Added groupby back into the bedtools package.

Makefile

@@ -29,6 +29,8 @@ SUBDIRS = $(SRC_DIR)/annotateBed \
 		  $(SRC_DIR)/fastaFromBed \
 		  $(SRC_DIR)/flankBed \
 		  $(SRC_DIR)/genomeCoverageBed \
+		  $(SRC_DIR)/getOverlap \
+		  $(SRC_DIR)/groupBy \
 		  $(SRC_DIR)/intersectBed \
 		  $(SRC_DIR)/linksBed \
 		  $(SRC_DIR)/maskFastaFromBed \
@@ -36,7 +38,6 @@ SUBDIRS = $(SRC_DIR)/annotateBed \
 		  $(SRC_DIR)/multiBamCov \
 		  $(SRC_DIR)/multiIntersectBed \
 		  $(SRC_DIR)/nucBed \
-		  $(SRC_DIR)/getOverlap \
 		  $(SRC_DIR)/pairToBed \
 		  $(SRC_DIR)/pairToPair \
 		  $(SRC_DIR)/shuffleBed \

scripts/makeBashScripts.py

@@ -15,12 +15,13 @@ def main():
                  'bedpetobam': 'bedpeToBam', 
                  'bedtobam': 'bedToBam', 
                  'closest': 'closestBed',
-                 'cluster': 'clusterBed',                 
+                 'cluster': 'clusterBed',
                  'complement': 'complementBed',
                  'coverage': 'coverageBed', 
                  'flank': 'flankBed', 
                  'genomecov': 'genomeCoverageBed', 
-                 'getfasta': 'fastaFromBed', 
+                 'getfasta': 'fastaFromBed',
+                 'groupby': 'groupBy',
                  'igv': 'bedToIgv', 
                  'intersect': 'intersectBed', 
                  'links': 'linksBed', 

src/bedtools.cpp

@@ -48,6 +48,7 @@ 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 groupby_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[]);//
@@ -120,7 +121,7 @@ int main(int argc, char *argv[])
     else if (sub_cmd == "igv")         return bedtoigv_main(argc-1, argv+1);
     else if (sub_cmd == "links")       return links_main(argc-1, argv+1);
     else if (sub_cmd == "makewindows") return windowmaker_main(argc-1, argv+1);
-
+    else if (sub_cmd == "groupby")     return groupby_main(argc-1, argv+1);
     // help
     else if (sub_cmd == "-h" || sub_cmd == "--help" ||
              sub_cmd == "-help")
@@ -216,6 +217,7 @@ int bedtools_help(void)
     cout  << "    igv           "  << "Create an IGV snapshot batch script.\n";
     cout  << "    links         "  << "Create a HTML page of links to UCSC locations.\n";
     cout  << "    makewindows   "  << "Make interval \"windows\" across a genome.\n";
+    cout  << "    groupby       "  << "Group by common cols. & summarize oth. cols. (~ SQL \"groupBy\")\n";
 
     cout  << endl;
     cout  << " -General help:\n";

src/groupBy/Makefile

+UTILITIES_DIR = ../utils/
+OBJ_DIR = ../../obj/
+BIN_DIR = ../../bin/
+
+
+# -------------------
+# define our includes
+# -------------------
+INCLUDES = -I$(UTILITIES_DIR)/tabFile/ \
+           -I$(UTILITIES_DIR)/lineFileUtilities/ \
+           -I$(UTILITIES_DIR)/gzstream/ \
+           -I$(UTILITIES_DIR)/fileType/ \
+           -I$(UTILITIES_DIR)/version/
+
+# ----------------------------------
+# define our source and object files
+# ----------------------------------
+SOURCES= groupBy.cpp
+OBJECTS= $(SOURCES:.cpp=.o)
+_EXT_OBJECTS=tabFile.o gzstream.o fileType.o
+EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
+BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
+
+all: $(BUILT_OBJECTS)
+
+.PHONY: all
+
+
+$(BUILT_OBJECTS): $(SOURCES)
+	@echo "  * compiling" $(*F).cpp
+	@$(CXX) -c -o $@ $(*F).cpp $(LDFLAGS) $(CXXFLAGS) $(INCLUDES)
+
+$(EXT_OBJECTS):
+	@$(MAKE) --no-print-directory -C $(UTILITIES_DIR)/tabFile/
+	@$(MAKE) --no-print-directory -C $(UTILITIES_DIR)/gzstream/
+	@$(MAKE) --no-print-directory -C $(UTILITIES_DIR)/lineFileUtilities/
+	@$(MAKE) --no-print-directory -C $(UTILITIES_DIR)/fileType/	
+	
+clean:
+	@echo "Cleaning up."
+	@rm -f $(OBJ_DIR)/* $(BIN_DIR)/*
+
+.PHONY: clean