[ENH] add -allowBeyondChromEnd feature to shuffle.

src/shuffleBed/Makefile

@@ -7,7 +7,7 @@ BIN_DIR = ../../bin/
 # -------------------
 INCLUDES = -I$(UTILITIES_DIR)/bedFile/ \
            -I$(UTILITIES_DIR)/bedFilePE/ \
-           -I$(UTILITIES_DIR)/GenomeFile/ \
+           -I$(UTILITIES_DIR)/genomeFile/ \
            -I$(UTILITIES_DIR)/lineFileUtilities/ \
            -I$(UTILITIES_DIR)/gzstream/ \
            -I$(UTILITIES_DIR)/fileType/ \

src/shuffleBed/shuffleBed.cpp

@@ -19,7 +19,8 @@ BedShuffle::BedShuffle(string &bedFile, string &genomeFile,
                        bool haveInclude, bool sameChrom, 
                        float overlapFraction, int seed,
                        bool chooseChrom, bool isBedpe, size_t maxTries,
-                       bool noOverlapping) {
+                       bool noOverlapping, bool preventExceedingChromEnd) 
+{
 
     _bedFile         = bedFile;
     _genomeFile      = genomeFile;
@@ -34,6 +35,8 @@ BedShuffle::BedShuffle(string &bedFile, string &genomeFile,
     _isBedpe         = isBedpe;
     _maxTries        = maxTries;
     _noOverlapping   = noOverlapping;
+    _preventExceedingChromEnd = preventExceedingChromEnd;
+
 
     // use the supplied seed for the random
     // number generation if given.  else,
@@ -337,6 +340,14 @@ void BedShuffle::ChooseLocus(BED &bedEntry) {
             bedEntry.start = location.second;
             bedEntry.end   = bedEntry.start + length;
             chromSize      = _genome->getChromSize(location.first);
+
+            if ((bedEntry.end > chromSize) && 
+                (_preventExceedingChromEnd == false))
+            {
+                bedEntry.end = chromSize;
+                break;
+            }
+
         } while (bedEntry.end > chromSize);
         // keep looking if we have exceeded the end of the chrom.
     }
@@ -361,6 +372,14 @@ void BedShuffle::ChooseLocus(BED &bedEntry) {
                 bedEntry.start = randomStart;
                 bedEntry.end   = randomStart + length;
             }
+
+            if ((bedEntry.end > chromSize) && 
+                (_preventExceedingChromEnd == false))
+            {
+                bedEntry.end = chromSize;
+                break;
+            }
+
         } while (bedEntry.end > chromSize);
     }
 }

src/shuffleBed/shuffleBed.h

@@ -11,7 +11,7 @@
 ******************************************************************************/
 #include "bedFile.h"
 #include "bedFilePE.h"
-#include "GenomeFile.h"
+#include "genomeFile.h"
 
 #include <vector>
 #include <iostream>
@@ -39,7 +39,8 @@ public:
                bool haveInclude, bool sameChrom, 
                float overlapFraction, int seed, 
                bool chooseChrom, bool isBedpe,
-               size_t _maxTries, bool noOverlapping);
+               size_t _maxTries, bool noOverlapping,
+               bool preventExceedingChromEnd);
 
     // destructor
     ~BedShuffle(void);
@@ -60,6 +61,7 @@ private:
     bool _isBedpe;
     size_t _maxTries;
     bool _noOverlapping;
+    bool _preventExceedingChromEnd;
 
 
     // The BED file from which to compute coverage.

src/shuffleBed/shuffleBedMain.cpp

@@ -47,6 +47,7 @@ int shuffle_main(int argc, char* argv[]) {
     bool isBedpe          = false;
     size_t maxTries       = 1000;
     bool noOverlapping    = false;
+    bool allowBeyondChromEnd = false;
 
 
     for(int i = 1; i < argc; i++) {
@@ -125,6 +126,9 @@ int shuffle_main(int argc, char* argv[]) {
         else if(PARAMETER_CHECK("-noOverlapping", 14, parameterLength)) {
             noOverlapping = true;
         }
+        else if(PARAMETER_CHECK("-allowBeyondChromEnd", 20, parameterLength)) {
+            allowBeyondChromEnd = true;
+        }
         else {
           cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
             showHelp = true;
@@ -143,7 +147,8 @@ int shuffle_main(int argc, char* argv[]) {
                                         haveInclude, sameChrom, 
                                         overlapFraction, seed, 
                                         chooseChrom, isBedpe,
-                                        maxTries, noOverlapping);
+                                        maxTries, noOverlapping,
+                                        !(allowBeyondChromEnd));
         delete bc;
         return 0;
     }
@@ -198,7 +203,14 @@ void shuffle_help(void) {
     cerr << "\t-maxTries\t"         << "\n\t\tMax. number of attempts to find a home for a shuffled interval" << endl;
     cerr                            << "\t\tin the presence of -incl or -excl." << endl;
     cerr                            << "\t\tDefault = 1000." << endl;
-    cerr << "\t-noOverlapping\t"         << "\n\t\tDon't allow shuffled intervals to overlap." << endl;
+    cerr << "\t-noOverlapping\t"    << "\n\t\tDon't allow shuffled intervals to overlap." << endl;
+
+    cerr << "\t-allowBeyondChromEnd\t"  << "\n\t\tAllow shuffled intervals to be relocated to a position" << endl;
+    cerr                                 << "\t\tin which the entire original interval cannot fit w/o exceeding" << endl;
+    cerr                                 << "\t\tthe end of the chromosome.  In this case, the end coordinate of the" << endl;
+    cerr                                 << "\t\tshuffled interval will be set to the chromosome's length." << endl;
+    cerr                                 << "\t\tBy default, an interval's original length must be fully-contained" << endl;
+    cerr                                 << "\t\twithin the chromosome." << endl;
 
     cerr << "Notes: " << endl;
     cerr << "\t(1)  The genome file should tab delimited and structured as follows:" << endl;