- // spawn the proper tool.
- if (strcmp(argv[1], "intersect") == 0) return intersect_main(argc-1, argv+1);
- else if (strcmp(argv[1], "coverage") == 0) return coverage_main(argc-1, argv+1);
- else if (strcmp(argv[1], "merge") == 0) return merge_main(argc-1, argv+1);
- else if (strcmp(argv[1], "subtract") == 0) return subtract_main(argc-1, argv+1);
+ // unknown
else {
- fprintf(stderr, "[main] unrecognized command '%s'\n", argv[1]);
+ cerr << "error: unrecognized command: " << argv[1] << endl << endl;
return 1;
}
return 0;
diff --git a/src/closestBed/Makefile b/src/closestBed/Makefile
index 00cfd889d6af6d88b6eeb88826859f9999cc38b6..859bfcf118cee4a81741f10e3729ac3445cc072a 100644
--- a/src/closestBed/Makefile
+++ b/src/closestBed/Makefile
@@ -16,14 +16,10 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= closestBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/closestBed/closestMain.cpp b/src/closestBed/closestMain.cpp
index 0128af24fc6edbb4fa20558e56a1b541774be9b1..e4f1b7a51def47bf2a8095da7c278e53b7c356b5 100644
--- a/src/closestBed/closestMain.cpp
+++ b/src/closestBed/closestMain.cpp
@@ -15,15 +15,15 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "closestBed"
+#define PROGRAM_NAME "bedtools closest"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void closest_help(void);
-int main(int argc, char* argv[]) {
+int closest_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -58,7 +58,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) closest_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -144,19 +144,16 @@ int main(int argc, char* argv[]) {
diffStrand, tieMode, reportDistance,
signDistance, strandedDistMode, ignoreOverlaps, printHeader);
delete bc;
- return 0;
}
else {
- ShowHelp();
+ closest_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Authors: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
- cerr << "\t Erik Arner, Riken" << endl << endl;
+void closest_help(void) {
+
+ cerr << "\nTool: bedtools closest (aka closestBed)" << endl;
cerr << "Summary: For each feature in A, finds the closest " << endl;
cerr << "\t feature (upstream or downstream) in B." << endl << endl;
@@ -164,12 +161,12 @@ void ShowHelp(void) {
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <bed/gff/vcf> -b <bed/gff/vcf>" << endl << endl;
cerr << "Options: " << endl;
- cerr << "\t-s\t" << "Require same strandedness. That is, find the closest feature in B" << endl;
- cerr << "\t\tthat overlaps A on the _same_ strand." << endl;
+ cerr << "\t-s\t" << "Req. same strandedness. That is, find the closest feature in" << endl;
+ cerr << "\t\tB that overlaps A on the _same_ strand." << endl;
cerr << "\t\t- By default, overlaps are reported without respect to strand." << endl << endl;
- cerr << "\t-S\t" << "Require opposite strandedness. That is, find the closest feature in B" << endl;
- cerr << "\t\tthat overlaps A on the _opposite_ strand." << endl;
+ cerr << "\t-S\t" << "Req. opposite strandedness. That is, find the closest feature" << endl;
+ cerr << "\t\tin B that overlaps A on the _opposite_ strand." << endl;
cerr << "\t\t- By default, overlaps are reported without respect to strand." << endl << endl;
cerr << "\t-d\t" << "In addition to the closest feature in B, " << endl;
@@ -178,17 +175,19 @@ void ShowHelp(void) {
cerr << "\t-D\t" << "Like -d, report the closest feature in B, and its distance to A" << endl;
cerr << "\t\tas an extra column. Unlike -d, use negative distances to report" << endl;
- cerr << "\t\tupstream features. You must specify which orientation defines \"upstream\"." << endl;
- cerr << "\t\tThe options are:" << endl;
+ cerr << "\t\tupstream features." << endl;
+ cerr << "\t\tThe options for defining which orientation is \"upstream\" are:" << endl;
cerr << "\t\t- \"ref\" Report distance with respect to the reference genome. " << endl;
cerr << "\t\t B features with a lower (start, stop) are upstream" << endl;
cerr << "\t\t- \"a\" Report distance with respect to A." << endl;
- cerr << "\t\t When A is on the - strand, \"upstream\" means B has a higher (start,stop)." << endl;
+ cerr << "\t\t When A is on the - strand, \"upstream\" means B has a" << endl;
+ cerr << "\t\t higher (start,stop)." << endl;
cerr << "\t\t- \"b\" Report distance with respect to B." << endl;
- cerr << "\t\t When B is on the - strand, \"upstream\" means A has a higher (start,stop)." << endl << endl;
+ cerr << "\t\t When B is on the - strand, \"upstream\" means A has a" << endl;
+ cerr << "\t\t higher (start,stop)." << endl << endl;
- cerr << "\t-io\t" << "Ignore features in B that overlap A. That is, we want close, but " << endl;
- cerr << "\t\tnot touching features only." << endl << endl;
+ cerr << "\t-io\t" << "Ignore features in B that overlap A. That is, we want close," << endl;
+ cerr << "\t\tyet not touching features only." << endl << endl;
cerr << "\t-t\t" << "How ties for closest feature are handled. This occurs when two" << endl;
cerr << "\t\tfeatures in B have exactly the same \"closeness\" with A." << endl;
diff --git a/src/complementBed/Makefile b/src/complementBed/Makefile
index 2f542063e48f88b4aad550cbc8136f0739709f41..bc56b4dd173a35d04af8654f56178b74cca5df0c 100644
--- a/src/complementBed/Makefile
+++ b/src/complementBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= complementBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/complementBed/complementMain.cpp b/src/complementBed/complementMain.cpp
index 7eb34d5f5f3745eac2b9340ddc258464afc431af..98b1a786b7cdd5d3cf97640edb5f9839b4f9e42a 100644
--- a/src/complementBed/complementMain.cpp
+++ b/src/complementBed/complementMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "complementBed"
+#define PROGRAM_NAME "bedtools complement"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void complement_help(void);
-int main(int argc, char* argv[]) {
+int complement_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -45,7 +45,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) complement_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -73,25 +73,23 @@ int main(int argc, char* argv[]) {
// make sure we have both input files
if (!haveBed || !haveGenome) {
- cerr << endl << "*****" << endl << "*****ERROR: Need -i BED file and -g Genome file. " << endl << "*****" << endl;
+ cerr << endl << "*****" << endl << "*****ERROR: Need -i BED file and -g genome file. " << endl << "*****" << endl;
showHelp = true;
}
if (!showHelp) {
BedComplement *bc = new BedComplement(bedFile, genomeFile);
bc->ComplementBed();
- return 0;
}
else {
- ShowHelp();
+ complement_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void complement_help(void) {
+ cerr << "\nTool: bedtools complement (aka complementBed)" << endl;
+
cerr << "Summary: Returns the base pair complement of a feature file." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome>" << endl << endl;
diff --git a/src/coverageBed/Makefile b/src/coverageBed/Makefile
index 48ccb37fa903f4c79e2ee69aa957237b09295744..ea5d777bc0680972e8cbdc8e58f3fbca9e501b55 100644
--- a/src/coverageBed/Makefile
+++ b/src/coverageBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= coverageBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/coverageBed/coverageMain.cpp b/src/coverageBed/coverageMain.cpp
index 6a01fdc6521298b59954f404eef43a8d711efacc..a0b839f8ed0b7ef4c9da612fae306b5d4e76f5a4 100644
--- a/src/coverageBed/coverageMain.cpp
+++ b/src/coverageBed/coverageMain.cpp
@@ -15,15 +15,15 @@
using namespace std;
// define the version
-#define PROGRAM_NAME "coverageBed"
+#define PROGRAM_NAME "bedtools coverage"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void coverage_help(void);
-int main(int argc, char* argv[]) {
+int coverage_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -55,7 +55,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) coverage_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -123,19 +123,17 @@ int main(int argc, char* argv[]) {
BedCoverage *bg = new BedCoverage(bedAFile, bedBFile, sameStrand, diffStrand,
writeHistogram, bamInput, obeySplits, eachBase, countsOnly);
delete bg;
- return 0;
}
else {
- ShowHelp();
+ coverage_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void coverage_help(void) {
+ cerr << "\nTool: bedtools coverage (aka coverageBed)" << endl;
+
cerr << "Summary: Returns the depth and breadth of coverage of features from A" << endl;
cerr << "\t on the intervals in B." << endl << endl;
@@ -149,8 +147,8 @@ void ShowHelp(void) {
cerr << "\t\toverlap B on the _same_ strand." << endl;
cerr << "\t\t- By default, overlaps are counted without respect to strand." << endl << endl;
- cerr << "\t-S\t" << "Require different strandedness. That is, only report hits in A that" << endl;
- cerr << "\t\toverlap B on the _opposite_ strand." << endl;
+ cerr << "\t-S\t" << "Require different strandedness. That is, only report hits in A" << endl;
+ cerr << "\t\tthat overlap B on the _opposite_ strand." << endl;
cerr << "\t\t- By default, overlaps are counted without respect to strand." << endl << endl;
cerr << "\t-hist\t" << "Report a histogram of coverage for each feature in B" << endl;
diff --git a/src/fastaFromBed/Makefile b/src/fastaFromBed/Makefile
index 3cd9e892da53c4c6fa6a91b9435ba037594657e2..701a660492978d67b8f3562f865d38cd1faf0905 100644
--- a/src/fastaFromBed/Makefile
+++ b/src/fastaFromBed/Makefile
@@ -25,14 +25,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= fastaFromBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/fastaFromBed/fastaFromBedMain.cpp b/src/fastaFromBed/fastaFromBedMain.cpp
index c31676d93228329dad44141c43b5d790aa8471d9..06bce7754e7e1af3a431a604f539857267ee8811 100644
--- a/src/fastaFromBed/fastaFromBedMain.cpp
+++ b/src/fastaFromBed/fastaFromBedMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "fastaFromBed"
+#define PROGRAM_NAME "bedtools getfasta"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void fastafrombed_help(void);
-int main(int argc, char* argv[]) {
+int fastafrombed_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -56,7 +56,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) fastafrombed_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -107,19 +107,16 @@ int main(int argc, char* argv[]) {
Bed2Fa *b2f = new Bed2Fa(useNameOnly, fastaDbFile, bedFile, fastaOutFile, useFasta, useStrand);
delete b2f;
-
- return 0;
}
else {
- ShowHelp();
+ fastafrombed_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void fastafrombed_help(void) {
+
+ cerr << "\nTool: bedtools getfasta (aka fastaFromBed)" << endl;
cerr << "Summary: Extract DNA sequences into a fasta file based on feature coordinates." << endl << endl;
@@ -134,12 +131,10 @@ void ShowHelp(void) {
cerr << "\t-tab\tWrite output in TAB delimited format." << endl;
cerr << "\t\t- Default is FASTA format." << endl << endl;
- cerr << "\t-s\tForce strandedness. If the feature occupies the antisense strand," << endl;
- cerr << "\t\tthe sequence will be reverse complemented." << endl;
+ cerr << "\t-s\tForce strandedness. If the feature occupies the antisense," << endl;
+ cerr << "\t\tstrand, the sequence will be reverse complemented." << endl;
cerr << "\t\t- By default, strand information is ignored." << endl << endl;
-
-
// end the program here
exit(1);
diff --git a/src/flankBed/Makefile b/src/flankBed/Makefile
index 1da2069305eca0bb9b6c64c40b3b5f3defc37aae..bc53291bf943fce1c92736ed3c066c1f4aed91d8 100644
--- a/src/flankBed/Makefile
+++ b/src/flankBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= flankBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/flankBed/flankBedMain.cpp b/src/flankBed/flankBedMain.cpp
index 062839e6a377298ce849bd209ea5d9e093df46f5..5600cec0acae200fc58c6b5ad8e3c14f70da69de 100644
--- a/src/flankBed/flankBedMain.cpp
+++ b/src/flankBed/flankBedMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "flankBed"
+#define PROGRAM_NAME "bedtools flank"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void flank_help(void);
-int main(int argc, char* argv[]) {
+int flank_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -54,7 +54,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) flank_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -137,15 +137,14 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ flank_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void flank_help(void) {
+
+ cerr << "\nTool: bedtools flank (aka flankBed)" << endl;
cerr << "Summary: Creates flanking interval(s) for each BED/GFF/VCF feature." << endl << endl;
@@ -155,10 +154,12 @@ void ShowHelp(void) {
cerr << "\t-b\t" << "Create flanking intervak using -b base pairs in each direction." << endl;
cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
- cerr << "\t-l\t" << "The number of base pairs that a flank should start from orig. start coordinate." << endl;
+ cerr << "\t-l\t" << "The number of base pairs that a flank should start from" << endl;
+ cerr << "\t\torig. start coordinate." << endl;
cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
- cerr << "\t-r\t" << "The number of base pairs that a flank should end from orig. end coordinate." << endl;
+ cerr << "\t-r\t" << "The number of base pairs that a flank should end from" << endl;
+ cerr << "\t\torig. end coordinate." << endl;
cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
cerr << "\t-s\t" << "Define -l and -r based on strand." << endl;
diff --git a/src/genomeCoverageBed/Makefile b/src/genomeCoverageBed/Makefile
index 4194fdb581eecaef89964490ddb8c4f146270593..a77257674c3ff69d7f0d348e85c0bbc197ed0c34 100644
--- a/src/genomeCoverageBed/Makefile
+++ b/src/genomeCoverageBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= genomeCoverageBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/genomeCoverageBed/genomeCoverageMain.cpp b/src/genomeCoverageBed/genomeCoverageMain.cpp
index a316e45acd1bb28ed0571695cf175652a3420d97..43cb0246f6ec85ffe00891fb7792cc92a724acdd 100644
--- a/src/genomeCoverageBed/genomeCoverageMain.cpp
+++ b/src/genomeCoverageBed/genomeCoverageMain.cpp
@@ -15,16 +15,16 @@ Licenced under the GNU General Public License 2.0 license.
using namespace std;
// define our program name
-#define PROGRAM_NAME "genomeCoverageBed"
+#define PROGRAM_NAME "bedtools genomecov"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void genomecoverage_help(void);
-int main(int argc, char* argv[]) {
+int genomecoverage_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -64,7 +64,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) genomecoverage_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -205,21 +205,17 @@ int main(int argc, char* argv[]) {
eachBaseZeroBased,
add_gb_track_line, gb_track_opts);
delete bc;
-
- return 0;
}
else {
- ShowHelp();
+ genomecoverage_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Authors: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
- cerr << " Assaf Gordon, CSHL" << endl << endl;
+void genomecoverage_help(void) {
+ cerr << "\nTool: bedtools genomecov (aka genomeCoverageBed)" << endl;
+
cerr << "Summary: Compute the coverage of a feature file among a genome." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome>" << endl << endl;
diff --git a/src/getOverlap/Makefile b/src/getOverlap/Makefile
index be9ff2faed3acf0d5ab32c2690d058ae0acb9526..6767447bcb3ba3dbd3900ed0721ee93616f32360 100644
--- a/src/getOverlap/Makefile
+++ b/src/getOverlap/Makefile
@@ -19,15 +19,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= getOverlap
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/getOverlap/getOverlap.cpp b/src/getOverlap/getOverlap.cpp
index 20bb597bffae49b32446fd9b789567f9f8ae336f..ec981a2db065ce1feca85c13ec9238e4527d8866 100644
--- a/src/getOverlap/getOverlap.cpp
+++ b/src/getOverlap/getOverlap.cpp
@@ -28,11 +28,11 @@ using namespace std;
// function declarations
-void ShowHelp(void);
+void getoverlap_help(void);
void DetermineInput(string &inFile, short &s1Col, short &e1Col, short &s2Col, short &e2Col);
void ComputeOverlaps(istream &input, short &s1Col, short &e1Col, short &s2Col, short &e2Col);
-int main(int argc, char* argv[]) {
+int getoverlap_main(int argc, char* argv[]) {
// input files
string inFile = "stdin";
@@ -53,7 +53,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) getoverlap_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -92,7 +92,7 @@ int main(int argc, char* argv[]) {
if (posColumns.size() != 4) {
cerr << endl << "*****" << endl << "*****ERROR: Please specify 4, comma-separated position columns. " << endl << "*****" << endl;
- ShowHelp();
+ getoverlap_help();
}
else {
short s1, e1, s2, e2;
@@ -105,16 +105,15 @@ int main(int argc, char* argv[]) {
}
}
else {
- ShowHelp();
+ getoverlap_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void getoverlap_help(void) {
+ cerr << "\nTool: bedtools overlap (aka getOverlap)" << endl;
+
cerr << "Summary: Computes the amount of overlap (positive values)" << endl;
cerr << "\t or distance (negative values) between genome features" << endl;
cerr << "\t and reports the result at the end of the same line." << endl << endl;
diff --git a/src/intersectBed/Makefile b/src/intersectBed/Makefile
index 71cba93dd95b06e319ad3d2621874859985373fc..a6919d99c1af7e4c7677af53101fd4a43dfab9af 100644
--- a/src/intersectBed/Makefile
+++ b/src/intersectBed/Makefile
@@ -25,17 +25,13 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= intersectBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
- @$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
+ @$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(DFLAGS) $(INCLUDES)
$(EXT_OBJECTS):
@$(MAKE) --no-print-directory -C $(UTILITIES_DIR)/bedFile/
diff --git a/src/intersectBed/intersectMain.cpp b/src/intersectBed/intersectMain.cpp
index f1af974ad2aed7a00c36514ece89bf3fa29b91fe..d5602e7c8363172b4ccef15bd043c747901940ee 100644
--- a/src/intersectBed/intersectMain.cpp
+++ b/src/intersectBed/intersectMain.cpp
@@ -224,15 +224,14 @@ int intersect_main(int argc, char* argv[]) {
}
else {
intersect_help();
+ return 0;
}
}
void intersect_help(void) {
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
-
+ cerr << "\nTool: bedtools intersect (aka intersectBed)" << endl;
+
cerr << "Summary: Report overlaps between two feature files." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <bed/gff/vcf> -b <bed/gff/vcf>" << endl << endl;
@@ -241,7 +240,7 @@ void intersect_help(void) {
cerr << "\t-abam\t" << "The A input file is in BAM format. Output will be BAM as well." << endl << endl;
- cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
+ cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default writes compressed BAM." << endl << endl;
cerr << "\t-bed\t" << "When using BAM input (-abam), write output as BED. The default" << endl;
cerr << "\t\tis to write output in BAM when using -abam." << endl << endl;
@@ -281,18 +280,17 @@ void intersect_help(void) {
cerr << "\t\t- In other words, if -f is 0.90 and -r is used, this requires" << endl;
cerr << "\t\t that B overlap 90% of A and A _also_ overlaps 90% of B." << endl << endl;
- cerr << "\t-s\t" << "Require same strandedness. That is, only report hits in B that" << endl;
- cerr << "\t\toverlap A on the _same_ strand." << endl;
+ cerr << "\t-s\t" << "Require same strandedness. That is, only report hits in B" << endl;
+ cerr << "\t\tthat overlap A on the _same_ strand." << endl;
cerr << "\t\t- By default, overlaps are reported without respect to strand." << endl << endl;
- cerr << "\t-S\t" << "Require different strandedness. That is, only report hits in B that" << endl;
- cerr << "\t\toverlap A on the _opposite_ strand." << endl;
+ cerr << "\t-S\t" << "Require different strandedness. That is, only report hits in B" << endl;
+ cerr << "\t\tthat overlap A on the _opposite_ strand." << endl;
cerr << "\t\t- By default, overlaps are reported without respect to strand." << endl << endl;
cerr << "\t-split\t" << "Treat \"split\" BAM or BED12 entries as distinct BED intervals." << endl << endl;
- cerr << "\t-sorted\t" << "Use the \"chromsweep\" algorithm for sorted (-k1,1 -k2,2n) input" << endl;
- cerr << "\t\tNOTE: this will trust, but not enforce that data is sorted. Caveat emptor." << endl << endl;
+ cerr << "\t-sorted\t" << "Use the \"chromsweep\" algorithm for sorted (-k1,1 -k2,2n) input" << endl << endl;
cerr << "\t-header\t" << "Print the header from the A file prior to results." << endl << endl;
diff --git a/src/linksBed/Makefile b/src/linksBed/Makefile
index 75317bf20e885fc00e7af6506f90e159474b42d0..7c354db999788cdbea989a044531080b412f6a8c 100644
--- a/src/linksBed/Makefile
+++ b/src/linksBed/Makefile
@@ -18,14 +18,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= linksBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/linksBed/linksMain.cpp b/src/linksBed/linksMain.cpp
index f4e5500d68686694b9262285f43d9608bf05828c..8009d8842cbec9b8ff72b6a24132f05dd5351216 100644
--- a/src/linksBed/linksMain.cpp
+++ b/src/linksBed/linksMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "linksBed"
+#define PROGRAM_NAME "bedtools links"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void links_help(void);
-int main(int argc, char* argv[]) {
+int links_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -47,7 +47,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) links_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -93,19 +93,17 @@ int main(int argc, char* argv[]) {
if (!showHelp) {
BedLinks *bl = new BedLinks(bedFile, base, org, db);
delete bl;
- return 0;
}
else {
- ShowHelp();
+ links_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void links_help(void) {
+ cerr << "\nTool: bedtools links (aka linksBed)" << endl;
+
cerr << "Summary: Creates HTML links to an UCSC Genome Browser from a feature file." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> > out.html" << endl << endl;
diff --git a/src/maskFastaFromBed/Makefile b/src/maskFastaFromBed/Makefile
index 9b65fc5255774fab8ecf831aae62226066064058..3b5ec085d3f8560f5cff365748ed123999159a70 100644
--- a/src/maskFastaFromBed/Makefile
+++ b/src/maskFastaFromBed/Makefile
@@ -17,14 +17,10 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= maskFastaFromBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/maskFastaFromBed/maskFastaFromBedMain.cpp b/src/maskFastaFromBed/maskFastaFromBedMain.cpp
index 7fce56bbbe913fca5384b19dba4e5fb86ad99e82..5bb73908a5d307a0fe90fb3b83eee1bc04035a0d 100644
--- a/src/maskFastaFromBed/maskFastaFromBedMain.cpp
+++ b/src/maskFastaFromBed/maskFastaFromBedMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "maskFastaFromBed"
+#define PROGRAM_NAME "bedtools maskfasta"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void maskfastafrombed_help(void);
-int main(int argc, char* argv[]) {
+int maskfastafrombed_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -55,7 +55,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) maskfastafrombed_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -113,24 +113,20 @@ int main(int argc, char* argv[]) {
MaskFastaFromBed *maskFasta = new MaskFastaFromBed(fastaInFile, bedFile, fastaOutFile, softMask, maskChar);
delete maskFasta;
- return 0;
}
else {
- ShowHelp();
+ maskfastafrombed_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
-
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void maskfastafrombed_help(void) {
+ cerr << "\nTool: bedtools maskfasta (aka maskFastaFromBed)" << endl;
+
cerr << "Summary: Mask a fasta file based on feature coordinates." << endl << endl;
- cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -fi <fasta> -out <fasta> -bed <bed/gff/vcf>" << endl << endl;
+ cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -fi <fasta> -fo <fasta> -bed <bed/gff/vcf>" << endl << endl;
cerr << "Options:" << endl;
cerr << "\t-fi\tInput FASTA file" << endl;
@@ -138,8 +134,8 @@ void ShowHelp(void) {
cerr << "\t-fo\tOutput FASTA file" << endl;
cerr << "\t-soft\tEnforce \"soft\" masking. That is, instead of masking with Ns," << endl;
cerr << "\t\tmask with lower-case bases." << endl;
- cerr << "\t-mc\tReplace masking character. That is, instead of masking with Ns, use another character." << endl;
-
+ cerr << "\t-mc\tReplace masking character. That is, instead of masking" << endl;
+ cerr << "\t\twith Ns, use another character." << endl << endl;
// end the program here
exit(1);
diff --git a/src/mergeBed/Makefile b/src/mergeBed/Makefile
index c6a08b6dae4f95bdfeb7dbefc43de0354698fd1a..345571e282a22181384ac876c8d3f7da35f73e2f 100644
--- a/src/mergeBed/Makefile
+++ b/src/mergeBed/Makefile
@@ -18,14 +18,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= mergeBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/mergeBed/mergeMain.cpp b/src/mergeBed/mergeMain.cpp
index 03ee935511763992ad5503f2c34d2222570c2516..b91d4e1fac3f8ef06a2165157352a18c062a06f8 100644
--- a/src/mergeBed/mergeMain.cpp
+++ b/src/mergeBed/mergeMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "mergeBed"
+#define PROGRAM_NAME "bedtools merge"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void merge_help(void);
-int main(int argc, char* argv[]) {
+int merge_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -51,7 +51,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) merge_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -112,19 +112,17 @@ int main(int argc, char* argv[]) {
if (!showHelp) {
BedMerge *bm = new BedMerge(bedFile, numEntries, maxDistance, forceStrand, reportNames, reportScores, scoreOp);
delete bm;
- return 0;
}
else {
- ShowHelp();
+ merge_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
-
+void merge_help(void) {
+
+ cerr << "\nTool: bedtools merge (aka mergeBed)" << endl;
+
cerr << "Summary: Merges overlapping BED/GFF/VCF entries into a single interval." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf>" << endl << endl;
diff --git a/src/multiBamCov/Makefile b/src/multiBamCov/Makefile
index 9313ab959e995bdeea6427c29eb42924045826c1..cf992632b30f53501abcfd00e371415cce440ef3 100644
--- a/src/multiBamCov/Makefile
+++ b/src/multiBamCov/Makefile
@@ -22,14 +22,10 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= multiBamCov
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/multiBamCov/multiBamCovMain.cpp b/src/multiBamCov/multiBamCovMain.cpp
index 69b340dc975ba4c9cc7d8215bb03ea14e5e881a8..6b3c1b8544391d2ba45f1cb5756612aba8076060 100644
--- a/src/multiBamCov/multiBamCovMain.cpp
+++ b/src/multiBamCov/multiBamCovMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "multiBamCov"
+#define PROGRAM_NAME "bedtools multicov"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void multibamcov_help(void);
-int main(int argc, char* argv[]) {
+int multibamcov_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -53,7 +53,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) multibamcov_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -107,19 +107,17 @@ int main(int argc, char* argv[]) {
MultiCovBam *mc = new MultiCovBam(bamFiles, bedFile, minQual, properOnly, keepDuplicates, keepFailedQC);
mc->CollectCoverage();
delete mc;
- return 0;
}
else {
- ShowHelp();
+ multibamcov_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void multibamcov_help(void) {
+ cerr << "\nTool: bedtools multicov (aka multiBamCov)" << endl;
+
cerr << "Summary: Counts sequence coverage for multiple bams at specific loci." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -bams aln.1.bam aln.2.bam ... aln.n.bam -bed <bed/gff/vcf>" << endl << endl;
@@ -132,12 +130,12 @@ void ShowHelp(void) {
cerr << "\t-q\t" << "Minimum mapping quality allowed. Default is 0." << endl << endl;
- cerr << "\t-D\t" << "Include duplicate-marked reads. Default is to count non-duplicates only" << endl << endl;
+ cerr << "\t-D\t" << "Include duplicate reads. Default counts non-duplicates only" << endl << endl;
- cerr << "\t-F\t" << "Include failed-QC reads. Default is to count pass-QC reads only" << endl << endl;
+ cerr << "\t-F\t" << "Include failed-QC reads. Default counts pass-QC reads only" << endl << endl;
- cerr << "\t-p\t" << "Only count proper pairs. Default is to count all alignments with MAPQ" << endl;
- cerr << "\t\t" << "greater than the -q argument, regardless of the BAM FLAG field." << endl << endl;
+ cerr << "\t-p\t" << "Only count proper pairs. Default counts all alignments with" << endl;
+ cerr << "\t\t" << "MAPQ > -q argument, regardless of the BAM FLAG field." << endl << endl;
// end the program here
exit(1);
diff --git a/src/multiIntersectBed/Makefile b/src/multiIntersectBed/Makefile
index a076f3b411dbb81a4083f95b88edc03350799d84..4dfe6113c6fc55af5548037ca531da6e6e0cc43c 100644
--- a/src/multiIntersectBed/Makefile
+++ b/src/multiIntersectBed/Makefile
@@ -23,14 +23,10 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= multiIntersectBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/multiIntersectBed/multiIntersectBedMain.cpp b/src/multiIntersectBed/multiIntersectBedMain.cpp
index 5b4b627b3f5cda39450147dbf5000c83d2b6abd1..d67bdb4cc34032a10b79579dfd26887247599630 100644
--- a/src/multiIntersectBed/multiIntersectBedMain.cpp
+++ b/src/multiIntersectBed/multiIntersectBedMain.cpp
@@ -26,7 +26,7 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "multiIntersectBed"
+#define PROGRAM_NAME "bedtools multiinter"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
@@ -37,11 +37,10 @@ using namespace std;
std::string stl_basename(const std::string& path);
// function declarations
-void ShowHelp(void);
-void ShowExamples(void);
+void multiintersect_help(void);
+void multiintersect_examples(void);
-
-int main(int argc, char* argv[])
+int multiintersect_main(int argc, char* argv[])
{
bool haveFiles = false;
bool haveTitles = false;
@@ -59,7 +58,7 @@ int main(int argc, char* argv[])
//Parse command line options
if(argc <= 1)
- ShowHelp();
+ multiintersect_help();
for(int i = 1; i < argc; i++) {
int parameterLength = (int)strlen(argv[i]);
@@ -71,7 +70,7 @@ int main(int argc, char* argv[])
}
if(showHelp == true) {
- ShowHelp();
+ multiintersect_help();
exit(1);
}
@@ -132,8 +131,8 @@ int main(int argc, char* argv[])
cluster = true;
}
else if(PARAMETER_CHECK("-examples", 9, parameterLength)) {
- ShowHelp();
- ShowExamples();
+ multiintersect_help();
+ multiintersect_examples();
exit(1);
}
}
@@ -163,15 +162,14 @@ int main(int argc, char* argv[])
mbi.MultiIntersect();
else
mbi.Cluster();
+
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Authors: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
- cerr << " Assaf Gordon" << endl << endl;
+void multiintersect_help(void) {
+ cerr << "\nTool: bedtools multiinter (aka multiIntersectBed)" << endl;
+
cerr << "Summary: Identifies common intervals among multiple" << endl;
cerr << "\t BED/GFF/VCF files." << endl << endl;
@@ -180,12 +178,12 @@ void ShowHelp(void) {
cerr << "Options: " << endl;
- cerr << "\t-cluster\t\t" << "Invoke Ryan's algorithm." << endl << endl;
+ cerr << "\t-cluster\t" << "Invoke Ryan Layers's clustering algorithm." << endl << endl;
cerr << "\t-header\t\t" << "Print a header line." << endl;
cerr << "\t\t\t(chrom/start/end + names of each file)." << endl << endl;
- cerr << "\t-names\t\t" << "A list of names (one / file) to describe each file in -i." << endl;
+ cerr << "\t-names\t\t" << "A list of names (one/file) to describe each file in -i." << endl;
cerr << "\t\t\tThese names will be printed in the header line." << endl << endl;
cerr << "\t-g\t\t" << "Use genome file to calculate empty regions." << endl;
@@ -196,14 +194,14 @@ void ShowHelp(void) {
cerr << "\t\t\t- Requires the '-g FILE' parameter.\n" << endl;
cerr << "\t-filler TEXT\t" << "Use TEXT when representing intervals having no value." << endl;
- cerr << "\t\t\t- Default is '0', but you can use 'N/A' or any other text." << endl << endl;
+ cerr << "\t\t\t- Default is '0', but you can use 'N/A' or any text." << endl << endl;
cerr << "\t-examples\t" << "Show detailed usage examples." << endl << endl;
}
-void ShowExamples()
+void multiintersect_examples()
{
cerr << "Example usage:\n\n" \
"== Input files: ==\n" \
diff --git a/src/nucBed/Makefile b/src/nucBed/Makefile
index b824c458f891bd43434113957283e84e8374c8bf..95ba76f0686626313adc221d47c83365595fdcd4 100644
--- a/src/nucBed/Makefile
+++ b/src/nucBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= nucBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/nucBed/nucBedMain.cpp b/src/nucBed/nucBedMain.cpp
index f92ddd883d6bca862bc2beea51a7a1f0d8825ad6..a56cf814f6059f289480aac3efd3a304087b6ff1 100644
--- a/src/nucBed/nucBedMain.cpp
+++ b/src/nucBed/nucBedMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "nucBed"
+#define PROGRAM_NAME "bedtools nuc"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void nuc_help(void);
-int main(int argc, char* argv[]) {
+int nuc_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -53,7 +53,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) nuc_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -101,20 +101,17 @@ int main(int argc, char* argv[]) {
NucBed *nuc = new NucBed(fastaDbFile, bedFile, printSeq, hasPattern, pattern, forceStrand);
delete nuc;
-
- return 0;
}
else {
- ShowHelp();
+ nuc_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void nuc_help(void) {
+ cerr << "\nTool: bedtools nuc (aka nucBed)" << endl;
+
cerr << "Summary: Profiles the nucleotide content of intervals in a fasta file." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -fi <fasta> -bed <bed/gff/vcf>" << endl << endl;
@@ -124,11 +121,12 @@ void ShowHelp(void) {
cerr << "\t-bed\tBED/GFF/VCF file of ranges to extract from -fi" << endl << endl;
cerr << "\t-s\tProfile the sequence according to strand." << endl << endl;
cerr << "\t-seq\tPrint the extracted sequence" << endl << endl;
- cerr << "\t-pattern\tReport the number of times a user-defined sequence is observed (case-insensitive)." << endl << endl;
+ cerr << "\t-pattern\tReport the number of times a user-defined sequence" << endl;
+ cerr << "\t\t\tis observed (case-insensitive)." << endl << endl;
cerr << "Output format: " << endl;
- cerr << "\tThe following information will be reported after each original BED entry:" << endl;
+ cerr << "\tThe following information will be reported after each BED entry:" << endl;
cerr << "\t 1) %AT content" << endl;
cerr << "\t 2) %GC content" << endl;
cerr << "\t 3) Number of As observed" << endl;
@@ -138,9 +136,9 @@ void ShowHelp(void) {
cerr << "\t 7) Number of Ns observed" << endl;
cerr << "\t 8) Number of other bases observed" << endl;
cerr << "\t 9) The length of the explored sequence/interval." << endl;
- cerr << "\t 10) The sequence extracted from the FASTA file. (optional, if -seq is used)" << endl;
- cerr << "\t 11) The number of times a user defined pattern was observed. (optional, if -pattern is used.)" << endl;
-
+ cerr << "\t 10) The seq. extracted from the FASTA file. (opt., if -seq is used)" << endl;
+ cerr << "\t 11) The number of times a user's pattern was observed." << endl;
+ cerr << "\t (opt., if -pattern is used.)" << endl << endl;
// end the program here
exit(1);
diff --git a/src/pairToBed/Makefile b/src/pairToBed/Makefile
index 9d94beb499169cbaec07b1ee36ea5f28ed44ac14..015b2aa53df949c94efcdcbdc8279b035be6e2ec 100644
--- a/src/pairToBed/Makefile
+++ b/src/pairToBed/Makefile
@@ -24,14 +24,10 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= pairToBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/pairToBed/pairToBedMain.cpp b/src/pairToBed/pairToBedMain.cpp
index 38b8714a13ec6ba0638882961757fb2c0c90c08d..d84ed33f506dbaa578f802ee24f3a7a9f485f4e0 100644
--- a/src/pairToBed/pairToBedMain.cpp
+++ b/src/pairToBed/pairToBedMain.cpp
@@ -15,15 +15,15 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "pairToBed"
+#define PROGRAM_NAME "bedtools pairtobed"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void pairtobed_help(void);
-int main(int argc, char* argv[]) {
+int pairtobed_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -60,7 +60,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) pairtobed_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -160,20 +160,18 @@ int main(int argc, char* argv[]) {
searchType, sameStrand, diffStrand, inputIsBam,
outputIsBam, uncompressedBam, useEditDistance);
delete bi;
- return 0;
}
else {
- ShowHelp();
+ pairtobed_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
-
+void pairtobed_help(void) {
+
+ cerr << "\nTool: bedtools pairtobed (aka pairToBed)" << endl;
+
cerr << "Summary: Report overlaps between a BEDPE file and a BED/GFF/VCF file." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <bedpe> -b <bed/gff/vcf>" << endl << endl;
@@ -183,7 +181,7 @@ void ShowHelp(void) {
cerr << "\t-abam\t" << "The A input file is in BAM format. Output will be BAM as well." << endl;
cerr << "\t\t- Requires BAM to be grouped or sorted by query." << endl << endl;
- cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
+ cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default writes compressed BAM." << endl << endl;
cerr << "\t\tis to write output in BAM when using -abam." << endl << endl;
cerr << "\t-bedpe\t" << "When using BAM input (-abam), write output as BEDPE. The default" << endl;
diff --git a/src/pairToPair/Makefile b/src/pairToPair/Makefile
index 4d75faaa4c6e850b86c08a89c920a2f543c331e8..0e994bb902837ed3b6b6831a9577eaf346ee03ce 100644
--- a/src/pairToPair/Makefile
+++ b/src/pairToPair/Makefile
@@ -18,14 +18,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= pairToPair
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/pairToPair/pairToPairMain.cpp b/src/pairToPair/pairToPairMain.cpp
index 7f5761d01f25d40438d0f9a2a3fdd54d737ab245..5182df655aafec4c0d6aafcf8c0e8679d4cd2993 100644
--- a/src/pairToPair/pairToPairMain.cpp
+++ b/src/pairToPair/pairToPairMain.cpp
@@ -15,15 +15,15 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "pairToPair"
+#define PROGRAM_NAME "bedtools pairtopair"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void pairtopair_help(void);
-int main(int argc, char* argv[]) {
+int pairtopair_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -58,7 +58,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) pairtopair_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -140,16 +140,16 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ pairtopair_help();
}
+ return 0;
}
-void ShowHelp(void) {
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void pairtopair_help(void) {
+ cerr << "\nTool: bedtools pairtopair (aka pairToPair)" << endl;
+
cerr << "Summary: Report overlaps between two paired-end BED files (BEDPE)." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <BEDPE> -b <BEDPE>" << endl << endl;
@@ -162,13 +162,14 @@ void ShowHelp(void) {
cerr << "\t\tneither\tReport overlaps if neither end of A overlaps B." << endl;
cerr << "\t\teither\tReport overlaps if either ends of A overlap B." << endl;
cerr << "\t\tboth\tReport overlaps if both ends of A overlap B." << endl;
- cerr << "\t\tnotboth\tReport overlaps if one or neither of ends of A overlap B." << endl;
+ cerr << "\t\tnotboth\tReport overlaps if one or neither of A's overlap B." << endl;
cerr << "\t\t- Default = both." << endl << endl;
cerr << "\t-slop \t" << "The amount of slop (in b.p.). to be added to each footprint." << endl;
- cerr << "\t\t*Note*: Slop is subtracted from start1 and start2 and added to end1 and end2." << endl << endl;
-
+ cerr << "\t\t*Note*: Slop is subtracted from start1 and start2" << endl;
+ cerr << "\t\t\tand added to end1 and end2." << endl << endl;
+
cerr << "\t-ss\t" << "Add slop based to each BEDPE footprint based on strand." << endl;
cerr << "\t\t- If strand is \"+\", slop is only added to the end coordinates." << endl;
cerr << "\t\t- If strand is \"-\", slop is only added to the start coordinates." << endl;
diff --git a/src/shuffleBed/Makefile b/src/shuffleBed/Makefile
index 0265ae05dc87aed9923a70c257b846858d6db569..a20290cda2d51aabf1dca37b765bae3a925d14c6 100644
--- a/src/shuffleBed/Makefile
+++ b/src/shuffleBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= shuffleBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/shuffleBed/shuffleBedMain.cpp b/src/shuffleBed/shuffleBedMain.cpp
index d731d1a712a752a633e58d13bbaafcdc922bfc1f..38f3dacfc57e899b9401a62e240dc801d422db73 100644
--- a/src/shuffleBed/shuffleBedMain.cpp
+++ b/src/shuffleBed/shuffleBedMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "shuffleBed"
+#define PROGRAM_NAME "bedtools shuffle"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void shuffle_help(void);
-int main(int argc, char* argv[]) {
+int shuffle_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -54,7 +54,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) shuffle_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -129,16 +129,15 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ shuffle_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void shuffle_help(void) {
+ cerr << "\nTool: bedtools shuffle (aka shuffleBed)" << endl;
+
cerr << "Summary: Randomly permute the locations of a feature file among a genome." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome>" << endl << endl;
diff --git a/src/slopBed/Makefile b/src/slopBed/Makefile
index aed0b82d0d02fbd6cb6fb086aef7f3dc0733399c..9f8881064ee920f4e07eb4680f983f4a4633883e 100644
--- a/src/slopBed/Makefile
+++ b/src/slopBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= slopBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/slopBed/slopBedMain.cpp b/src/slopBed/slopBedMain.cpp
index b5f148e56dc6aab8aa714a1f7cc45ba8f01ded44..c7d870f1b803482754cfd57faf3f9cdffb4f25e8 100644
--- a/src/slopBed/slopBedMain.cpp
+++ b/src/slopBed/slopBedMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "slopBed"
+#define PROGRAM_NAME "bedtools slop"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void slop_help(void);
-int main(int argc, char* argv[]) {
+int slop_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -54,7 +54,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) slop_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -136,22 +136,21 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ slop_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void slop_help(void) {
+ cerr << "\nTool: bedtools slop (aka slopBed)" << endl;
+
cerr << "Summary: Add requested base pairs of \"slop\" to each feature." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome> [-b <int> or (-l and -r)]" << endl << endl;
cerr << "Options: " << endl;
- cerr << "\t-b\t" << "Increase the BED/GFF/VCF entry by -b base pairs in each direction." << endl;
+ cerr << "\t-b\t" << "Increase the BED/GFF/VCF entry -b base pairs in each direction." << endl;
cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
cerr << "\t-l\t" << "The number of base pairs to subtract from the start coordinate." << endl;
diff --git a/src/sortBed/Makefile b/src/sortBed/Makefile
index c675e68aa40425ab48cd29799a4548b4e1641075..55037c045aadcc84f4f1c0d0de38eb118d5120a5 100644
--- a/src/sortBed/Makefile
+++ b/src/sortBed/Makefile
@@ -18,14 +18,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= sortBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/sortBed/sortMain.cpp b/src/sortBed/sortMain.cpp
index e58faa207fd66e97b8c51555a405f56120083469..767c9b65f8b6b9c7bcb09680be8c905fee378efb 100644
--- a/src/sortBed/sortMain.cpp
+++ b/src/sortBed/sortMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "sortBed"
+#define PROGRAM_NAME "bedtools sort"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void sort_help(void);
-int main(int argc, char* argv[]) {
+int sort_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -51,7 +51,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) sort_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -135,15 +135,15 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ sort_help();
}
+ return 0;
}
-void ShowHelp(void) {
+void sort_help(void) {
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+ cerr << "\nTool: bedtools sort (aka sortBed)" << endl;
+
cerr << "Summary: Sorts a feature file in various and useful ways." << endl << endl;
cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf>" << endl << endl;
diff --git a/src/subtractBed/Makefile b/src/subtractBed/Makefile
index f67ac7b666bf3ed697b4e282bcc0c13708e33903..4d11368aba26a36e80ff7fef392e83ae428d80b0 100644
--- a/src/subtractBed/Makefile
+++ b/src/subtractBed/Makefile
@@ -22,14 +22,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= subtractBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/subtractBed/subtractMain.cpp b/src/subtractBed/subtractMain.cpp
index 58b4b412045027c64fc66bd3efc333dd11cbdcff..8ea557d52f73a513243a79a544050bb847706dd2 100644
--- a/src/subtractBed/subtractMain.cpp
+++ b/src/subtractBed/subtractMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "subtractBed"
+#define PROGRAM_NAME "bedtools subtract"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void subtract_help(void);
-int main(int argc, char* argv[]) {
+int subtract_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -54,7 +54,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) subtract_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -112,16 +112,15 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ subtract_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void subtract_help(void) {
+ cerr << "\nTool: bedtools subtract (aka subtractBed)" << endl;
+
cerr << "Summary: Removes the portion(s) of an interval that is overlapped" << endl;
cerr << "\t by another feature(s)." << endl << endl;
@@ -132,8 +131,8 @@ void ShowHelp(void) {
cerr << "\t\t- Default is 1E-9 (i.e., 1bp)." << endl;
cerr << "\t\t- (FLOAT) (e.g. 0.50)" << endl << endl;
- cerr << "\t-s\t" << "Require same strandedness. That is, only subtract hits in B that" << endl;
- cerr << "\t\toverlap A on the _same_ strand." << endl;
+ cerr << "\t-s\t" << "Require same strandedness. That is, only subtract hits in B" << endl;
+ cerr << "\t\tthat overlap A on the _same_ strand." << endl;
cerr << "\t\t- By default, overlaps are subtracted without respect to strand." << endl << endl;
cerr << "\t-S\t" << "Force strandedness. That is, only subtract hits in B that" << endl;
diff --git a/src/tagBam/Makefile b/src/tagBam/Makefile
index 35b97a2cc093abd757eef86be6c3a53f0ff69ebf..6b48cc33395d235cf9ca92d663bc48346b9e891c 100644
--- a/src/tagBam/Makefile
+++ b/src/tagBam/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= tagBam
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/tagBam/tagBamMain.cpp b/src/tagBam/tagBamMain.cpp
index b4c750dcfdd36bc0c6f79c4da7cc8072ea3b82d1..1a98528092e248d5cd01f073b47d044aaece9eb8 100644
--- a/src/tagBam/tagBamMain.cpp
+++ b/src/tagBam/tagBamMain.cpp
@@ -15,15 +15,15 @@
using namespace std;
// define the version
-#define PROGRAM_NAME "tagBam"
+#define PROGRAM_NAME "bedtools tag"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void tagbam_help(void);
-int main(int argc, char* argv[]) {
+int tagbam_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -62,7 +62,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) tagbam_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -188,16 +188,15 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ tagbam_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void tagbam_help(void) {
+ cerr << "\nTool: bedtools tag (aka tagBam)" << endl;
+
cerr << "Summary: Annotates a BAM file based on overlaps with multiple BED/GFF/VCF files" << endl;
cerr << "\t on the intervals in -i." << endl << endl;
diff --git a/src/unionBedGraphs/Makefile b/src/unionBedGraphs/Makefile
index bca92f26e0c9ab1ed2d6ce567a39968d62e3b82a..56622af088641b36548c1dcc57eead49bc07afac 100755
--- a/src/unionBedGraphs/Makefile
+++ b/src/unionBedGraphs/Makefile
@@ -23,14 +23,10 @@ EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= unionBedGraphs
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/unionBedGraphs/unionBedGraphsMain.cpp b/src/unionBedGraphs/unionBedGraphsMain.cpp
index b140b013d84b74ce2d5a2c27337431072b07d21c..69c7546ba51b1cbbb92cdd33c4469147d3126553 100644
--- a/src/unionBedGraphs/unionBedGraphsMain.cpp
+++ b/src/unionBedGraphs/unionBedGraphsMain.cpp
@@ -26,7 +26,7 @@
using namespace std;
// define our program name
-#define PROGRAM_NAME "unionBedGraphs"
+#define PROGRAM_NAME "bedtools unionbedg"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
@@ -34,14 +34,14 @@ using namespace std;
//STLized version of basename()
// (because POSIX basename() modifies the input string pointer)
// Additionally: removes any extension the basename might have.
-std::string stl_basename(const std::string& path);
+std::string ubg_stl_basename(const std::string& path);
// function declarations
-void ShowHelp(void);
-void ShowExamples(void);
+void unionbedgraphs_help(void);
+void unionbedgraphs_showexamples(void);
-int main(int argc, char* argv[])
+int unionbedgraphs_main(int argc, char* argv[])
{
bool haveFiles = false;
bool haveTitles = false;
@@ -58,7 +58,7 @@ int main(int argc, char* argv[])
//Parse command line options
if(argc <= 1)
- ShowHelp();
+ unionbedgraphs_help();
for(int i = 1; i < argc; i++) {
int parameterLength = (int)strlen(argv[i]);
@@ -70,7 +70,7 @@ int main(int argc, char* argv[])
}
if(showHelp == true) {
- ShowHelp();
+ unionbedgraphs_help();
exit(1);
}
@@ -128,8 +128,8 @@ int main(int argc, char* argv[])
printEmptyRegions = true;
}
else if(PARAMETER_CHECK("-examples", 9, parameterLength)) {
- ShowHelp();
- ShowExamples();
+ unionbedgraphs_help();
+ unionbedgraphs_showexamples();
exit(1);
}
}
@@ -156,15 +156,14 @@ int main(int argc, char* argv[])
if (printHeader)
ubg.PrintHeader();
ubg.Union();
+
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Authors: Assaf Gordon, CSHL" << endl;
- cerr << " Aaron Quinlan (aaronquinlan@gmail.com)" << endl << endl;
+void unionbedgraphs_help(void) {
+ cerr << "\nTool: bedtools unionbedg (aka unionBedGraphs)" << endl;
+
cerr << "Summary: Combines multiple BedGraph files into a single file," << endl;
cerr << "\t allowing coverage comparisons between them." << endl << endl;
@@ -177,7 +176,7 @@ void ShowHelp(void) {
cerr << "\t-header\t\t" << "Print a header line." << endl;
cerr << "\t\t\t(chrom/start/end + names of each file)." << endl << endl;
- cerr << "\t-names\t\t" << "A list of names (one / file) to describe each file in -i." << endl;
+ cerr << "\t-names\t\t" << "A list of names (one/file) to describe each file in -i." << endl;
cerr << "\t\t\tThese names will be printed in the header line." << endl << endl;
cerr << "\t-g\t\t" << "Use genome file to calculate empty regions." << endl;
@@ -188,14 +187,14 @@ void ShowHelp(void) {
cerr << "\t\t\t- Requires the '-g FILE' parameter.\n" << endl;
cerr << "\t-filler TEXT\t" << "Use TEXT when representing intervals having no value." << endl;
- cerr << "\t\t\t- Default is '0', but you can use 'N/A' or any other text." << endl << endl;
+ cerr << "\t\t\t- Default is '0', but you can use 'N/A' or any text." << endl << endl;
cerr << "\t-examples\t" << "Show detailed usage examples." << endl << endl;
}
-void ShowExamples()
+void unionbedgraphs_showexamples()
{
cerr << "Example usage:\n\n" \
"== Input files: ==\n" \
@@ -276,7 +275,7 @@ void ShowExamples()
;
}
-std::string stl_basename(const std::string& path)
+std::string ubg_stl_basename(const std::string& path)
{
string result;
diff --git a/src/utils/BamTools/include/api/BamAlignment.h b/src/utils/BamTools/include/api/BamAlignment.h
new file mode 100644
index 0000000000000000000000000000000000000000..7535d93f8f7a8fcd1062348b7d406a4082fa0abe
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamAlignment.h
@@ -0,0 +1,207 @@
+// ***************************************************************************
+// BamAlignment.h (c) 2009 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 22 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides the BamAlignment data structure
+// ***************************************************************************
+
+#ifndef BAMALIGNMENT_H
+#define BAMALIGNMENT_H
+
+#include <api/api_global.h>
+#include <api/BamAux.h>
+#include <string>
+#include <vector>
+
+namespace BamTools {
+
+// forward declaration of BamAlignment's friend classes
+namespace Internal {
+ class BamReaderPrivate;
+ class BamWriterPrivate;
+} // namespace Internal
+
+// BamAlignment data structure
+struct API_EXPORT BamAlignment {
+
+ // constructors & destructor
+ public:
+ BamAlignment(void);
+ BamAlignment(const BamAlignment& other);
+ ~BamAlignment(void);
+
+ // queries against alignment flags
+ public:
+ bool IsDuplicate(void) const; // returns true if this read is a PCR duplicate
+ bool IsFailedQC(void) const; // returns true if this read failed quality control
+ bool IsFirstMate(void) const; // returns true if alignment is first mate on read
+ bool IsMapped(void) const; // returns true if alignment is mapped
+ bool IsMateMapped(void) const; // returns true if alignment's mate is mapped
+ bool IsMateReverseStrand(void) const; // returns true if alignment's mate mapped to reverse strand
+ bool IsPaired(void) const; // returns true if alignment part of paired-end read
+ bool IsPrimaryAlignment(void) const; // returns true if reported position is primary alignment
+ bool IsProperPair(void) const; // returns true if alignment is part of read that satisfied paired-end resolution
+ bool IsReverseStrand(void) const; // returns true if alignment mapped to reverse strand
+ bool IsSecondMate(void) const; // returns true if alignment is second mate on read
+
+ // manipulate alignment flags
+ public:
+ void SetIsDuplicate(bool ok); // sets value of "PCR duplicate" flag
+ void SetIsFailedQC(bool ok); // sets value of "failed quality control" flag
+ void SetIsFirstMate(bool ok); // sets value of "alignment is first mate" flag
+ void SetIsMapped(bool ok); // sets value of "alignment is mapped" flag
+ void SetIsMateMapped(bool ok); // sets value of "alignment's mate is mapped" flag
+ void SetIsMateReverseStrand(bool ok); // sets value of "alignment's mate mapped to reverse strand" flag
+ void SetIsPaired(bool ok); // sets value of "alignment part of paired-end read" flag
+ void SetIsPrimaryAlignment(bool ok); // sets value of "position is primary alignment" flag
+ void SetIsProperPair(bool ok); // sets value of "alignment is part of read that satisfied paired-end resolution" flag
+ void SetIsReverseStrand(bool ok); // sets value of "alignment mapped to reverse strand" flag
+ void SetIsSecondMate(bool ok); // sets value of "alignment is second mate on read" flag
+
+ // legacy methods (consider deprecated, but still available)
+ void SetIsMateUnmapped(bool ok); // complement of using SetIsMateMapped()
+ void SetIsSecondaryAlignment(bool ok); // complement of using SetIsPrimaryAlignment()
+ void SetIsUnmapped(bool ok); // complement of using SetIsMapped()
+
+ // tag data access methods
+ public:
+
+ // -------------------------------------------------------------------------------------
+ // N.B. - The following tag access methods may not be used on BamAlignments fetched
+ // using BamReader::GetNextAlignmentCore(). Attempting to use them will not result in
+ // error message (to keep output clean) but will ALWAYS return false. Only user-created
+ // BamAlignments or those retrieved using BamReader::GetNextAlignment() are valid here.
+ //
+ // You can call BuildCharData() on such an alignment retrieved by GetNextAlignmentCore().
+ // This populates all the character data, and will enable subsequent queries on tag data.
+ // -------------------------------------------------------------------------------------
+
+ // adds a tag
+ bool AddTag(const std::string& tag, const std::string& type, const std::string& value);
+ bool AddTag(const std::string& tag, const std::string& type, const uint32_t& value);
+ bool AddTag(const std::string& tag, const std::string& type, const int32_t& value);
+ bool AddTag(const std::string& tag, const std::string& type, const float& value);
+
+ // adds a "binary array" tag
+ bool AddTag(const std::string& tag, const std::vector<uint8_t>& values);
+ bool AddTag(const std::string& tag, const std::vector<int8_t>& values);
+ bool AddTag(const std::string& tag, const std::vector<uint16_t>& values);
+ bool AddTag(const std::string& tag, const std::vector<int16_t>& values);
+ bool AddTag(const std::string& tag, const std::vector<uint32_t>& values);
+ bool AddTag(const std::string& tag, const std::vector<int32_t>& values);
+ bool AddTag(const std::string& tag, const std::vector<float>& values);
+
+ // edits a tag
+ bool EditTag(const std::string& tag, const std::string& type, const std::string& value);
+ bool EditTag(const std::string& tag, const std::string& type, const uint32_t& value);
+ bool EditTag(const std::string& tag, const std::string& type, const int32_t& value);
+ bool EditTag(const std::string& tag, const std::string& type, const float& value);
+
+ // edits a "binary array" tag
+ bool EditTag(const std::string& tag, const std::vector<uint8_t>& values);
+ bool EditTag(const std::string& tag, const std::vector<int8_t>& values);
+ bool EditTag(const std::string& tag, const std::vector<uint16_t>& values);
+ bool EditTag(const std::string& tag, const std::vector<int16_t>& values);
+ bool EditTag(const std::string& tag, const std::vector<uint32_t>& values);
+ bool EditTag(const std::string& tag, const std::vector<int32_t>& values);
+ bool EditTag(const std::string& tag, const std::vector<float>& values);
+
+ // retrieves data for a tag
+ bool GetTag(const std::string& tag, std::string& destination) const;
+ bool GetTag(const std::string& tag, uint32_t& destination) const;
+ bool GetTag(const std::string& tag, int32_t& destination) const;
+ bool GetTag(const std::string& tag, float& destination) const;
+
+ // retrieves data for a "binary array" tag
+ bool GetTag(const std::string& tag, std::vector<uint32_t>& destination) const;
+ bool GetTag(const std::string& tag, std::vector<int32_t>& destination) const;
+ bool GetTag(const std::string& tag, std::vector<float>& destination) const;
+
+ // retrieves the BAM tag-type character for a tag
+ bool GetTagType(const std::string& tag, char& type) const;
+
+ // legacy methods (consider deprecated, but still available)
+ bool GetEditDistance(uint32_t& editDistance) const; // retrieves value of "NM" tag
+ bool GetReadGroup(std::string& readGroup) const; // retrieves value of "RG" tag
+
+ // returns true if alignment has a record for this tag name
+ bool HasTag(const std::string& tag) const;
+
+ // removes a tag
+ bool RemoveTag(const std::string& tag);
+
+ // additional methods
+ public:
+ // populates alignment string fields
+ bool BuildCharData(void);
+ // calculates alignment end position
+ int GetEndPosition(bool usePadded = false, bool zeroBased = true) const;
+
+ // public data fields
+ public:
+ std::string Name; // read name
+ int32_t Length; // length of query sequence
+ std::string QueryBases; // 'original' sequence (as reported from sequencing machine)
+ std::string AlignedBases; // 'aligned' sequence (includes any indels, padding, clipping)
+ std::string Qualities; // FASTQ qualities (ASCII characters, not numeric values)
+ std::string TagData; // tag data (use provided methods to query/modify)
+ int32_t RefID; // ID number for reference sequence
+ int32_t Position; // position (0-based) where alignment starts
+ uint16_t Bin; // BAM (standard) index bin number for this alignment
+ uint16_t MapQuality; // mapping quality score
+ uint32_t AlignmentFlag; // alignment bit-flag (use provided methods to query/modify)
+ std::vector<CigarOp> CigarData; // CIGAR operations for this alignment
+ int32_t MateRefID; // ID number for reference sequence where alignment's mate was aligned
+ int32_t MatePosition; // position (0-based) where alignment's mate starts
+ int32_t InsertSize; // mate-pair insert size
+ std::string Filename; // name of BAM file which this alignment comes from
+
+ //! \cond
+ // internal utility methods
+ private:
+ bool FindTag(const std::string& tag,
+ char*& pTagData,
+ const unsigned int& tagDataLength,
+ unsigned int& numBytesParsed) const;
+ bool IsValidSize(const std::string& tag,
+ const std::string& type) const;
+ bool SkipToNextTag(const char storageType,
+ char*& pTagData,
+ unsigned int& numBytesParsed) const;
+
+ // internal data
+ private:
+
+ struct BamAlignmentSupportData {
+
+ // data members
+ std::string AllCharData;
+ uint32_t BlockLength;
+ uint32_t NumCigarOperations;
+ uint32_t QueryNameLength;
+ uint32_t QuerySequenceLength;
+ bool HasCoreOnly;
+
+ // constructor
+ BamAlignmentSupportData(void)
+ : BlockLength(0)
+ , NumCigarOperations(0)
+ , QueryNameLength(0)
+ , QuerySequenceLength(0)
+ , HasCoreOnly(false)
+ { }
+ };
+ BamAlignmentSupportData SupportData;
+ friend class Internal::BamReaderPrivate;
+ friend class Internal::BamWriterPrivate;
+ //! \endcond
+};
+
+typedef std::vector<BamAlignment> BamAlignmentVector;
+
+} // namespace BamTools
+
+#endif // BAMALIGNMENT_H
diff --git a/src/utils/BamTools/include/api/BamAux.h b/src/utils/BamTools/include/api/BamAux.h
new file mode 100644
index 0000000000000000000000000000000000000000..d171e7069db998da82ac346689cea68fee068611
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamAux.h
@@ -0,0 +1,457 @@
+// ***************************************************************************
+// BamAux.h (c) 2009 Derek Barnett, Michael Str�mberg
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 4 March 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides data structures & utility methods that are used throughout the API.
+// ***************************************************************************
+
+#ifndef BAMAUX_H
+#define BAMAUX_H
+
+#include <api/api_global.h>
+#include <fstream>
+#include <iostream>
+#include <string>
+#include <vector>
+
+/*! \file BamAux.h
+
+ Provides data structures & utility methods that are used throughout the API.
+*/
+/*! \namespace BamTools
+ \brief Contains all BamTools classes & methods.
+
+ The BamTools API contained in this namespace contains classes and methods
+ for reading, writing, and manipulating BAM alignment files.
+*/
+namespace BamTools {
+
+// ----------------------------------------------------------------
+// CigarOp
+
+/*! \struct BamTools::CigarOp
+ \brief Represents a CIGAR alignment operation.
+
+ \sa http://samtools.sourceforge.net/SAM-1.3.pdf for more details on using CIGAR operations.
+*/
+struct API_EXPORT CigarOp {
+
+ char Type; //!< CIGAR operation type (MIDNSHP)
+ uint32_t Length; //!< CIGAR operation length (number of bases)
+
+ //! constructor
+ CigarOp(const char type = '\0',
+ const uint32_t& length = 0)
+ : Type(type)
+ , Length(length)
+ { }
+};
+
+// ----------------------------------------------------------------
+// RefData
+
+/*! \struct BamTools::RefData
+ \brief Represents a reference sequence entry
+*/
+struct API_EXPORT RefData {
+
+ std::string RefName; //!< name of reference sequence
+ int32_t RefLength; //!< length of reference sequence
+
+ //! constructor
+ RefData(const std::string& name = "",
+ const int32_t& length = 0)
+ : RefName(name)
+ , RefLength(length)
+ { }
+};
+
+//! convenience typedef for vector of RefData entries
+typedef std::vector<RefData> RefVector;
+
+// ----------------------------------------------------------------
+// BamRegion
+
+/*! \struct BamTools::BamRegion
+ \brief Represents a sequential genomic region
+
+ Allowed to span multiple (sequential) references.
+*/
+struct API_EXPORT BamRegion {
+
+ int LeftRefID; //!< reference ID for region's left boundary
+ int LeftPosition; //!< position for region's left boundary
+ int RightRefID; //!< reference ID for region's right boundary
+ int RightPosition; //!< position for region's right boundary
+
+ //! constructor
+ BamRegion(const int& leftID = -1,
+ const int& leftPos = -1,
+ const int& rightID = -1,
+ const int& rightPos = -1)
+ : LeftRefID(leftID)
+ , LeftPosition(leftPos)
+ , RightRefID(rightID)
+ , RightPosition(rightPos)
+ { }
+
+ //! copy constructor
+ BamRegion(const BamRegion& other)
+ : LeftRefID(other.LeftRefID)
+ , LeftPosition(other.LeftPosition)
+ , RightRefID(other.RightRefID)
+ , RightPosition(other.RightPosition)
+ { }
+
+ //! Clears region boundaries
+ void clear(void) {
+ LeftRefID = -1; LeftPosition = -1;
+ RightRefID = -1; RightPosition = -1;
+ }
+
+ //! Returns true if region has a left boundary
+ bool isLeftBoundSpecified(void) const {
+ return ( LeftRefID >= 0 && LeftPosition >= 0 );
+ }
+
+ //! Returns true if region boundaries are not defined
+ bool isNull(void) const {
+ return ( !isLeftBoundSpecified() && !isRightBoundSpecified() );
+ }
+
+ //! Returns true if region has a right boundary
+ bool isRightBoundSpecified(void) const {
+ return ( RightRefID >= 0 && RightPosition >= 0 );
+ }
+};
+
+// ----------------------------------------------------------------
+// General utility methods
+
+/*! \fn bool FileExists(const std::string& filename)
+ \brief checks if file exists
+
+ Attempts to open file in a read-only mode.
+
+ \return \c true if file can be opened successfully
+*/
+API_EXPORT inline bool FileExists(const std::string& filename) {
+ std::ifstream f(filename.c_str(), std::ifstream::in);
+ return !f.fail();
+}
+
+/*! \fn void SwapEndian_16(int16_t& x)
+ \brief swaps endianness of signed 16-bit integer, in place
+
+ Swaps endian representation of value in \a x.
+*/
+API_EXPORT inline void SwapEndian_16(int16_t& x) {
+ x = ((x >> 8) | (x << 8));
+}
+
+/*! \fn void SwapEndian_16(uint16_t& x)
+ \brief swaps endianness of unsigned 16-bit integer, in place
+
+ Swaps endian representation of value in \a x.
+*/
+API_EXPORT inline void SwapEndian_16(uint16_t& x) {
+ x = ((x >> 8) | (x << 8));
+}
+
+/*! \fn void SwapEndian_32(int32_t& x)
+ \brief swaps endianness of signed 32-bit integer, in place
+
+ Swaps endian representation of value in \a x.
+*/
+API_EXPORT inline void SwapEndian_32(int32_t& x) {
+ x = ( (x >> 24) |
+ ((x << 8) & 0x00FF0000) |
+ ((x >> 8) & 0x0000FF00) |
+ (x << 24)
+ );
+}
+
+/*! \fn void SwapEndian_32(uint32_t& x)
+ \brief swaps endianness of unsigned 32-bit integer, in place
+
+ Swaps endian representation of value in \a x.
+*/
+API_EXPORT inline void SwapEndian_32(uint32_t& x) {
+ x = ( (x >> 24) |
+ ((x << 8) & 0x00FF0000) |
+ ((x >> 8) & 0x0000FF00) |
+ (x << 24)
+ );
+}
+
+/*! \fn void SwapEndian_64(int64_t& x)
+ \brief swaps endianness of signed 64-bit integer, in place
+
+ Swaps endian representation of value in \a x.
+*/
+API_EXPORT inline void SwapEndian_64(int64_t& x) {
+ x = ( (x >> 56) |
+ ((x << 40) & 0x00FF000000000000ll) |
+ ((x << 24) & 0x0000FF0000000000ll) |
+ ((x << 8) & 0x000000FF00000000ll) |
+ ((x >> 8) & 0x00000000FF000000ll) |
+ ((x >> 24) & 0x0000000000FF0000ll) |
+ ((x >> 40) & 0x000000000000FF00ll) |
+ (x << 56)
+ );
+}
+
+/*! \fn void SwapEndian_64(uint64_t& x)
+ \brief swaps endianness of unsigned 64-bit integer, in place
+
+ Swaps endian representation of value in \a x.
+*/
+API_EXPORT inline void SwapEndian_64(uint64_t& x) {
+ x = ( (x >> 56) |
+ ((x << 40) & 0x00FF000000000000ll) |
+ ((x << 24) & 0x0000FF0000000000ll) |
+ ((x << 8) & 0x000000FF00000000ll) |
+ ((x >> 8) & 0x00000000FF000000ll) |
+ ((x >> 24) & 0x0000000000FF0000ll) |
+ ((x >> 40) & 0x000000000000FF00ll) |
+ (x << 56)
+ );
+}
+
+/*! \fn void SwapEndian_16p(char* data)
+ \brief swaps endianness of the next 2 bytes in a buffer, in place
+
+ Swaps endian representation the next 2 bytes in \a data.
+*/
+API_EXPORT inline void SwapEndian_16p(char* data) {
+ uint16_t& value = (uint16_t&)*data;
+ SwapEndian_16(value);
+}
+
+/*! \fn void SwapEndian_32p(char* data)
+ \brief swaps endianness of the next 4 bytes in a buffer, in place
+
+ Swaps endian representation the next 4 bytes in \a data.
+*/
+API_EXPORT inline void SwapEndian_32p(char* data) {
+ uint32_t& value = (uint32_t&)*data;
+ SwapEndian_32(value);
+}
+
+/*! \fn void SwapEndian_64p(char* data)
+ \brief swaps endianness of the next 8 bytes in a buffer, in place
+
+ Swaps endian representation the next 8 bytes in \a data.
+*/
+API_EXPORT inline void SwapEndian_64p(char* data) {
+ uint64_t& value = (uint64_t&)*data;
+ SwapEndian_64(value);
+}
+
+/*! \fn bool SystemIsBigEndian(void)
+ \brief checks host architecture's byte order
+ \return \c true if system uses big-endian ordering
+*/
+API_EXPORT inline bool SystemIsBigEndian(void) {
+ const uint16_t one = 0x0001;
+ return ((*(char*) &one) == 0 );
+}
+
+/*! \fn void PackUnsignedInt(char* buffer, unsigned int value)
+ \brief stores unsigned integer value in a byte buffer
+
+ \param buffer destination buffer
+ \param value unsigned integer to 'pack' in buffer
+*/
+API_EXPORT inline void PackUnsignedInt(char* buffer, unsigned int value) {
+ buffer[0] = (char)value;
+ buffer[1] = (char)(value >> 8);
+ buffer[2] = (char)(value >> 16);
+ buffer[3] = (char)(value >> 24);
+}
+
+/*! \fn void PackUnsignedShort(char* buffer, unsigned short value)
+ \brief stores unsigned short integer value in a byte buffer
+
+ \param buffer destination buffer
+ \param value unsigned short integer to 'pack' in buffer
+*/
+API_EXPORT inline void PackUnsignedShort(char* buffer, unsigned short value) {
+ buffer[0] = (char)value;
+ buffer[1] = (char)(value >> 8);
+}
+
+/*! \fn double UnpackDouble(const char* buffer)
+ \brief reads a double value from byte buffer
+
+ \param buffer source byte buffer
+ \return the (double) value read from the buffer
+*/
+API_EXPORT inline double UnpackDouble(const char* buffer) {
+ union { double value; unsigned char valueBuffer[sizeof(double)]; } un;
+ un.value = 0;
+ un.valueBuffer[0] = buffer[0];
+ un.valueBuffer[1] = buffer[1];
+ un.valueBuffer[2] = buffer[2];
+ un.valueBuffer[3] = buffer[3];
+ un.valueBuffer[4] = buffer[4];
+ un.valueBuffer[5] = buffer[5];
+ un.valueBuffer[6] = buffer[6];
+ un.valueBuffer[7] = buffer[7];
+ return un.value;
+}
+
+/*! \fn double UnpackDouble(char* buffer)
+ \brief reads a double value from byte buffer
+
+ This is an overloaded function.
+
+ \param buffer source byte buffer
+ \return the (double) value read from the buffer
+*/
+API_EXPORT inline double UnpackDouble(char* buffer) {
+ return UnpackDouble( (const char*)buffer );
+}
+
+/*! \fn double UnpackFloat(const char* buffer)
+ \brief reads a float value from byte buffer
+
+ \param buffer source byte buffer
+ \return the (float) value read from the buffer
+*/
+API_EXPORT inline float UnpackFloat(const char* buffer) {
+ union { float value; unsigned char valueBuffer[sizeof(float)]; } un;
+ un.value = 0;
+ un.valueBuffer[0] = buffer[0];
+ un.valueBuffer[1] = buffer[1];
+ un.valueBuffer[2] = buffer[2];
+ un.valueBuffer[3] = buffer[3];
+ return un.value;
+}
+
+/*! \fn double UnpackFloat(char* buffer)
+ \brief reads a float value from byte buffer
+
+ This is an overloaded function.
+
+ \param buffer source byte buffer
+ \return the (float) value read from the buffer
+*/
+API_EXPORT inline float UnpackFloat(char* buffer) {
+ return UnpackFloat( (const char*)buffer );
+}
+
+/*! \fn signed int UnpackSignedInt(const char* buffer)
+ \brief reads a signed integer value from byte buffer
+
+ \param buffer source byte buffer
+ \return the (signed int) value read from the buffer
+*/
+API_EXPORT inline signed int UnpackSignedInt(const char* buffer) {
+ union { signed int value; unsigned char valueBuffer[sizeof(signed int)]; } un;
+ un.value = 0;
+ un.valueBuffer[0] = buffer[0];
+ un.valueBuffer[1] = buffer[1];
+ un.valueBuffer[2] = buffer[2];
+ un.valueBuffer[3] = buffer[3];
+ return un.value;
+}
+
+/*! \fn signed int UnpackSignedInt(char* buffer)
+ \brief reads a signed integer value from byte buffer
+
+ This is an overloaded function.
+
+ \param buffer source byte buffer
+ \return the (signed int) value read from the buffer
+*/
+API_EXPORT inline signed int UnpackSignedInt(char* buffer) {
+ return UnpackSignedInt( (const char*) buffer );
+}
+
+/*! \fn signed short UnpackSignedShort(const char* buffer)
+ \brief reads a signed short integer value from byte buffer
+
+ \param buffer source byte buffer
+ \return the (signed short) value read from the buffer
+*/
+API_EXPORT inline signed short UnpackSignedShort(const char* buffer) {
+ union { signed short value; unsigned char valueBuffer[sizeof(signed short)]; } un;
+ un.value = 0;
+ un.valueBuffer[0] = buffer[0];
+ un.valueBuffer[1] = buffer[1];
+ return un.value;
+}
+
+/*! \fn signed short UnpackSignedShort(char* buffer)
+ \brief reads a signed short integer value from byte buffer
+
+ This is an overloaded function.
+
+ \param buffer source byte buffer
+ \return the (signed short) value read from the buffer
+*/
+API_EXPORT inline signed short UnpackSignedShort(char* buffer) {
+ return UnpackSignedShort( (const char*)buffer );
+}
+
+/*! \fn unsigned int UnpackUnsignedInt(const char* buffer)
+ \brief reads an unsigned integer value from byte buffer
+
+ \param buffer source byte buffer
+ \return the (unsigned int) value read from the buffer
+*/
+API_EXPORT inline unsigned int UnpackUnsignedInt(const char* buffer) {
+ union { unsigned int value; unsigned char valueBuffer[sizeof(unsigned int)]; } un;
+ un.value = 0;
+ un.valueBuffer[0] = buffer[0];
+ un.valueBuffer[1] = buffer[1];
+ un.valueBuffer[2] = buffer[2];
+ un.valueBuffer[3] = buffer[3];
+ return un.value;
+}
+
+/*! \fn unsigned int UnpackUnsignedInt(char* buffer)
+ \brief reads an unsigned integer value from byte buffer
+
+ This is an overloaded function.
+
+ \param buffer source byte buffer
+ \return the (unsigned int) value read from the buffer
+*/
+API_EXPORT inline unsigned int UnpackUnsignedInt(char* buffer) {
+ return UnpackUnsignedInt( (const char*)buffer );
+}
+
+/*! \fn unsigned short UnpackUnsignedShort(const char* buffer)
+ \brief reads an unsigned short integer value from byte buffer
+
+ \param buffer source byte buffer
+ \return the (unsigned short) value read from the buffer
+*/
+API_EXPORT inline unsigned short UnpackUnsignedShort(const char* buffer) {
+ union { unsigned short value; unsigned char valueBuffer[sizeof(unsigned short)]; } un;
+ un.value = 0;
+ un.valueBuffer[0] = buffer[0];
+ un.valueBuffer[1] = buffer[1];
+ return un.value;
+}
+
+/*! \fn unsigned short UnpackUnsignedShort(char* buffer)
+ \brief reads an unsigned short integer value from byte buffer
+
+ This is an overloaded function.
+
+ \param buffer source byte buffer
+ \return the (unsigned short) value read from the buffer
+*/
+API_EXPORT inline unsigned short UnpackUnsignedShort(char* buffer) {
+ return UnpackUnsignedShort( (const char*)buffer );
+}
+
+} // namespace BamTools
+
+#endif // BAMAUX_H
diff --git a/src/utils/BamTools/include/api/BamConstants.h b/src/utils/BamTools/include/api/BamConstants.h
new file mode 100644
index 0000000000000000000000000000000000000000..e433c8e79df37d318a6f4a9c7b94b060494b9a92
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamConstants.h
@@ -0,0 +1,128 @@
+// ***************************************************************************
+// BamConstants.h (c) 2011 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides basic constants for handling BAM files.
+// ***************************************************************************
+
+#ifndef BAM_CONSTANTS_H
+#define BAM_CONSTANTS_H
+
+#include <string>
+
+/*! \namespace BamTools::Constants
+ \brief Provides basic constants for handling BAM files.
+*/
+
+namespace BamTools {
+namespace Constants {
+
+const int BAM_SIZEOF_INT = 4;
+
+// header magic number
+const char* const BAM_HEADER_MAGIC = "BAM\1";
+const unsigned int BAM_HEADER_MAGIC_LENGTH = 4;
+
+// BAM alignment core size
+const int BAM_CORE_SIZE = 32;
+const int BAM_CORE_BUFFER_SIZE = 8;
+
+// BAM alignment flags
+const int BAM_ALIGNMENT_PAIRED = 0x0001;
+const int BAM_ALIGNMENT_PROPER_PAIR = 0x0002;
+const int BAM_ALIGNMENT_UNMAPPED = 0x0004;
+const int BAM_ALIGNMENT_MATE_UNMAPPED = 0x0008;
+const int BAM_ALIGNMENT_REVERSE_STRAND = 0x0010;
+const int BAM_ALIGNMENT_MATE_REVERSE_STRAND = 0x0020;
+const int BAM_ALIGNMENT_READ_1 = 0x0040;
+const int BAM_ALIGNMENT_READ_2 = 0x0080;
+const int BAM_ALIGNMENT_SECONDARY = 0x0100;
+const int BAM_ALIGNMENT_QC_FAILED = 0x0200;
+const int BAM_ALIGNMENT_DUPLICATE = 0x0400;
+
+// CIGAR constants
+const char* const BAM_CIGAR_LOOKUP = "MIDNSHP=X";
+const int BAM_CIGAR_MATCH = 0;
+const int BAM_CIGAR_INS = 1;
+const int BAM_CIGAR_DEL = 2;
+const int BAM_CIGAR_REFSKIP = 3;
+const int BAM_CIGAR_SOFTCLIP = 4;
+const int BAM_CIGAR_HARDCLIP = 5;
+const int BAM_CIGAR_PAD = 6;
+const int BAM_CIGAR_SEQMATCH = 7;
+const int BAM_CIGAR_MISMATCH = 8;
+
+const char BAM_CIGAR_MATCH_CHAR = 'M';
+const char BAM_CIGAR_INS_CHAR = 'I';
+const char BAM_CIGAR_DEL_CHAR = 'D';
+const char BAM_CIGAR_REFSKIP_CHAR = 'N';
+const char BAM_CIGAR_SOFTCLIP_CHAR = 'S';
+const char BAM_CIGAR_HARDCLIP_CHAR = 'H';
+const char BAM_CIGAR_PAD_CHAR = 'P';
+const char BAM_CIGAR_SEQMATCH_CHAR = '=';
+const char BAM_CIGAR_MISMATCH_CHAR = 'X';
+
+const int BAM_CIGAR_SHIFT = 4;
+const int BAM_CIGAR_MASK = ((1 << BAM_CIGAR_SHIFT) - 1);
+
+// BAM tag types
+const char BAM_TAG_TYPE_ASCII = 'A';
+const char BAM_TAG_TYPE_UINT8 = 'c';
+const char BAM_TAG_TYPE_INT8 = 'C';
+const char BAM_TAG_TYPE_UINT16 = 's';
+const char BAM_TAG_TYPE_INT16 = 'S';
+const char BAM_TAG_TYPE_UINT32 = 'i';
+const char BAM_TAG_TYPE_INT32 = 'I';
+const char BAM_TAG_TYPE_FLOAT = 'f';
+const char BAM_TAG_TYPE_STRING = 'Z';
+const char BAM_TAG_TYPE_HEX = 'H';
+const char BAM_TAG_TYPE_ARRAY = 'B';
+
+const size_t BAM_TAG_TAGSIZE = 2;
+const size_t BAM_TAG_TYPESIZE = 1;
+const int BAM_TAG_ARRAYBASE_SIZE = 8;
+
+// DNA bases
+const char* const BAM_DNA_LOOKUP = "=ACMGRSVTWYHKDBN";
+const unsigned char BAM_BASECODE_EQUAL = 0;
+const unsigned char BAM_BASECODE_A = 1;
+const unsigned char BAM_BASECODE_C = 2;
+const unsigned char BAM_BASECODE_G = 4;
+const unsigned char BAM_BASECODE_T = 8;
+const unsigned char BAM_BASECODE_N = 15;
+
+const char BAM_DNA_EQUAL = '=';
+const char BAM_DNA_A = 'A';
+const char BAM_DNA_C = 'C';
+const char BAM_DNA_G = 'G';
+const char BAM_DNA_T = 'T';
+const char BAM_DNA_N = 'N';
+const char BAM_DNA_DEL = '-';
+const char BAM_DNA_PAD = '*';
+
+// zlib constants
+const int GZIP_ID1 = 31;
+const int GZIP_ID2 = 139;
+const int CM_DEFLATE = 8;
+const int FLG_FEXTRA = 4;
+const int OS_UNKNOWN = 255;
+const int BGZF_XLEN = 6;
+const int BGZF_ID1 = 66;
+const int BGZF_ID2 = 67;
+const int BGZF_LEN = 2;
+const int GZIP_WINDOW_BITS = -15;
+const int Z_DEFAULT_MEM_LEVEL = 8;
+
+// BZGF constants
+const int BGZF_BLOCK_HEADER_LENGTH = 18;
+const int BGZF_BLOCK_FOOTER_LENGTH = 8;
+const int BGZF_MAX_BLOCK_SIZE = 65536;
+const int BGZF_DEFAULT_BLOCK_SIZE = 65536;
+
+} // namespace Constants
+} // namespace BamTools
+
+#endif // BAM_CONSTANTS_H
diff --git a/src/utils/BamTools/include/api/BamIndex.h b/src/utils/BamTools/include/api/BamIndex.h
new file mode 100644
index 0000000000000000000000000000000000000000..00a8f0174458f542268944d30aa39997400d0338
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamIndex.h
@@ -0,0 +1,80 @@
+// ***************************************************************************
+// BamIndex.h (c) 2009 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 5 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides basic BAM index interface
+// ***************************************************************************
+
+#ifndef BAM_INDEX_H
+#define BAM_INDEX_H
+
+#include <api/api_global.h>
+#include <api/BamAux.h>
+#include <string>
+
+namespace BamTools {
+
+namespace Internal {
+ class BamReaderPrivate;
+} // namespace Internal
+
+/*! \class BamTools::BamIndex
+ \brief Provides methods for generating & loading BAM index files.
+
+ This class straddles the line between public API and internal
+ implementation detail. Most client code should never have to use this
+ class directly.
+
+ It is exposed to the public API to allow advanced users to implement
+ their own custom indexing schemes.
+
+ More documentation on methods & enums coming soon.
+*/
+
+class API_EXPORT BamIndex {
+
+ // enums
+ public:
+ // specify index-caching behavior
+ enum IndexCacheMode { FullIndexCaching = 0 // store entire index file contents in memory
+ , LimitedIndexCaching // store only index data for current reference
+ , NoIndexCaching // do not store any index data between jumps
+ };
+
+ // list of supported BamIndex types
+ enum IndexType { BAMTOOLS = 0
+ , STANDARD
+ };
+
+ // ctor & dtor
+ public:
+ BamIndex(Internal::BamReaderPrivate* reader) : m_reader(reader) { }
+ virtual ~BamIndex(void) { }
+
+ // index interface
+ public:
+ // builds index from associated BAM file & writes out to index file
+ virtual bool Create(void) =0; // creates index file from BAM file
+ // returns whether reference has alignments or no
+ virtual bool HasAlignments(const int& referenceID) const =0;
+ // attempts to use index data to jump to @region, returns success/fail
+ // a "successful" jump indicates no error, but not whether this region has data
+ // * thus, the method sets a flag to indicate whether there are alignments
+ // available after the jump position
+ virtual bool Jump(const BamTools::BamRegion& region, bool* hasAlignmentsInRegion) =0;
+ // loads existing data from file into memory
+ virtual bool Load(const std::string& filename) =0;
+ // change the index caching behavior
+ virtual void SetCacheMode(const BamIndex::IndexCacheMode& mode) =0;
+
+ // data members
+ protected:
+ Internal::BamReaderPrivate* m_reader; // copy, not ownedprivate:
+};
+
+} // namespace BamTools
+
+#endif // BAM_INDEX_H
diff --git a/src/utils/BamTools/include/api/BamMultiReader.h b/src/utils/BamTools/include/api/BamMultiReader.h
new file mode 100644
index 0000000000000000000000000000000000000000..cc49ec8eb2076469be2585bdf978bd603acec8a3
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamMultiReader.h
@@ -0,0 +1,127 @@
+// ***************************************************************************
+// BamMultiReader.h (c) 2010 Erik Garrison, Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 15 March 2011 (DB)
+// ---------------------------------------------------------------------------
+// Convenience class for reading multiple BAM files.
+// ***************************************************************************
+
+#ifndef BAMMULTIREADER_H
+#define BAMMULTIREADER_H
+
+#include <api/api_global.h>
+#include <api/BamReader.h>
+#include <map>
+#include <sstream>
+#include <string>
+#include <utility>
+
+namespace BamTools {
+
+namespace Internal {
+ class BamMultiReaderPrivate;
+} // namespace Internal
+
+class API_EXPORT BamMultiReader {
+
+ public:
+ enum SortOrder { SortedByPosition = 0
+ , SortedByReadName
+ , Unsorted
+ };
+
+ // constructor / destructor
+ public:
+ BamMultiReader(void);
+ ~BamMultiReader(void);
+
+ // public interface
+ public:
+
+ // ----------------------
+ // BAM file operations
+ // ----------------------
+
+ // closes all open BAM files
+ void Close(void);
+ // close only the requested BAM file
+ void CloseFile(const std::string& filename);
+ // returns list of filenames for all open BAM files
+ const std::vector<std::string> Filenames(void) const;
+ // returns true if multireader has any open BAM files
+ bool HasOpenReaders(void) const;
+ // performs random-access jump within current BAM files
+ bool Jump(int refID, int position = 0);
+ // opens BAM files
+ bool Open(const std::vector<std::string>& filenames);
+ // opens a single BAM file, adding to any other current BAM files
+ bool OpenFile(const std::string& filename);
+ // returns file pointers to beginning of alignments
+ bool Rewind(void);
+ // sets the target region of interest
+ bool SetRegion(const BamRegion& region);
+ // sets the target region of interest
+ bool SetRegion(const int& leftRefID,
+ const int& leftPosition,
+ const int& rightRefID,
+ const int& rightPosition);
+
+ // ----------------------
+ // access alignment data
+ // ----------------------
+
+ // retrieves next available alignment
+ bool GetNextAlignment(BamAlignment& alignment);
+ // retrieves next available alignmnet (without populating the alignment's string data fields)
+ bool GetNextAlignmentCore(BamAlignment& alignment);
+
+ // sets the expected sorting order for reading across multiple BAM files
+ void SetSortOrder(const SortOrder& order);
+
+ // ----------------------
+ // access auxiliary data
+ // ----------------------
+
+ // returns unified SAM header for all files
+ SamHeader GetHeader(void) const;
+ // returns unified SAM header text for all files
+ std::string GetHeaderText(void) const;
+ // returns number of reference sequences
+ int GetReferenceCount(void) const;
+ // returns all reference sequence entries.
+ const BamTools::RefVector GetReferenceData(void) const;
+ // returns the ID of the reference with this name.
+ int GetReferenceID(const std::string& refName) const;
+
+ // ----------------------
+ // BAM index operations
+ // ----------------------
+
+ // creates index files for current BAM files
+ bool CreateIndexes(const BamIndex::IndexType& type = BamIndex::STANDARD);
+ // returns true if all BAM files have index data available
+ bool HasIndexes(void) const;
+ // looks for index files that match current BAM files
+ bool LocateIndexes(const BamIndex::IndexType& preferredType = BamIndex::STANDARD);
+ // opens index files for current BAM files.
+ bool OpenIndexes(const std::vector<std::string>& indexFilenames);
+ // changes the caching behavior of the index data
+ void SetIndexCacheMode(const BamIndex::IndexCacheMode& mode);
+
+ // deprecated methods
+ public:
+ // returns \c true if all BAM files have index data available.
+ bool IsIndexLoaded(void) const;
+ // convenience method for printing filenames to stdout
+ void PrintFilenames(void) const;
+
+ // private implementation
+ private:
+ Internal::BamMultiReaderPrivate* d;
+};
+
+} // namespace BamTools
+
+#endif // BAMMULTIREADER_H
diff --git a/src/utils/BamTools/include/api/BamReader.h b/src/utils/BamTools/include/api/BamReader.h
new file mode 100644
index 0000000000000000000000000000000000000000..85b0c0d5713ed15f4a616ec4a6f7e1c2ab9d3c1b
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamReader.h
@@ -0,0 +1,118 @@
+// ***************************************************************************
+// BamReader.h (c) 2009 Derek Barnett, Michael Str�mberg
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 4 March 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides read access to BAM files.
+// ***************************************************************************
+
+#ifndef BAMREADER_H
+#define BAMREADER_H
+
+#include <api/api_global.h>
+#include <api/BamAlignment.h>
+#include <api/BamIndex.h>
+#include <api/SamHeader.h>
+#include <string>
+
+namespace BamTools {
+
+namespace Internal {
+ class BamReaderPrivate;
+} // namespace Internal
+
+class API_EXPORT BamReader {
+
+ // constructor / destructor
+ public:
+ BamReader(void);
+ ~BamReader(void);
+
+ // public interface
+ public:
+
+ // ----------------------
+ // BAM file operations
+ // ----------------------
+
+ // closes the current BAM file
+ void Close(void);
+ // returns filename of current BAM file
+ const std::string GetFilename(void) const;
+ // returns true if a BAM file is open for reading
+ bool IsOpen(void) const;
+ // performs random-access jump within BAM file
+ bool Jump(int refID, int position = 0);
+ // opens a BAM file
+ bool Open(const std::string& filename);
+ // returns internal file pointer to beginning of alignment data
+ bool Rewind(void);
+ // sets the target region of interest
+ bool SetRegion(const BamRegion& region);
+ // sets the target region of interest
+ bool SetRegion(const int& leftRefID,
+ const int& leftPosition,
+ const int& rightRefID,
+ const int& rightPosition);
+
+ // ----------------------
+ // access alignment data
+ // ----------------------
+
+ // retrieves next available alignment
+ bool GetNextAlignment(BamAlignment& alignment);
+ // retrieves next available alignmnet (without populating the alignment's string data fields)
+ bool GetNextAlignmentCore(BamAlignment& alignment);
+
+ // ----------------------
+ // access header data
+ // ----------------------
+
+ // returns SAM header data
+ SamHeader GetHeader(void) const;
+ // returns SAM header data, as SAM-formatted text
+ std::string GetHeaderText(void) const;
+
+ // ----------------------
+ // access reference data
+ // ----------------------
+
+ // returns the number of reference sequences
+ int GetReferenceCount(void) const;
+ // returns all reference sequence entries
+ const RefVector& GetReferenceData(void) const;
+ // returns the ID of the reference with this name
+ int GetReferenceID(const std::string& refName) const;
+
+ // ----------------------
+ // BAM index operations
+ // ----------------------
+
+ // creates an index file for current BAM file, using the requested index type
+ bool CreateIndex(const BamIndex::IndexType& type = BamIndex::STANDARD);
+ // returns true if index data is available
+ bool HasIndex(void) const;
+ // looks in BAM file's directory for a matching index file
+ bool LocateIndex(const BamIndex::IndexType& preferredType = BamIndex::STANDARD);
+ // opens a BAM index file
+ bool OpenIndex(const std::string& indexFilename);
+ // sets a custom BamIndex on this reader
+ void SetIndex(BamIndex* index);
+ // changes the caching behavior of the index data
+ void SetIndexCacheMode(const BamIndex::IndexCacheMode& mode);
+
+ // deprecated methods
+ public:
+ // returns true if index data is available
+ bool IsIndexLoaded(void) const;
+
+ // private implementation
+ private:
+ Internal::BamReaderPrivate* d;
+};
+
+} // namespace BamTools
+
+#endif // BAMREADER_H
diff --git a/src/utils/BamTools/include/api/BamWriter.h b/src/utils/BamTools/include/api/BamWriter.h
new file mode 100644
index 0000000000000000000000000000000000000000..476dbecf271e35ea8db22883055faa69c62babf0
--- /dev/null
+++ b/src/utils/BamTools/include/api/BamWriter.h
@@ -0,0 +1,64 @@
+// ***************************************************************************
+// BamWriter.h (c) 2009 Michael Str�mberg, Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 4 March 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides the basic functionality for producing BAM files
+// ***************************************************************************
+
+#ifndef BAMWRITER_H
+#define BAMWRITER_H
+
+#include <api/api_global.h>
+#include <api/BamAux.h>
+#include <string>
+
+namespace BamTools {
+
+class BamAlignment;
+class SamHeader;
+
+namespace Internal {
+ class BamWriterPrivate;
+} // namespace Internal
+
+class API_EXPORT BamWriter {
+
+ public: enum CompressionMode { Compressed = 0
+ , Uncompressed
+ };
+
+ // ctor & dtor
+ public:
+ BamWriter(void);
+ ~BamWriter(void);
+
+ // public interface
+ public:
+ // closes the current BAM file
+ void Close(void);
+ // returns true if BAM file is open for writing
+ bool IsOpen(void) const;
+ // opens a BAM file for writing
+ bool Open(const std::string& filename,
+ const std::string& samHeaderText,
+ const RefVector& referenceSequences);
+ // opens a BAM file for writing
+ bool Open(const std::string& filename,
+ const SamHeader& samHeader,
+ const RefVector& referenceSequences);
+ // saves the alignment to the alignment archive
+ void SaveAlignment(const BamAlignment& alignment);
+ // sets the output compression mode
+ void SetCompressionMode(const CompressionMode& compressionMode);
+
+ // private implementation
+ private:
+ Internal::BamWriterPrivate* d;
+};
+
+} // namespace BamTools
+
+#endif // BAMWRITER_H
diff --git a/src/utils/BamTools/include/api/SamConstants.h b/src/utils/BamTools/include/api/SamConstants.h
new file mode 100644
index 0000000000000000000000000000000000000000..d34592027891a8233e6e27b6b4f2cb8a84a8f9a6
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamConstants.h
@@ -0,0 +1,96 @@
+// ***************************************************************************
+// SamConstants.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides constants for SAM header
+// ***************************************************************************
+
+#ifndef SAM_CONSTANTS_H
+#define SAM_CONSTANTS_H
+
+#include <api/api_global.h>
+#include <string>
+
+namespace BamTools {
+namespace Constants {
+
+// basic char constants used in SAM format
+const char SAM_COLON = ':';
+const char SAM_EQUAL = '=';
+const char SAM_PERIOD = '.';
+const char SAM_STAR = '*';
+const char SAM_TAB = '\t';
+const std::string SAM_DIGITS = "0123456789";
+
+// HD entries
+const std::string SAM_HD_BEGIN_TOKEN = "@HD";
+const std::string SAM_HD_VERSION_TAG = "VN";
+const std::string SAM_HD_SORTORDER_TAG = "SO";
+const std::string SAM_HD_GROUPORDER_TAG = "GO";
+
+// SQ entries
+const std::string SAM_SQ_BEGIN_TOKEN = "@SQ";
+const std::string SAM_SQ_ASSEMBLYID_TAG = "AS";
+const std::string SAM_SQ_CHECKSUM_TAG = "M5";
+const std::string SAM_SQ_LENGTH_TAG = "LN";
+const std::string SAM_SQ_NAME_TAG = "SN";
+const std::string SAM_SQ_SPECIES_TAG = "SP";
+const std::string SAM_SQ_URI_TAG = "UR";
+
+// RG entries
+const std::string SAM_RG_BEGIN_TOKEN = "@RG";
+const std::string SAM_RG_DESCRIPTION_TAG = "DS";
+const std::string SAM_RG_FLOWORDER_TAG = "FO";
+const std::string SAM_RG_ID_TAG = "ID";
+const std::string SAM_RG_KEYSEQUENCE_TAG = "KS";
+const std::string SAM_RG_LIBRARY_TAG = "LB";
+const std::string SAM_RG_PLATFORMUNIT_TAG = "PU";
+const std::string SAM_RG_PREDICTEDINSERTSIZE_TAG = "PI";
+const std::string SAM_RG_PRODUCTIONDATE_TAG = "DT";
+const std::string SAM_RG_PROGRAM_TAG = "PG";
+const std::string SAM_RG_SAMPLE_TAG = "SM";
+const std::string SAM_RG_SEQCENTER_TAG = "CN";
+const std::string SAM_RG_SEQTECHNOLOGY_TAG = "PL";
+
+// PG entries
+const std::string SAM_PG_BEGIN_TOKEN = "@PG";
+const std::string SAM_PG_COMMANDLINE_TAG = "CL";
+const std::string SAM_PG_ID_TAG = "ID";
+const std::string SAM_PG_NAME_TAG = "PN";
+const std::string SAM_PG_PREVIOUSPROGRAM_TAG = "PP";
+const std::string SAM_PG_VERSION_TAG = "VN";
+
+// CO entries
+const std::string SAM_CO_BEGIN_TOKEN = "@CO";
+
+// HD:SO values
+const std::string SAM_HD_SORTORDER_COORDINATE = "coordinate";
+const std::string SAM_HD_SORTORDER_QUERYNAME = "queryname";
+const std::string SAM_HD_SORTORDER_UNKNOWN = "unknown";
+const std::string SAM_HD_SORTORDER_UNSORTED = "unsorted";
+
+// HD:GO values
+const std::string SAM_HD_GROUPORDER_NONE = "none";
+const std::string SAM_HD_GROUPORDER_QUERY = "query";
+const std::string SAM_HD_GROUPORDER_REFERENCE = "reference";
+
+// SQ:LN values
+const unsigned int SAM_SQ_LENGTH_MIN = 1;
+const unsigned int SAM_SQ_LENGTH_MAX = 536870911; // 2^29 - 1
+
+// RG:PL values
+const std::string SAM_RG_SEQTECHNOLOGY_CAPILLARY = "CAPILLARY";
+const std::string SAM_RG_SEQTECHNOLOGY_HELICOS = "HELICOS";
+const std::string SAM_RG_SEQTECHNOLOGY_ILLUMINA = "ILLUMINA";
+const std::string SAM_RG_SEQTECHNOLOGY_IONTORRENT = "IONTORRENT";
+const std::string SAM_RG_SEQTECHNOLOGY_LS454 = "LS454";
+const std::string SAM_RG_SEQTECHNOLOGY_PACBIO = "PACBIO";
+const std::string SAM_RG_SEQTECHNOLOGY_SOLID = "SOLID";
+
+} // namespace Constants
+} // namespace BamTools
+
+#endif // SAM_CONSTANTS_H
diff --git a/src/utils/BamTools/include/api/SamHeader.h b/src/utils/BamTools/include/api/SamHeader.h
new file mode 100644
index 0000000000000000000000000000000000000000..5c7a1019120f74981ad8ee3e73559d8247b569a9
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamHeader.h
@@ -0,0 +1,69 @@
+// ***************************************************************************
+// SamHeader.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 18 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM header data fields.
+// ***************************************************************************
+
+#ifndef SAM_HEADER_H
+#define SAM_HEADER_H
+
+#include <api/api_global.h>
+#include <api/SamProgramChain.h>
+#include <api/SamReadGroupDictionary.h>
+#include <api/SamSequenceDictionary.h>
+#include <string>
+#include <vector>
+
+namespace BamTools {
+
+struct API_EXPORT SamHeader {
+
+ // ctor & dtor
+ SamHeader(const std::string& headerText = "");
+ SamHeader(const SamHeader& other);
+ ~SamHeader(void);
+
+ // query/modify entire SamHeader
+ void Clear(void); // clears all header contents
+ bool IsValid(bool verbose = false) const; // returns true if SAM header is well-formed
+ void SetHeaderText(const std::string& headerText); // replaces data fields with contents of SAM-formatted text
+ std::string ToString(void) const; // returns the printable, SAM-formatted header text
+
+ // convenience query methods
+ bool HasVersion(void) const; // returns true if header contains format version entry
+ bool HasSortOrder(void) const; // returns true if header contains sort order entry
+ bool HasGroupOrder(void) const; // returns true if header contains group order entry
+ bool HasSequences(void) const; // returns true if header contains any sequence entries
+ bool HasReadGroups(void) const; // returns true if header contains any read group entries
+ bool HasPrograms(void) const; // returns true if header contains any program record entries
+ bool HasComments(void) const; // returns true if header contains comments
+
+ // --------------
+ // data members
+ // --------------
+
+ // header metadata (@HD line)
+ std::string Version; // VN:<Version> *Required for valid SAM header, if @HD record is present*
+ std::string SortOrder; // SO:<SortOrder>
+ std::string GroupOrder; // GO:<GroupOrder>
+
+ // header sequences (@SQ entries)
+ SamSequenceDictionary Sequences;
+
+ // header read groups (@RG entries)
+ SamReadGroupDictionary ReadGroups;
+
+ // header program data (@PG entries)
+ SamProgramChain Programs;
+
+ // header comments (@CO entries)
+ std::vector<std::string> Comments;
+};
+
+} // namespace BamTools
+
+#endif // SAM_HEADER_H
diff --git a/src/utils/BamTools/include/api/SamProgram.h b/src/utils/BamTools/include/api/SamProgram.h
new file mode 100644
index 0000000000000000000000000000000000000000..3c89059bdaedeeaf0b287b9c4fd3a2aae794002e
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamProgram.h
@@ -0,0 +1,62 @@
+// ***************************************************************************
+// SamProgram.h (c) 2011 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM header program records.
+// ***************************************************************************
+
+#ifndef SAM_PROGRAM_H
+#define SAM_PROGRAM_H
+
+#include "api/api_global.h"
+#include <string>
+
+namespace BamTools {
+
+class SamProgramChain;
+
+struct API_EXPORT SamProgram {
+
+ // ctor & dtor
+ SamProgram(void);
+ SamProgram(const std::string& id);
+ SamProgram(const SamProgram& other);
+ ~SamProgram(void);
+
+ // query/modify entire program record
+ void Clear(void); // clears all data fields
+
+ // convenience query methods
+ bool HasCommandLine(void) const; // returns true if program record has a command line entry
+ bool HasID(void) const; // returns true if program record has an ID
+ bool HasName(void) const; // returns true if program record has a name
+ bool HasPreviousProgramID(void) const; // returns true if program record has a 'previous program ID'
+ bool HasVersion(void) const; // returns true if program record has a version
+
+ // data members
+ std::string CommandLine; // CL:<CommandLine>
+ std::string ID; // ID:<ID> *Required for valid SAM header*
+ std::string Name; // PN:<Name>
+ std::string PreviousProgramID; // PP:<PreviousProgramID>
+ std::string Version; // VN:<Version>
+
+ // internal (non-standard) methods & fields
+ private:
+ bool HasNextProgramID(void) const;
+ std::string NextProgramID;
+ friend class BamTools::SamProgramChain;
+};
+
+/*! \fn bool operator==(const SamProgram& lhs, const SamProgram& rhs)
+ \brief tests equality by comparing program IDs
+*/
+API_EXPORT inline bool operator==(const SamProgram& lhs, const SamProgram& rhs) {
+ return lhs.ID == rhs.ID;
+}
+
+} // namespace BamTools
+
+#endif // SAM_PROGRAM_H
diff --git a/src/utils/BamTools/include/api/SamProgramChain.h b/src/utils/BamTools/include/api/SamProgramChain.h
new file mode 100644
index 0000000000000000000000000000000000000000..4cb16fc337fc2b834dded709804f8879f15968f2
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamProgramChain.h
@@ -0,0 +1,86 @@
+// ***************************************************************************
+// SamProgramChain.h (c) 2011 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides methods for operating on a SamProgram record "chain"
+// ***************************************************************************
+
+#ifndef SAM_PROGRAMCHAIN_H
+#define SAM_PROGRAMCHAIN_H
+
+#include <api/api_global.h>
+#include <api/SamProgram.h>
+#include <string>
+#include <vector>
+
+namespace BamTools {
+
+// chain is *NOT* sorted in any order
+// use First()/Last() to retrieve oldest/newest programs, respectively
+typedef std::vector<SamProgram> SamProgramContainer;
+typedef SamProgramContainer::iterator SamProgramIterator;
+typedef SamProgramContainer::const_iterator SamProgramConstIterator;
+
+class API_EXPORT SamProgramChain {
+
+ // ctor & dtor
+ public:
+ SamProgramChain(void);
+ SamProgramChain(const SamProgramChain& other);
+ ~SamProgramChain(void);
+
+ // query/modify program data
+ public:
+ // appends a program record to the chain
+ void Add(SamProgram& program);
+ void Add(std::vector<SamProgram>& programs);
+
+ // clears all read group entries
+ void Clear(void);
+
+ // returns true if chain contains this program record (matches on ID)
+ bool Contains(const SamProgram& program) const;
+ bool Contains(const std::string& programId) const;
+
+ // returns the first (oldest) program in the chain
+ SamProgram& First(void);
+ const SamProgram& First(void) const;
+
+ // returns true if chain is empty
+ bool IsEmpty(void) const;
+
+ // returns last (most recent) program in the chain
+ SamProgram& Last(void);
+ const SamProgram& Last(void) const;
+
+ // returns number of program records in the chain
+ int Size(void) const;
+
+ // retrieves a modifiable reference to the SamProgram object associated with this ID
+ SamProgram& operator[](const std::string& programId);
+
+ // retrieve STL-compatible iterators
+ public:
+ SamProgramIterator Begin(void); // returns iterator to begin()
+ SamProgramConstIterator Begin(void) const; // returns const_iterator to begin()
+ SamProgramConstIterator ConstBegin(void) const; // returns const_iterator to begin()
+ SamProgramIterator End(void); // returns iterator to end()
+ SamProgramConstIterator End(void) const; // returns const_iterator to end()
+ SamProgramConstIterator ConstEnd(void) const; // returns const_iterator to end()
+
+ // internal methods
+ private:
+ int IndexOf(const std::string& programId) const;
+ const std::string NextIdFor(const std::string& programId) const;
+
+ // data members
+ private:
+ SamProgramContainer m_data;
+};
+
+} // namespace BamTools
+
+#endif // SAM_PROGRAMCHAIN_H
diff --git a/src/utils/BamTools/include/api/SamReadGroup.h b/src/utils/BamTools/include/api/SamReadGroup.h
new file mode 100644
index 0000000000000000000000000000000000000000..b203d3cdb16927c5363272c9810b74b84bd2ecba
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamReadGroup.h
@@ -0,0 +1,69 @@
+// ***************************************************************************
+// SamReadGroup.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 18 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM read group data fields.
+// ***************************************************************************
+
+#ifndef SAM_READGROUP_H
+#define SAM_READGROUP_H
+
+#include "api/api_global.h"
+#include <string>
+
+namespace BamTools {
+
+struct API_EXPORT SamReadGroup {
+
+ // ctor & dtor
+ SamReadGroup(void);
+ SamReadGroup(const std::string& id);
+ SamReadGroup(const SamReadGroup& other);
+ ~SamReadGroup(void);
+
+ // query/modify entire read group
+ void Clear(void); // clears all data fields
+
+ // convenience query methods
+ bool HasDescription(void) const; // returns true if read group has a description
+ bool HasFlowOrder(void) const; // returns true if read group has a flow order entry
+ bool HasID(void) const; // returns true if read group has a group ID
+ bool HasKeySequence(void) const; // returns true if read group has a key sequence
+ bool HasLibrary(void) const; // returns true if read group has a library name
+ bool HasPlatformUnit(void) const; // returns true if read group has a platform unit ID
+ bool HasPredictedInsertSize(void) const; // returns true if read group has a predicted insert size
+ bool HasProductionDate(void) const; // returns true if read group has a production date
+ bool HasProgram(void) const; // returns true if read group has a program entry
+ bool HasSample(void) const; // returns true if read group has a sample name
+ bool HasSequencingCenter(void) const; // returns true if read group has a sequencing center ID
+ bool HasSequencingTechnology(void) const; // returns true if read group has a sequencing technology ID
+
+
+ // data fields
+ std::string Description; // DS:<Description>
+ std::string FlowOrder; // FO:<FlowOrder>
+ std::string ID; // ID:<ID> *Required for valid SAM header*
+ std::string KeySequence; // KS:<KeySequence>
+ std::string Library; // LB:<Library>
+ std::string PlatformUnit; // PU:<PlatformUnit>
+ std::string PredictedInsertSize; // PI:<PredictedInsertSize>
+ std::string ProductionDate; // DT:<ProductionDate>
+ std::string Program; // PG:<Program>
+ std::string Sample; // SM:<Sample>
+ std::string SequencingCenter; // CN:<SequencingCenter>
+ std::string SequencingTechnology; // PL:<SequencingTechnology>
+};
+
+/*! \fn bool operator==(const SamReadGroup& lhs, const SamReadGroup& rhs)
+ \brief tests equality by comparing read group IDs
+*/
+API_EXPORT inline bool operator==(const SamReadGroup& lhs, const SamReadGroup& rhs) {
+ return lhs.ID == rhs.ID;
+}
+
+} // namespace BamTools
+
+#endif // SAM_READGROUP_H
diff --git a/src/utils/BamTools/include/api/SamReadGroupDictionary.h b/src/utils/BamTools/include/api/SamReadGroupDictionary.h
new file mode 100644
index 0000000000000000000000000000000000000000..8ec40e227ba5f020699cd1a0021a2785587e447c
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamReadGroupDictionary.h
@@ -0,0 +1,87 @@
+// ***************************************************************************
+// SamReadGroupDictionary.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 18 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides methods for operating on a collection of SamReadGroup entries.
+// ***************************************************************************
+
+#ifndef SAM_READGROUP_DICTIONARY_H
+#define SAM_READGROUP_DICTIONARY_H
+
+#include <api/api_global.h>
+#include <api/SamReadGroup.h>
+#include <string>
+#include <vector>
+
+namespace BamTools {
+
+typedef std::vector<SamReadGroup> SamReadGroupContainer;
+typedef SamReadGroupContainer::iterator SamReadGroupIterator;
+typedef SamReadGroupContainer::const_iterator SamReadGroupConstIterator;
+
+class API_EXPORT SamReadGroupDictionary {
+
+ // ctor & dtor
+ public:
+ SamReadGroupDictionary(void);
+ SamReadGroupDictionary(const SamReadGroupDictionary& other);
+ ~SamReadGroupDictionary(void);
+
+ // query/modify read group data
+ public:
+ // adds a read group
+ void Add(const SamReadGroup& readGroup);
+ void Add(const std::string& readGroupId);
+
+ // adds multiple read groups
+ void Add(const std::vector<SamReadGroup>& readGroups);
+ void Add(const std::vector<std::string>& readGroupIds);
+
+ // clears all read group entries
+ void Clear(void);
+
+ // returns true if dictionary contains this read group
+ bool Contains(const SamReadGroup& readGroup) const;
+ bool Contains(const std::string& readGroupId) const;
+
+ // returns true if dictionary is empty
+ bool IsEmpty(void) const;
+
+ // removes read group, if found
+ void Remove(const SamReadGroup& readGroup);
+ void Remove(const std::string& readGroupId);
+
+ // removes multiple read groups
+ void Remove(const std::vector<SamReadGroup>& readGroups);
+ void Remove(const std::vector<std::string>& readGroupIds);
+
+ // returns number of read groups in dictionary
+ int Size(void) const;
+
+ // retrieves a modifiable reference to the SamReadGroup object associated with this ID
+ SamReadGroup& operator[](const std::string& readGroupId);
+
+ // retrieve STL-compatible iterators
+ public:
+ SamReadGroupIterator Begin(void); // returns iterator to begin()
+ SamReadGroupConstIterator Begin(void) const; // returns const_iterator to begin()
+ SamReadGroupConstIterator ConstBegin(void) const; // returns const_iterator to begin()
+ SamReadGroupIterator End(void); // returns iterator to end()
+ SamReadGroupConstIterator End(void) const; // returns const_iterator to end()
+ SamReadGroupConstIterator ConstEnd(void) const; // returns const_iterator to end()
+
+ // internal methods
+ private:
+ int IndexOf(const std::string& readGroupId) const;
+
+ // data members
+ private:
+ SamReadGroupContainer m_data;
+};
+
+} // namespace BamTools
+
+#endif // SAM_READGROUP_DICTIONARY_H
diff --git a/src/utils/BamTools/include/api/SamSequence.h b/src/utils/BamTools/include/api/SamSequence.h
new file mode 100644
index 0000000000000000000000000000000000000000..054e58f985ec40c74820f37a8318157202dfd2ed
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamSequence.h
@@ -0,0 +1,61 @@
+// ***************************************************************************
+// SamSequence.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 18 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM sequence data fields.
+// ***************************************************************************
+
+#ifndef SAM_SEQUENCE_H
+#define SAM_SEQUENCE_H
+
+#include <api/api_global.h>
+#include <string>
+
+namespace BamTools {
+
+struct API_EXPORT SamSequence {
+
+ // ctor & dtor
+ SamSequence(void);
+ SamSequence(const std::string& name, const int& length);
+ SamSequence(const std::string& name, const std::string& length);
+ SamSequence(const SamSequence& other);
+ ~SamSequence(void);
+
+ // query/modify entire sequence
+ void Clear(void); // clears all contents
+
+ // convenience query methods
+ bool HasAssemblyID(void) const; // returns true if sequence has an assembly ID
+ bool HasChecksum(void) const; // returns true if sequence has an MD5 checksum
+ bool HasLength(void) const; // returns true if sequence has a length
+ bool HasName(void) const; // returns true if sequence has a name
+ bool HasSpecies(void) const; // returns true if sequence has a species ID
+ bool HasURI(void) const; // returns true if sequence has a URI
+
+ // data members
+ std::string AssemblyID; // AS:<AssemblyID>
+ std::string Checksum; // M5:<Checksum>
+ std::string Length; // LN:<Length> *Required for valid SAM header*
+ std::string Name; // SN:<Name> *Required for valid SAM header*
+ std::string Species; // SP:<Species>
+ std::string URI; // UR:<URI>
+};
+
+/*! \fn bool operator==(const SamSequence& lhs, const SamSequence& rhs)
+ \brief tests equality by comparing sequence names, lengths, & checksums (if available)
+*/
+API_EXPORT inline bool operator==(const SamSequence& lhs, const SamSequence& rhs) {
+ if ( lhs.Name != rhs.Name ) return false;
+ if ( lhs.Length != rhs.Length ) return false;
+ if ( lhs.HasChecksum() && rhs.HasChecksum() )
+ return (lhs.Checksum == rhs.Checksum);
+ else return true;
+}
+
+} // namespace BamTools
+
+#endif // SAM_SEQUENCE_H
diff --git a/src/utils/BamTools/include/api/SamSequenceDictionary.h b/src/utils/BamTools/include/api/SamSequenceDictionary.h
new file mode 100644
index 0000000000000000000000000000000000000000..1ac73261fef989f3f38adf9d6a9d1db1c16701b7
--- /dev/null
+++ b/src/utils/BamTools/include/api/SamSequenceDictionary.h
@@ -0,0 +1,89 @@
+// ***************************************************************************
+// SamSequenceDictionary.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 18 April 2011
+// ---------------------------------------------------------------------------
+// Provides methods for operating on a collection of SamSequence entries.
+// ***************************************************************************
+
+#ifndef SAM_SEQUENCE_DICTIONARY_H
+#define SAM_SEQUENCE_DICTIONARY_H
+
+#include <api/api_global.h>
+#include <api/SamSequence.h>
+#include <string>
+#include <map>
+#include <vector>
+
+namespace BamTools {
+
+typedef std::vector<SamSequence> SamSequenceContainer;
+typedef SamSequenceContainer::iterator SamSequenceIterator;
+typedef SamSequenceContainer::const_iterator SamSequenceConstIterator;
+
+class API_EXPORT SamSequenceDictionary {
+
+ // ctor & dtor
+ public:
+ SamSequenceDictionary(void);
+ SamSequenceDictionary(const SamSequenceDictionary& other);
+ ~SamSequenceDictionary(void);
+
+ // query/modify sequence data
+ public:
+ // adds a sequence
+ void Add(const SamSequence& sequence);
+ void Add(const std::string& name, const int& length);
+
+ // adds multiple sequences
+ void Add(const std::vector<SamSequence>& sequences);
+ void Add(const std::map<std::string, int>& sequenceMap);
+
+ // clears all sequence entries
+ void Clear(void);
+
+ // returns true if dictionary contains this sequence
+ bool Contains(const SamSequence& sequence) const;
+ bool Contains(const std::string& sequenceName) const;
+
+ // returns true if dictionary is empty
+ bool IsEmpty(void) const;
+
+ // removes sequence, if found
+ void Remove(const SamSequence& sequence);
+ void Remove(const std::string& sequenceName);
+
+ // removes multiple sequences
+ void Remove(const std::vector<SamSequence>& sequences);
+ void Remove(const std::vector<std::string>& sequenceNames);
+
+ // returns number of sequences in dictionary
+ int Size(void) const;
+
+ // retrieves a modifiable reference to the SamSequence object associated with this name
+ SamSequence& operator[](const std::string& sequenceName);
+
+ // retrieve STL-compatible iterators
+ public:
+ SamSequenceIterator Begin(void); // returns iterator to begin()
+ SamSequenceConstIterator Begin(void) const; // returns const_iterator to begin()
+ SamSequenceConstIterator ConstBegin(void) const; // returns const_iterator to begin()
+ SamSequenceIterator End(void); // returns iterator to end()
+ SamSequenceConstIterator End(void) const; // returns const_iterator to end()
+ SamSequenceConstIterator ConstEnd(void) const; // returns const_iterator to end()
+
+ // internal methods
+ private:
+ int IndexOf(const std::string& name) const;
+
+ // data members
+ private:
+ SamSequenceContainer m_data;
+};
+
+} // namespace BamTools
+
+#endif // SAM_SEQUENCE_DICTIONARY_H
+
diff --git a/src/utils/BamTools/include/api/api_global.h b/src/utils/BamTools/include/api/api_global.h
new file mode 100644
index 0000000000000000000000000000000000000000..84fcad214c1e141d8f227dba3961e6570f03aec8
--- /dev/null
+++ b/src/utils/BamTools/include/api/api_global.h
@@ -0,0 +1,22 @@
+// ***************************************************************************
+// api_global.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 19 November 2010 (DB)
+// ---------------------------------------------------------------------------
+// Provides macros for exporting & importing BamTools API library symbols
+// ***************************************************************************
+
+#ifndef API_GLOBAL_H
+#define API_GLOBAL_H
+
+#include "shared/bamtools_global.h"
+
+#ifdef BAMTOOLS_API_LIBRARY
+# define API_EXPORT BAMTOOLS_LIBRARY_EXPORT
+#else
+# define API_EXPORT BAMTOOLS_LIBRARY_IMPORT
+#endif
+
+#endif // API_GLOBAL_H
diff --git a/src/utils/BamTools/include/shared/bamtools_global.h b/src/utils/BamTools/include/shared/bamtools_global.h
new file mode 100644
index 0000000000000000000000000000000000000000..6e3cb39e5a69fca28ef2196cdb0407671e8805c0
--- /dev/null
+++ b/src/utils/BamTools/include/shared/bamtools_global.h
@@ -0,0 +1,79 @@
+// ***************************************************************************
+// bamtools_global.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// All rights reserved.
+// ---------------------------------------------------------------------------
+// Last modified: 3 March 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides the basic definitions for exporting & importing library symbols.
+// Also provides some platform-specific rules for definitions.
+// ***************************************************************************
+
+#ifndef BAMTOOLS_GLOBAL_H
+#define BAMTOOLS_GLOBAL_H
+
+/*! \brief Library export macro
+ \internal
+*/
+#ifndef BAMTOOLS_LIBRARY_EXPORT
+# if defined(WIN32)
+# define BAMTOOLS_LIBRARY_EXPORT __declspec(dllexport)
+# else
+# define BAMTOOLS_LIBRARY_EXPORT __attribute__((visibility("default")))
+# endif
+#endif // BAMTOOLS_LIBRARY_EXPORT
+
+/*! \brief Library import macro
+ \internal
+*/
+#ifndef BAMTOOLS_LIBRARY_IMPORT
+# if defined(WIN32)
+# define BAMTOOLS_LIBRARY_IMPORT __declspec(dllimport)
+# else
+# define BAMTOOLS_LIBRARY_IMPORT
+# endif
+#endif // BAMTOOLS_LIBRARY_IMPORT
+
+/*! \brief Platform-specific type definitions
+ \internal
+*/
+#ifndef BAMTOOLS_LFS
+#define BAMTOOLS_LFS
+ #ifdef WIN32
+ #define ftell64(a) _ftelli64(a)
+ #define fseek64(a,b,c) _fseeki64(a,b,c)
+ #else
+ #define ftell64(a) ftello(a)
+ #define fseek64(a,b,c) fseeko(a,b,c)
+ #endif
+#endif // BAMTOOLS_LFS
+
+/*! \def ftell64(a)
+ \brief Platform-independent tell() operation.
+ \internal
+*/
+/*! \def fseek64(a,b,c)
+ \brief Platform-independent seek() operation.
+ \internal
+*/
+
+/*! \brief Platform-specific type definitions
+ \internal
+*/
+#ifndef BAMTOOLS_TYPES
+#define BAMTOOLS_TYPES
+ #ifdef _MSC_VER
+ typedef char int8_t;
+ typedef unsigned char uint8_t;
+ typedef short int16_t;
+ typedef unsigned short uint16_t;
+ typedef int int32_t;
+ typedef unsigned int uint32_t;
+ typedef long long int64_t;
+ typedef unsigned long long uint64_t;
+ #else
+ #include <stdint.h>
+ #endif
+#endif // BAMTOOLS_TYPES
+
+#endif // BAMTOOLS_GLOBAL_H
diff --git a/src/utils/BamTools/lib/libbamtools.a b/src/utils/BamTools/lib/libbamtools.a
new file mode 100644
index 0000000000000000000000000000000000000000..ecf2d8ef5c48c40969a454a3b7ea5f34d422c7ba
Binary files /dev/null and b/src/utils/BamTools/lib/libbamtools.a differ
diff --git a/src/utils/BamTools/src/api/BamAlignment.o b/src/utils/BamTools/src/api/BamAlignment.o
new file mode 100644
index 0000000000000000000000000000000000000000..36be1eced244916a6535ffda1c4eecb49d5bfcd9
Binary files /dev/null and b/src/utils/BamTools/src/api/BamAlignment.o differ
diff --git a/src/utils/BamTools/src/api/BamMultiReader.o b/src/utils/BamTools/src/api/BamMultiReader.o
new file mode 100644
index 0000000000000000000000000000000000000000..f053ec2ab82d48b489244296b413cd1e6091c4f4
Binary files /dev/null and b/src/utils/BamTools/src/api/BamMultiReader.o differ
diff --git a/src/utils/BamTools/src/api/BamReader.o b/src/utils/BamTools/src/api/BamReader.o
new file mode 100644
index 0000000000000000000000000000000000000000..1fd86c23022b6a281f4a1135690e3477e21e989e
Binary files /dev/null and b/src/utils/BamTools/src/api/BamReader.o differ
diff --git a/src/utils/BamTools/src/api/BamWriter.o b/src/utils/BamTools/src/api/BamWriter.o
new file mode 100644
index 0000000000000000000000000000000000000000..9f7e8a4a859cf95c05edeb768a381936c4c96eba
Binary files /dev/null and b/src/utils/BamTools/src/api/BamWriter.o differ
diff --git a/src/utils/BamTools/src/api/SamHeader.o b/src/utils/BamTools/src/api/SamHeader.o
new file mode 100644
index 0000000000000000000000000000000000000000..10e49204ad057a11f159aec86122566f75b9c651
Binary files /dev/null and b/src/utils/BamTools/src/api/SamHeader.o differ
diff --git a/src/utils/BamTools/src/api/SamProgram.o b/src/utils/BamTools/src/api/SamProgram.o
new file mode 100644
index 0000000000000000000000000000000000000000..b7d6c41bd895efdfc342560e2d97daca9cded631
Binary files /dev/null and b/src/utils/BamTools/src/api/SamProgram.o differ
diff --git a/src/utils/BamTools/src/api/SamProgramChain.o b/src/utils/BamTools/src/api/SamProgramChain.o
new file mode 100644
index 0000000000000000000000000000000000000000..ddda8067930d6db4f4bf7e095d82278b12ab1146
Binary files /dev/null and b/src/utils/BamTools/src/api/SamProgramChain.o differ
diff --git a/src/utils/BamTools/src/api/SamReadGroup.o b/src/utils/BamTools/src/api/SamReadGroup.o
new file mode 100644
index 0000000000000000000000000000000000000000..644aff69b9462791a36b3276be9da5f3eb5ad553
Binary files /dev/null and b/src/utils/BamTools/src/api/SamReadGroup.o differ
diff --git a/src/utils/BamTools/src/api/SamReadGroupDictionary.o b/src/utils/BamTools/src/api/SamReadGroupDictionary.o
new file mode 100644
index 0000000000000000000000000000000000000000..d30ca9b094b895cb8bbf27e8565b3ab58d1d2e48
Binary files /dev/null and b/src/utils/BamTools/src/api/SamReadGroupDictionary.o differ
diff --git a/src/utils/BamTools/src/api/SamSequence.o b/src/utils/BamTools/src/api/SamSequence.o
new file mode 100644
index 0000000000000000000000000000000000000000..2ecce255d2c6d056bd1a8ff79ecb5b29b3f64106
Binary files /dev/null and b/src/utils/BamTools/src/api/SamSequence.o differ
diff --git a/src/utils/BamTools/src/api/SamSequenceDictionary.o b/src/utils/BamTools/src/api/SamSequenceDictionary.o
new file mode 100644
index 0000000000000000000000000000000000000000..972c52becc0169c84f19efa0ba3b09057c1f1e72
Binary files /dev/null and b/src/utils/BamTools/src/api/SamSequenceDictionary.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamHeader_p.o b/src/utils/BamTools/src/api/internal/BamHeader_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..b2bbda4485bec554f64bc899b7c2e02e0fdf26ec
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamHeader_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamIndexFactory_p.o b/src/utils/BamTools/src/api/internal/BamIndexFactory_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..728bae8935b88c0924dce74d4e11bb7f2bcb02c2
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamIndexFactory_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamMultiReader_p.o b/src/utils/BamTools/src/api/internal/BamMultiReader_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..c9099ab005622687bc8552bd18c7e738c6b35b22
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamMultiReader_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamRandomAccessController_p.o b/src/utils/BamTools/src/api/internal/BamRandomAccessController_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..7e96470d377a791f130ee6b21b1bd424831a55dc
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamRandomAccessController_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamReader_p.o b/src/utils/BamTools/src/api/internal/BamReader_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..14568a96d3b7e4b02056d723ea996d1d35001ecf
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamReader_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamStandardIndex_p.o b/src/utils/BamTools/src/api/internal/BamStandardIndex_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..38afbbae0441bad94687b5146afcc17d303fa4da
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamStandardIndex_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamToolsIndex_p.o b/src/utils/BamTools/src/api/internal/BamToolsIndex_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..e81c1139a314fa4a0ff8997e8bd2769827cc250e
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamToolsIndex_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BamWriter_p.o b/src/utils/BamTools/src/api/internal/BamWriter_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..d1900e310615606bd94fe45f0d0e43483c6a4488
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BamWriter_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/BgzfStream_p.o b/src/utils/BamTools/src/api/internal/BgzfStream_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..5160c6dee86ecb2a0c6ac6be1b8e5a816e72ef2d
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/BgzfStream_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/SamFormatParser_p.o b/src/utils/BamTools/src/api/internal/SamFormatParser_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..30a31ef4d208423fe10bba1abcb65e373187773d
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/SamFormatParser_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/SamFormatPrinter_p.o b/src/utils/BamTools/src/api/internal/SamFormatPrinter_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..ce79b555b00a19347a15c0a36240de9c5fe009be
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/SamFormatPrinter_p.o differ
diff --git a/src/utils/BamTools/src/api/internal/SamHeaderValidator_p.o b/src/utils/BamTools/src/api/internal/SamHeaderValidator_p.o
new file mode 100644
index 0000000000000000000000000000000000000000..3eefb6dfc51969914cffd6dade2bd086c375931b
Binary files /dev/null and b/src/utils/BamTools/src/api/internal/SamHeaderValidator_p.o differ
diff --git a/src/windowBed/Makefile b/src/windowBed/Makefile
index 0b5303f3980c228c20294d984af3676877dc8e91..e32b09b3bbc1a3bf252a165bace9f53d21e6d434 100644
--- a/src/windowBed/Makefile
+++ b/src/windowBed/Makefile
@@ -24,14 +24,10 @@ BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
PROGRAM= windowBed
-all: $(PROGRAM)
+all: $(BUILT_OBJECTS)
.PHONY: all
-$(PROGRAM): $(BUILT_OBJECTS) $(EXT_OBJECTS)
- @echo " * linking $(PROGRAM)"
- @$(CXX) $(LDFLAGS) $(CXXFLAGS) -o $(BIN_DIR)/$@ $^ -L$(UTILITIES_DIR)/BamTools/lib/ -lbamtools $(LIBS)
-
$(BUILT_OBJECTS): $(SOURCES)
@echo " * compiling" $(*F).cpp
@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
diff --git a/src/windowBed/windowMain.cpp b/src/windowBed/windowMain.cpp
index cbb5739a14120596893da0d89c26e34a78809249..9bf830ee129540f4343e05620fbe6b7f172eb8b2 100644
--- a/src/windowBed/windowMain.cpp
+++ b/src/windowBed/windowMain.cpp
@@ -15,16 +15,16 @@
using namespace std;
// define the version
-#define PROGRAM_NAME "windowBed"
+#define PROGRAM_NAME "bedtools window"
// define our parameter checking macro
#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
// function declarations
-void ShowHelp(void);
+void window_help(void);
-int main(int argc, char* argv[]) {
+int window_main(int argc, char* argv[]) {
// our configuration variables
bool showHelp = false;
@@ -65,7 +65,7 @@ int main(int argc, char* argv[]) {
}
}
- if(showHelp) ShowHelp();
+ if(showHelp) window_help();
// do some parsing (all of these parameters require 2 strings)
for(int i = 1; i < argc; i++) {
@@ -198,17 +198,16 @@ int main(int argc, char* argv[]) {
return 0;
}
else {
- ShowHelp();
+ window_help();
}
+ return 0;
}
-void ShowHelp(void) {
-
- cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
-
- cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
+void window_help(void) {
+ cerr << "\nTool: bedtools window (aka windowBed)" << endl;
+
cerr << "Summary: Examines a \"window\" around each feature in A and" << endl;
cerr << "\t reports all features in B that overlap the window. For each" << endl;
cerr << "\t overlap the entire entry in A and B are reported." << endl << endl;
@@ -219,7 +218,7 @@ void ShowHelp(void) {
cerr << "\t-abam\t" << "The A input file is in BAM format. Output will be BAM as well." << endl << endl;
- cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
+ cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default writes compressed BAM." << endl << endl;
cerr << "\t-bed\t" << "When using BAM input (-abam), write output as BED. The default" << endl;
cerr << "\t\tis to write output in BAM when using -abam." << endl << endl;