Update GBSX digest script (input file location error fixed)

Updated GBSX demultiplexer for Single Read demultiplexer, issue with Enzymes with variable site. (This problem did not occure in Paired-End demultiplexing)

releases/latest/GBSX_digest_v1.0.pl → GBSX_digest_v1.1.pl

@@ -124,7 +124,9 @@
 my @ref_chr;
 open (ref_files,$fasta_file_locations)||die "could not open file $fasta_file_locations\n";
 while(<ref_files>){
-	chop;
+    #update 1.1: change chop to chomp
+	#chop;
+    chomp;
        	$_=~s/\n//;$_=~s/\r//;
 	if ($_ ne ""){ 
 		push(@ref_chr,$_); 

src/be/uzleuven/gc/logistics/GBSX/GBSX.java

@@ -35,7 +35,7 @@
 public class GBSX {
 
     public static final boolean DEBUG = false;
-    public final static String VERSION = "GBSX v1.1.4";
+    public final static String VERSION = "GBSX v1.1.5";
     private final static String LICENCE = "GPLv3";
 
     /**

src/be/uzleuven/gc/logistics/GBSX/demultiplexer/application/FastqDemultiplex.java

@@ -376,6 +376,7 @@ private void processPairEndFastqFiles(){
                     mostMismatches = sample.getBarcodeMismatches();
                 }
             }
+            //mostMismatches += this.parameters.getAllowedMismatchesEnzyme();
             statsFile.saveStats(mostMismatches, this.parameters.getOutputDirectory());
             this.correctionLog.writeToFile(this.parameters.getOutputDirectory());
         } catch (IOException ex) {
@@ -478,9 +479,10 @@ private void processSingleReadFastqFiles(){
             int mostMismatches = this.parameters.getAllowedMismatchesBarcode();
             for (Sample sample : this.sampleList){
                 if (sample.getBarcodeMismatches() > mostMismatches){
-                    mostMismatches = sample.getBarcodeMismatches();
+                    mostMismatches = sample.getBarcodeMismatches();                            
                 }
             }
+            //mostMismatches += this.parameters.getAllowedMismatchesEnzyme();
             statsFile.saveStats(mostMismatches, this.parameters.getOutputDirectory());
         } catch (IOException ex) {
             this.writeToLog("ERROR in reading the fastq files: " + ex.getMessage());
@@ -1114,22 +1116,30 @@ private SampleBarcodeCombination findGBSBarcode(String sequence, int startDistan
                 }
                 if (barcodeLocationLength[0] != -1){
                     //try every enzyme site
+                    boolean cutsiteFound = false;
+                    String exactEnzymeCutSite = "";
+                    int[] cutsiteLocationLength = new int[1];
+                    cutsiteLocationLength[0] = 0;
                     for (String enzymeCutSite : sample.getEnzyme().getInitialCutSiteRemnant()){
-                        int[] cutsiteLocationLength;
                         if ((cutsiteLocationLength = this.findingDistanceAlgorithm.calculateEquivalentDistance(sequence.substring(distance + barcodeLocationLength[1]), enzymeCutSite, this.parameters.getAllowedMismatchesEnzyme()))[0] != -1){
                             //check on adaptor ligase
                             if (this.parameters.getAdaptorLigaseMismatches() != -1){
                                 String adaptor = this.parameters.getCommonAdaptor();
                                 for (int l = cutsiteLocationLength[1]/2; l <= cutsiteLocationLength[1]; l++){
-                                    if (this.findingDistanceAlgorithm.calculateEquivalentDistance(sequence.substring(distance + barcodeLocationLength[1] + l), adaptor, this.parameters.getAdaptorLigaseMismatches())[0] != -1){
+                                    if (this.findingDistanceAlgorithm.calculateEquivalentDistance(sequence.substring(distance + barcodeLocationLength[1] + cutsiteLocationLength[1]), adaptor, this.parameters.getAdaptorLigaseMismatches())[0] != -1){
                                         return null;
+                                        //adaptor ligase
                                     }
                                 }
                             }
-                            foundSampleSet.add(new SampleBarcodeCombination(sample, enzymeCutSite, distance, barcodeLocationLength[0], barcodeLocationLength[1], cutsiteLocationLength[1]));
-//                            return new SampleBarcodeCombination(sample, enzymeCutSite, distance, barcodeLocationLength[0], barcodeLocationLength[1], cutsiteLocationLength[1]);
+                            exactEnzymeCutSite = enzymeCutSite;
+                            cutsiteFound = true;
                         }
                     }
+                    if (cutsiteFound){
+                        foundSampleSet.add(new SampleBarcodeCombination(sample, exactEnzymeCutSite, distance, barcodeLocationLength[0], barcodeLocationLength[1], cutsiteLocationLength[1]));
+
+                    }
                 }
             }
 

src/be/uzleuven/gc/logistics/GBSX/demultiplexer/model/SampleBarcodeCombination.java

@@ -88,4 +88,41 @@ public int getLengthFoundBarcode(){
         return this.lengthFoundBarcode;
     }
 
+    /**
+     * 
+     * @param s SampleBarcodeCombination
+     * @return true if the same sample is used
+     */
+   public boolean isSameSample(SampleBarcodeCombination s){
+       if (s.getSample().equals(this.getSample())){
+           return true;
+       }else{
+           return false;
+       }
+   }
+
+   /**
+    * 
+    * @param s SampleBarcodeCombination
+    * @return true is this sampleBarcodeCombination is better then s:
+    *   the location of this must be lower
+    *   when equal location this mismatches must be lower or equal
+    */
+   public boolean isThisBetter(SampleBarcodeCombination s){
+       if(isSameSample(s)){
+           if(s.getLocation() < this.getLocation()){
+               return false;
+           }else if(s.getLocation() > this.getLocation()){
+               return true;
+           }else{
+               if(s.getMismatches() >= this.getMismatches()){
+                   return true;
+               }else{
+                   return false;
+               }
+           }
+       }else{
+           return false;
+       }
+   }
 }