Fixed some merge errors

Makefile

@@ -179,7 +179,7 @@ $(UTIL_SUBDIRS) $(SUBDIRS): $(OBJ_DIR) $(BIN_DIR)
 clean:
 	@$(MAKE) --no-print-directory --directory=$(BT_ROOT) clean_api
 	@echo " * Cleaning up."
-	@rm -f $(OBJ_DIR)/* $(BIN_DIR)/*
+	@rm -f $(VERSION_FILE) $(OBJ_DIR)/* $(BIN_DIR)/*
 .PHONY: clean
 
 test: all

README.md

 bedtools - a swiss army knife for genome arithmetic         
 ===================================================
 
-**Current version**: 2.22.1
+**Current version**: 2.23.0
 
 Note
 -------

docs/content/tools/merge.rst

@@ -113,9 +113,17 @@ The ``-s`` option will only merge intervals that are overlapping/bookended
   chr1  501  1000  a4  4 +
 
   $ bedtools merge -i A.bed -s
-  chr1  100  250    +
-  chr1  501  1000   +
-  chr1  250  500    -
+  chr1  100  250
+  chr1  501  1000
+  chr1  250  500
+
+To also report the strand, you could use the ``-c`` and ``-o`` operators (see below for more details):
+
+.. code-block:: bash
+
+  $ bedtools merge -i A.bed -s -c 6 -o distinct
+  chr1  100 250 +
+  chr1  501 1000  +
 
 ==========================================================================
 ``-S`` Reporting merged intervals on a specific strand. 

src/bedtools.cpp

@@ -213,7 +213,7 @@ int bedtools_help(void)
     cout  << "    random        "  << "Generate random intervals in a genome.\n";
     cout  << "    shuffle       "  << "Randomly redistrubute intervals in a genome.\n";
     cout  << "    sample        "  << "Sample random records from file using reservoir sampling.\n";   
-    cout  << "    spacing       "  << "Report the spacing between intervals in a file.\n";     
+    cout  << "    spacing       "  << "Report the gap lengths between intervals in a file.\n";     
     cout  << "    annotate      "  << "Annotate coverage of features from multiple files.\n";
     
     cout  << endl;

src/closestFile/closestHelp.cpp

@@ -84,5 +84,3 @@ void closest_help(void) {
     exit(1);
 }
 
-
-

src/coverageFile/coverageHelp.cpp

@@ -16,8 +16,8 @@ void coverage_help(void) {
 
     cerr << "\nTool:    bedtools coverage (aka coverageBed)" << endl;
     cerr << "Version: " << VERSION << "\n";    
-    cerr << "Summary: Returns the depth and breadth of coverage of features from A" << endl;
-    cerr << "\t on the intervals in B." << endl << endl;
+    cerr << "Summary: Returns the depth and breadth of coverage of features from B" << endl;
+    cerr << "\t on the intervals in A." << endl << endl;
 
     cerr << "Usage:   " << "bedtools coverage" << " [OPTIONS] -a <bed/gff/vcf> -b <bed/gff/vcf>" << endl << endl;
 
@@ -40,11 +40,11 @@ void coverage_help(void) {
     allToolsCommonHelp();
 
     cerr << "Default Output:  " << endl;
-    cerr << "\t" << " After each entry in B, reports: " << endl;
-    cerr << "\t   1) The number of features in A that overlapped the B interval." << endl;
-    cerr << "\t   2) The number of bases in B that had non-zero coverage." << endl;
-    cerr << "\t   3) The length of the entry in B." << endl;
-    cerr << "\t   4) The fraction of bases in B that had non-zero coverage." << endl << endl;
+    cerr << "\t" << " After each entry in A, reports: " << endl;
+    cerr << "\t   1) The number of features in B that overlapped the A interval." << endl;
+    cerr << "\t   2) The number of bases in A that had non-zero coverage." << endl;
+    cerr << "\t   3) The length of the entry in A." << endl;
+    cerr << "\t   4) The fraction of bases in A that had non-zero coverage." << endl << endl;
 
     exit(1);
 }

src/fisher/fisher.h

@@ -15,10 +15,10 @@
 
 class Fisher : public Jaccard {
 public:
-	Fisher(ContextFisher *context);
-	bool init();
-	bool finalizeCalculations();
-	void giveFinalReport(RecordOutputMgr *outputMgr);
+    Fisher(ContextFisher *context);
+    bool init();
+    bool finalizeCalculations();
+    void giveFinalReport(RecordOutputMgr *outputMgr);
 
 protected:
     bool _haveExclude;
@@ -28,7 +28,7 @@ protected:
     vector<int> _qsizes;
     BedFile *_excludeFile;
     unsigned long getTotalIntersection(RecordKeyVector &hits);
-	virtual ContextFisher *upCast(ContextBase *context) { return static_cast<ContextFisher *>(context); }
+    virtual ContextFisher *upCast(ContextBase *context) { return static_cast<ContextFisher *>(context); }
 
 };
 

src/genomeCoverageBed/genomeCoverageBed.cpp

@@ -124,7 +124,7 @@ void BedGenomeCoverage::AddCoverage(int start, int end) {
     // make sure the coordinates fit within the chrom
     if (start < _currChromSize)
         _currChromCoverage[start].starts++;
-    if (end < _currChromSize)
+    if (end >= 0 && end < _currChromSize)
         _currChromCoverage[end].ends++;
     else
         _currChromCoverage[_currChromSize-1].ends++;

src/jaccard/jaccard.h

@@ -38,8 +38,4 @@ protected:
 
 };
 
-
-
-
-
 #endif /* JACCARD_H_ */

src/sortBed/sortBed.cpp

@@ -9,13 +9,14 @@
 
   Licenced under the GNU General Public License 2.0 license.
 ******************************************************************************/
+#include <set>
 #include "lineFileUtilities.h"
 #include "sortBed.h"
 
 //
 // Constructor
 //
-BedSort::BedSort(string &bedFile, bool printHeader) {
+BedSort::BedSort(string &bedFile, bool printHeader,string &faidxFile):_faidxFile(faidxFile) {
     _bedFile = bedFile;
     _bed = new BedFile(bedFile);
     
@@ -47,6 +48,72 @@ void BedSort::SortBed() {
     }
 }
 
+void BedSort::SortBedOnFaidx()
+		{
+		set<string> all_chromosomes;
+		if(_faidxFile.empty())
+				{
+				cerr << "[sortBed] File for fasta index undefined." << endl;
+				exit(EXIT_FAILURE);
+				}
+		/* read FAIDX file */
+		ifstream faidx(_faidxFile.c_str(),ios::in);
+		if(!faidx.is_open()) {
+				cerr << "Cannot open \""<< _faidxFile << "\""
+						<< strerror(errno)
+						<< endl;
+				exit(EXIT_FAILURE);
+				}
+		string line;
+		while(getline(faidx,line,'\n')) {
+				if(line.empty()) continue;
+				string::size_type tab= line.find('\t');
+				if(tab!=string::npos) {
+						line.erase(tab);
+						}
+				if(all_chromosomes.find(line) != all_chromosomes.end()) {
+						cerr  << "Chromosome \"" << line
+									<<"\" defined twice in "
+									<< _faidxFile
+									<< endl;
+						exit(EXIT_FAILURE);
+						}
+				_tid2chrom[_tid2chrom.size()]=line;
+				all_chromosomes.insert(line);
+				}
+		faidx.close();
+		/** end read FAIDX */
+		
+		//check BED chromosomes
+		for (masterBedMapNoBin::iterator m = _bed->bedMapNoBin.begin(); m != _bed->bedMapNoBin.end(); ++m) {
+					if( all_chromosomes.find(m->first) ==  all_chromosomes.end()) {
+						cerr  << "Chromosome \"" << m->first
+									<<"\" undefined in "
+									<< _faidxFile
+									<< endl;
+						exit(EXIT_FAILURE);
+						}
+				}	
+		
+		
+		//loop over each chromosome using the faidx index
+		for(size_t tid=0; tid <_tid2chrom.size(); ++tid)
+			{
+			string chrom = _tid2chrom[tid];
+			masterBedMapNoBin::iterator m = _bed->bedMapNoBin.find(chrom);
+			
+			if( m == _bed->bedMapNoBin.end() ) continue; //this chromosome is not present in BED
+			
+		    // bedList is already sorted by start position.
+		    vector<BED> bedList = m->second;
+
+		    for (unsigned int i = 0; i < bedList.size(); ++i) {
+		        _bed->reportBedNewLine(bedList[i]);
+			  }
+			}
+		
+		}
+
 
 void BedSort::SortBedBySizeAsc() {
 

src/sortBed/sortBed.h

@@ -14,6 +14,7 @@
 #include <algorithm>
 #include <iostream>
 #include <fstream>
+#include <map>
 
 using namespace std;
 
@@ -26,12 +27,13 @@ class BedSort {
 public:
 
     // constructor
-    BedSort(string &bedFile, bool printHeader);
+    BedSort(string &bedFile, bool printHeader,string &faidxFile);
 
     // destructor
     ~BedSort(void);
 
     void SortBed();             // the default.  sorts by chrom (asc.) then by start (asc.)
+    void SortBedOnFaidx();			// sort BED using a given Reference (e.g: faidx file )
     void SortBedBySizeAsc();
     void SortBedBySizeDesc();
     void SortBedByChromThenSizeAsc();
@@ -41,7 +43,8 @@ public:
 
 private:
     string _bedFile;
-
+    string _faidxFile;/* genome index, generated by samtools faidx 1st col is chromosome */
+		map<size_t,string> _tid2chrom; /** map chromosome name to sort order */
     // instance of a bed file class.
     BedFile *_bed;
 };

src/sortBed/sortMain.cpp

@@ -31,6 +31,7 @@ int sort_main(int argc, char* argv[]) {
 
     // input files
     string bedFile  = "stdin";
+    string faidxFile; 
     bool haveBed    = true;
     int sortChoices = 0;
 
@@ -40,6 +41,7 @@ int sort_main(int argc, char* argv[]) {
     bool sortByChromThenSizeDesc  = false;
     bool sortByChromThenScoreAsc  = false;
     bool sortByChromThenScoreDesc = false;
+    bool sortByFaidx              = false;
     bool printHeader        = false;
 
     for(int i = 1; i < argc; i++) {
@@ -88,6 +90,14 @@ int sort_main(int argc, char* argv[]) {
             sortByChromThenScoreDesc = true;
             sortChoices++;
         }
+        else if(PARAMETER_CHECK("-faidx", 6, parameterLength)) {
+             if ((i+1) < argc) {
+                faidxFile = argv[i + 1];
+                i++;
+           		  }
+           	sortByFaidx = true;
+            sortChoices++;
+        }
         else if(PARAMETER_CHECK("-header", 7, parameterLength)) {
             printHeader = true;
         }
@@ -109,7 +119,7 @@ int sort_main(int argc, char* argv[]) {
 
 
     if (!showHelp) {
-        BedSort *bm = new BedSort(bedFile, printHeader);
+        BedSort *bm = new BedSort(bedFile, printHeader,faidxFile);
 
         if (sortBySizeAsc) {
             bm->SortBedBySizeAsc();
@@ -129,6 +139,9 @@ int sort_main(int argc, char* argv[]) {
         else if (sortByChromThenScoreDesc) {
             bm->SortBedByChromThenScoreDesc();
         }
+        else if (sortByFaidx) {
+            bm->SortBedOnFaidx();
+        }       
         else {
             bm->SortBed();
         }
@@ -148,13 +161,14 @@ void sort_help(void) {
     cerr << "Usage:   " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf>" << endl << endl;
 
     cerr << "Options: " << endl;
-    cerr << "\t" << "-sizeA\t\t"    << "Sort by feature size in ascending order." << endl;
-    cerr << "\t" << "-sizeD\t\t"    << "Sort by feature size in descending order." << endl;
-    cerr << "\t" << "-chrThenSizeA\t"   << "Sort by chrom (asc), then feature size (asc)." << endl;
-    cerr << "\t" << "-chrThenSizeD\t"   << "Sort by chrom (asc), then feature size (desc)." << endl;
-    cerr << "\t" << "-chrThenScoreA\t"  << "Sort by chrom (asc), then score (asc)." << endl;
-    cerr << "\t" << "-chrThenScoreD\t"  << "Sort by chrom (asc), then score (desc)." << endl;
-
+    cerr << "\t" << "-sizeA\t\t\t"    << "Sort by feature size in ascending order." << endl;
+    cerr << "\t" << "-sizeD\t\t\t"    << "Sort by feature size in descending order." << endl;
+    cerr << "\t" << "-chrThenSizeA\t\t"   << "Sort by chrom (asc), then feature size (asc)." << endl;
+    cerr << "\t" << "-chrThenSizeD\t\t"   << "Sort by chrom (asc), then feature size (desc)." << endl;
+    cerr << "\t" << "-chrThenScoreA\t\t"  << "Sort by chrom (asc), then score (asc)." << endl;
+    cerr << "\t" << "-chrThenScoreD\t\t"  << "Sort by chrom (asc), then score (desc)." << endl;
+    cerr << "\t" << "-faidx (names.txt)\t"  << "Sort according to the chromosomes declared in \"names.txt\"" << endl;
+    
     cerr << "\t-header\t"       << "Print the header from the A file prior to results." << endl << endl;
 
     exit(1);

src/spacingFile/spacingFile.h

@@ -17,21 +17,21 @@ class SpacingFile : public ToolBase {
 public:
     SpacingFile(ContextSpacing *context);
     virtual ~SpacingFile();
-	virtual bool init();
-	virtual bool findNext(RecordKeyVector &hits);
-	virtual void processHits(RecordOutputMgr *outputMgr, RecordKeyVector &hits);
-	virtual void cleanupHits(RecordKeyVector &hits);
-	virtual bool finalizeCalculations() { return true; }
-	virtual void  giveFinalReport(RecordOutputMgr *outputMgr) {}
+    virtual bool init();
+    virtual bool findNext(RecordKeyVector &hits);
+    virtual void processHits(RecordOutputMgr *outputMgr, RecordKeyVector &hits);
+    virtual void cleanupHits(RecordKeyVector &hits);
+    virtual bool finalizeCalculations() { return true; }
+    virtual void  giveFinalReport(RecordOutputMgr *outputMgr) {}
 
 
 protected:
-	const Record *_prevRec;
-	const Record *_currRec;
-	QuickString _distance;
+    const Record *_prevRec;
+    const Record *_currRec;
+    QuickString _distance;
 
-	FileRecordMgr *_inputFile;
-	virtual ContextSpacing *upCast(ContextBase *context) { return static_cast<ContextSpacing *>(context); }
+    FileRecordMgr *_inputFile;
+    virtual ContextSpacing *upCast(ContextBase *context) { return static_cast<ContextSpacing *>(context); }
 
 
 

src/spacingFile/spacingHelp.cpp

@@ -11,14 +11,14 @@ void spacing_help(void) {
 
     cerr << "\nTool:    bedtools spacing" << endl;
     cerr << "Version: " << VERSION << "\n";
-    cerr << "Summary: Report (last col.) the length of the space between intervals." << endl << endl;
+    cerr << "Summary: Report (last col.) the gap lengths between intervals in a file." << endl << endl;
 
     cerr << "Usage:   " << "bedtools spacing" << " [OPTIONS] -i <bed/gff/vcf/bam>" << endl << endl;
     
     cerr << "Notes: " << endl;
     cerr << "\t(1)  Input must be sorted by chrom,start (sort -k1,1 -k2,2n for BED)." << endl;
     cerr << "\t(2)  The 1st element for each chrom will have NULL distance. (\".\")." << endl;
-    cerr << "\t(3)  The distance for both overlapping and bookended intervals is 0." << endl << endl;
+    cerr << "\t(3)  Distance for overlapping intervaks is -1 and bookended is 0." << endl << endl;
 
     cerr << "Example: " << endl;
     cerr << "\t$ cat test.bed " << endl;

src/split/splitBed.cpp

@@ -124,7 +124,7 @@ int BedSplit::main(int argc,char** argv)
         }
     if(num_chuncks<=0)
         {
-        cerr << "Error: num_chuncks==0.\n" << endl;
+        cerr << "Error: num_chunks==0.\n" << endl;
         usage(cerr);
         return EXIT_FAILURE;
         }
@@ -138,7 +138,7 @@ int BedSplit::main(int argc,char** argv)
         }
     else 
         {
-        cerr << "Unknow split algorithm " << algorithm << endl;
+        cerr << "Unknown split algorithm " << algorithm << endl;
         return EXIT_FAILURE;
         }
     }
@@ -151,6 +151,7 @@ std::FILE* BedSplit::saveFileChunk(std::string& filename,size_t file_index)
     sprintf(tmp,"%05d",(file_index+1));
     filename.append(".").append(tmp).append(".bed");
     FILE* out = fopen(filename.c_str(),"w");
+    
     if(out==NULL)
         {
         fprintf(stderr,"Cannot open \"%s\". %s\n",

src/split/splitBed.h

@@ -24,6 +24,7 @@
 // Class methods and elements
 //************************************************
 class BedSplit {
+
 private:
     std::string bedFileName;
     std::string outfileprefix;
@@ -37,10 +38,8 @@ private:
     void loadBed();
     int doSimpleSplit();
     int doEuristicSplitOnTotalSize();
-public:
-
-
 
+public:
     // constructor
     BedSplit();
     // destructor

src/utils/FileRecordTools/Makefile

@@ -28,7 +28,10 @@ _EXT_OBJECTS=SingleLineDelimTextFileReader.o BamFileReader.o Bed3Interval.o Bed6
 EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
 BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
 
-$(BUILT_OBJECTS): $(SOURCES) $(SUBDIRS)
+all: $(BUILT_OBJECTS) $(SUBDIRS)
+.PHONY: all
+
+$(BUILT_OBJECTS): $(SOURCES)
 	@echo "  * compiling FileRecordMgr.cpp"
 	@$(CXX) -c -o $(OBJ_DIR)/FileRecordMgr.o FileRecordMgr.cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
 	@echo "  * compiling FileRecordMergeMgr.cpp"
@@ -48,4 +51,4 @@ clean:
 	@rm -f $(OBJ_DIR)/FileRecordMergeMgr.o
 	
 
-.PHONY: clean
\ No newline at end of file
+.PHONY: clean

src/utils/NewChromsweep/NewChromsweep.h

@@ -12,8 +12,6 @@
 #ifndef NEW_CHROMSWEEP_H
 #define NEW_CHROMSWEEP_H
 
-using namespace std;
-
 #include <string>
 #include "BTlist.h"
 #include "RecordKeyList.h"
@@ -24,6 +22,8 @@ using namespace std;
 #include <stdlib.h>
 #include "QuickString.h"
 
+using namespace std;
+
 class Record;
 class FileRecordMgr;
 class ContextIntersect;

src/utils/RecordOutputMgr/RecordOutputMgr.h

@@ -8,12 +8,12 @@
 #ifndef RECORDOUTPUTMGR_H_
 #define RECORDOUTPUTMGR_H_
 
-using namespace std;
-
 #include "ContextBase.h"
 #include "RecordKeyVector.h"
 #include "api/BamWriter.h"
 
+using namespace std;
+
 class BlockMgr;
 
 class RecordOutputMgr {

src/utils/ToolBase/ToolBase.h

@@ -8,11 +8,11 @@
 #ifndef TOOLBASE_H_
 #define TOOLBASE_H_
 
-using namespace std;
-
 #include "RecordKeyVector.h"
 #include "RecordOutputMgr.h"
 
+using namespace std;
+
 class ContextBase;
 
 class ToolBase {