vidjil.cpp, windowExtractor.{h,cpp}: new/updated -u/-uu/-uuu unsegmentation options

Especially with RNAseq data, one may want to output reads with V/J similarities,
but not random reads for other parts of the genome.
The new behavior of the '-u' option is to return these interesting reads.

algo/core/windowExtractor.cpp

@@ -91,6 +91,7 @@ WindowsStorage *WindowExtractor::extract(OnlineFasta *reads,
         *out_unsegmented << *seg ;
       }
       if (out_unsegmented_detail && (seg->getSegmentationStatus() >= STATS_FIRST_UNSEG)) {
+        if (unsegmented_detail_full || (seg->getSegmentationStatus() != UNSEG_TOO_FEW_ZERO && seg->getSegmentationStatus() != UNSEG_TOO_SHORT))
         *out_unsegmented_detail[seg->getSegmentationStatus()] << *seg ;
       }
     }
@@ -153,8 +154,9 @@ void WindowExtractor::setUnsegmentedOutput(ostream *out) {
   out_unsegmented = out;
 }
 
-void WindowExtractor::setUnsegmentedDetailOutput(ofstream **outs) {
+void WindowExtractor::setUnsegmentedDetailOutput(ofstream **outs, bool unsegmented_detail_full) {
   out_unsegmented_detail = outs;
+  this->unsegmented_detail_full = unsegmented_detail_full;
 }
 
 void WindowExtractor::setAffectsOutput(ostream *out) {

algo/core/windowExtractor.h

@@ -29,6 +29,7 @@ class WindowExtractor {
   ostream *out_segmented;
   ostream *out_unsegmented;
   ofstream **out_unsegmented_detail;
+  bool unsegmented_detail_full;
   ostream *out_affects;
 
   Stats stats[STATS_SIZE];
@@ -113,9 +114,10 @@ class WindowExtractor {
   /**
    * Defines the output streams where the unsegmented sequences will be output, split by unsegmentation cause.
    * Otherwise no output will be given.
-   * @param out: The output streams
+   * @param outs: The output streams
+   * @param unsegmented_detail_full: Whether we should output UNSEG_TOO_FEW_ZERO reads
    */
-  void setUnsegmentedDetailOutput(ofstream **outs);
+  void setUnsegmentedDetailOutput(ofstream **outs, bool unsegmented_detail_full=false);
 
   /**
    * Defines the output stream where the detailed affects will be output.

algo/vidjil.cpp

@@ -240,7 +240,8 @@ void usage(char *progname, bool advanced)
   cerr << "Detailed output per read (generally not recommended, large files, but may be used for filtering, as in -uu -X 1000)" << endl
        << "  -U            output segmented reads (in " << SEGMENTED_FILENAME << " file)" << endl
        << "  -u            output unsegmented reads (in " << UNSEGMENTED_FILENAME << " file)" << endl
-       << "  -uu           output unsegmented reads, gathered by unsegmentation cause (in *" << UNSEGMENTED_DETAIL_FILENAME << " files)" << endl
+       << "  -uu           output unsegmented reads, gathered by unsegmentation cause, except for very short and 'too few V/J' reads (in *" << UNSEGMENTED_DETAIL_FILENAME << " files)" << endl
+       << "  -uuu          output unsegmented reads, gathered by unsegmentation cause, all reads (in *" << UNSEGMENTED_DETAIL_FILENAME << " files) (use only for debug)" << endl
        << "  -K            output detailed k-mer affectation on all reads (in " << AFFECTS_FILENAME << " file) (use only for debug, for example -KX 100)" << endl
        << endl
  
@@ -361,6 +362,7 @@ int main (int argc, char **argv)
   bool output_segmented = false;
   bool output_unsegmented = false;
   bool output_unsegmented_detail = false;
+  bool output_unsegmented_detail_full = false;
   bool output_affects = false;
   bool keep_unsegmented_as_clone = false;
 
@@ -648,8 +650,9 @@ int main (int argc, char **argv)
         break;
 
       case 'u':
-        output_unsegmented_detail = output_unsegmented; // -uu
-        output_unsegmented = true;
+        output_unsegmented = output_unsegmented_detail_full ;       // -uuu
+        output_unsegmented_detail_full = output_unsegmented_detail; // -uu
+        output_unsegmented_detail = true;                           // -u
         break;
       case 'U':
         output_segmented = true;
@@ -1067,7 +1070,7 @@ int main (int argc, char **argv)
           out_unsegmented_detail[i] = new ofstream(f_unsegmented_detail.c_str());
         }
 
-      we.setUnsegmentedDetailOutput(out_unsegmented_detail);
+      we.setUnsegmentedDetailOutput(out_unsegmented_detail, output_unsegmented_detail_full);
     }