Added makewindows sub-command

Makefile

@@ -44,7 +44,8 @@ SUBDIRS = $(SRC_DIR)/annotateBed \
 		  $(SRC_DIR)/subtractBed \
 		  $(SRC_DIR)/tagBam \
 		  $(SRC_DIR)/unionBedGraphs \
-		  $(SRC_DIR)/windowBed
+		  $(SRC_DIR)/windowBed \
+		  $(SRC_DIR)/windowMaker
 
 UTIL_SUBDIRS =	$(SRC_DIR)/utils/lineFileUtilities \
 				$(SRC_DIR)/utils/bedFile \

scripts/makeBashScripts.py

@@ -37,7 +37,8 @@ def main():
                  'subtract': 'subtractBed', 
                  'tag': 'tagBam', 
                  'unionbedg': 'unionBedGraphs', 
-                 'window': 'windowBed'}
+                 'window': 'windowBed',
+                 'makewindows': 'windowMaker'}
 
     # create a BASH script for each old tool, mapping to the new CLI command.
     for tool in tool_map:

src/bedtools.cpp

@@ -63,6 +63,7 @@ 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 windowmaker_main(int argc, char* argv[]); //
 int bedtools_help(void);
 int bedtools_faq(void);
 
@@ -116,6 +117,7 @@ int main(int argc, char *argv[])
     else if (sub_cmd == "overlap")     return getoverlap_main(argc-1, argv+1);
     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);
 
     // help
     else if (sub_cmd == "-h" || sub_cmd == "--help" ||
@@ -134,14 +136,16 @@ int main(int argc, char *argv[])
     // verison information
     else if (sub_cmd == "-contact" || sub_cmd == "--contact")
     {
-        cout << "For further help, please email the bedtools mailing list: " << endl;
-        cout << "bedtools-discuss@googlegroups.com" << endl << endl;
+        cout << endl;
+        cout << "- For further help, or to report a bug, please " << endl;
+        cout << "  email the bedtools mailing list: " << endl;
+        cout << "     bedtools-discuss@googlegroups.com" << endl << endl;
 
-        cout << "Stable releases of bedtools can be found at: " << endl;
-        cout << "http://bedtools.googlecode.com" << endl << endl;
+        cout << "- Stable releases of bedtools can be found at: " << endl;
+        cout << "     http://bedtools.googlecode.com" << endl << endl;
 
-        cout << "The development repository can be found at: " << endl;
-        cout << "https://github.com/arq5x/bedtools" << endl << endl;
+        cout << "- The development repository can be found at: " << endl;
+        cout << "     https://github.com/arq5x/bedtools" << endl << endl;
     }
     // unknown
     else {
@@ -208,11 +212,12 @@ int bedtools_help(void)
     cout  << "    overlap       "  << "Computes the amount of overlap from two intervals.\n"; 
     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  << endl;
     cout  << " -General help:\n";
     cout  << "    --help        "  << "Print this help menu.\n";
-    cout  << "    --faq         "  << "Frequently asked questions.\n";
+    //cout  << "    --faq         "  << "Frequently asked questions.\n";  TODO
     cout  << "    --version     "  << "What version of bedtools are you using?.\n";
     cout  << "    --contact     "  << "Feature requests, bugs, mailing lists, etc.\n";
 

src/windowMaker/Makefile

+UTILITIES_DIR = ../utils/
+OBJ_DIR = ../../obj/
+BIN_DIR = ../../bin/
+
+# -------------------
+# define our includes
+# -------------------
+INCLUDES = -I$(UTILITIES_DIR)/genomeFile/ \
+           -I$(UTILITIES_DIR)/BamTools/include
+
+# ----------------------------------
+# define our source and object files
+# ----------------------------------
+SOURCES= windowMakerMain.cpp windowMaker.cpp
+OBJECTS= $(SOURCES:.cpp=.o)
+_EXT_OBJECTS=genomeFile.o
+EXT_OBJECTS=$(patsubst %,$(OBJ_DIR)/%,$(_EXT_OBJECTS))
+BUILT_OBJECTS= $(patsubst %,$(OBJ_DIR)/%,$(OBJECTS))
+PROGRAM= windowMaker
+
+
+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)/genomeFile/
+	
+clean:
+	@echo "Cleaning up."
+	@rm -f $(OBJ_DIR)/* $(BIN_DIR)/*
+
+.PHONY: clean

src/windowMaker/windowMaker.cpp

+/*****************************************************************************
+  windowMaker.cpp
+
+  (c) 2009 - Aaron Quinlan
+  Hall Laboratory
+  Department of Biochemistry and Molecular Genetics
+  University of Virginia
+  aaronquinlan@gmail.com
+
+  Licenced under the GNU General Public License 2.0 license.
+******************************************************************************/
+#include "windowMaker.h"
+
+WindowMaker::WindowMaker(string &genomeFile, uint32_t size, uint32_t step) 
+    : _genomeFile(genomeFile)
+    , _size(size)
+    , _step(step)
+{
+    _genome = new GenomeFile(genomeFile);
+    MakeWindows();
+}
+
+WindowMaker::~WindowMaker(void) {}
+
+
+void WindowMaker::MakeWindows() {
+
+    // get a list of the chroms in the user's genome
+    vector<string> chromList =  _genome->getChromList();
+
+    // process each chrom in the genome
+    for (size_t c = 0; c < chromList.size(); ++c) {
+        string chrom = chromList[c];
+        uint32_t chrom_size = _genome->getChromSize(chrom);
+        
+    	for (uint32_t start = 0; start <= chrom_size; start += _step) {
+    		if ((start + _size) <= chrom_size) {
+    			cout << chrom << "\t" << start << "\t" << start + _size << endl;
+    		}
+    		else if (start < chrom_size) {
+    			cout << chrom << "\t" << start << "\t" << chrom_size << endl;
+    		}
+    	}
+    }
+}
+

src/windowMaker/windowMaker.h

+/*****************************************************************************
+  windowMaker.h
+
+  (c) 2009 - Aaron Quinlan
+  Hall Laboratory
+  Department of Biochemistry and Molecular Genetics
+  University of Virginia
+  aaronquinlan@gmail.com
+
+  Licenced under the GNU General Public License 2.0 license.
+******************************************************************************/
+#include "genomeFile.h"
+
+using namespace std;
+
+
+//************************************************
+// Class methods and elements
+//************************************************
+class WindowMaker {
+
+public:
+
+  // constructor
+  WindowMaker(string &genomeFile, uint32_t size, uint32_t step);
+
+  // destructor
+  ~WindowMaker(void);
+
+  void MakeWindows();
+
+private:
+    string _genomeFile;
+    GenomeFile *_genome;
+    uint32_t _size; 
+    uint32_t _step;
+};

src/windowMaker/windowMakerMain.cpp

+/*****************************************************************************
+  windowMakerMain.cpp
+
+  (c) 2009 - Aaron Quinlan
+  Hall Laboratory
+  Department of Biochemistry and Molecular Genetics
+  University of Virginia
+  aaronquinlan@gmail.com
+
+  Licenced under the GNU General Public License 2.0 license.
+******************************************************************************/
+#include "windowMaker.h"
+
+using namespace std;
+
+// define our program name
+#define PROGRAM_NAME "bedtools makewindows"
+
+
+// define our parameter checking macro
+#define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
+
+// function declarations
+void windowmaker_help(void);
+
+int windowmaker_main(int argc, char* argv[]) {
+
+    // our configuration variables
+    bool showHelp = false;
+
+    // input files
+    string genomeFile;
+
+    // parms
+    uint32_t size = 0;
+    uint32_t step = 0;
+    
+    bool haveGenome = false;
+    bool haveSize   = false;
+    
+    for(int i = 1; i < argc; i++) {
+        int parameterLength = (int)strlen(argv[i]);
+
+        if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
+        (PARAMETER_CHECK("--help", 5, parameterLength))) {
+            showHelp = true;
+        }
+    }
+
+    if(showHelp) windowmaker_help();
+
+    // do some parsing (all of these parameters require 2 strings)
+    for(int i = 1; i < argc; i++) {
+
+        int parameterLength = (int)strlen(argv[i]);
+
+        if(PARAMETER_CHECK("-g", 2, parameterLength)) {
+            if ((i+1) < argc) {
+                haveGenome = true;
+                genomeFile = argv[i + 1];
+                i++;
+            }
+        }
+        else if(PARAMETER_CHECK("-w", 2, parameterLength)) {
+            if ((i+1) < argc) {
+                haveSize = true;
+                size = atoi(argv[i + 1]);
+                step = size;
+                i++;
+            }
+        }
+        else if(PARAMETER_CHECK("-s", 2, parameterLength)) {
+            if ((i+1) < argc) {
+                step = atoi(argv[i + 1]);
+                i++;
+            }
+        }
+        else {
+          cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
+            showHelp = true;
+        }
+    }
+
+    // make sure we have both input files
+    if (!haveGenome || !haveSize) {
+      cerr << endl << "*****" << endl << "*****ERROR: Need -g (genome file) and -w (window size). " << endl << "*****" << endl;
+      showHelp = true;
+    }
+    if (!showHelp) {
+        WindowMaker *wm = new WindowMaker(genomeFile, size, step);
+        delete wm;
+    }
+    else {
+        windowmaker_help();
+    }
+    return 0;
+}
+
+void windowmaker_help(void) {
+
+    cerr << "\nTool:    bedtools makewindows" << endl;
+    
+    cerr << "Summary: Makes adjacent and/or sliding windows across a genome." << endl << endl;
+
+    cerr << "Usage:   " << PROGRAM_NAME << " [OPTIONS] -g <genome> -w <window_size>" << endl << endl;
+
+    cerr << "Options: " << endl;
+
+    cerr << "\t-s\t"            << "Step size: i.e., how many base pairs to step before" << endl;
+    cerr                        << "\t\tcreating a new window. Used to create \"sliding\" windows." << endl;
+    cerr                        << "\t\t- Defaults to -w (non-sliding windows)." << endl << endl;
+    
+    cerr << "Notes: " << endl;
+    cerr << "\t(1)  The genome file should tab delimited and structured as follows:" << endl;
+    cerr << "\t     <chromName><TAB><chromSize>" << endl << endl;
+    cerr << "\tFor example, Human (hg19):" << endl;
+    cerr << "\tchr1\t249250621" << endl;
+    cerr << "\tchr2\t243199373" << endl;
+    cerr << "\t..." << endl;
+    cerr << "\tchr18_gl000207_random\t4262" << endl << endl;
+
+    cerr << "Tips: " << endl;
+    cerr << "\tOne can use the UCSC Genome Browser's MySQL database to extract" << endl;
+    cerr << "\tchromosome sizes. For example, H. sapiens:" << endl << endl;
+    cerr << "\tmysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -e \\" << endl;
+    cerr << "\t\"select chrom, size from hg19.chromInfo\"  > hg19.genome" << endl << endl;
+
+    exit(1);
+
+}