GBSX v1.2

Code Clean-up
Deleted demultiplex parameter -m

Author: koen

README.md

@@ -190,6 +190,7 @@ Possible Standard Enzymes for the info file: (NA is no enzyme)
 * `RBSTA`  
 * `RBSCG`  
 * `NspI`  
+* `AvaII`  
 * `NAN`  
 
 
@@ -256,4 +257,12 @@ v1.1.3
 v1.1.4
 * Update adaptor ligase finding algorithm
 * Removed unneeded, confusing parameters -cc and -al
-* Removed unused code
+* Removed unused code
+
+v1.1.5
+ * Update digest (removed possible input file parsing error)
+ * Updated Single Read Demultiplexing
+
+v1.2
+ * Deleted Demultiplexer option -m
+ * Code Clean-up

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 v1.1.5";
+    public final static String VERSION = "GBSX v1.2";
     private final static String LICENCE = "GPLv3";
 
     /**

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

@@ -57,7 +57,6 @@ public class BarcodeDiscovery {
     public static void main (String[] args){
         try{
             if (BarcodeDiscovery.DEBUG){
-//                String input = "-help";
                 String input ="-f1 /home/koen/Downloads/GBS_Example/dem/dem2/output.fastq";
                 args = input.split(" ");
             }

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

@@ -36,7 +36,7 @@
 public class GBSdemultiplex {
 
     public static final boolean DEBUG = false;
-    public final static String VERSION = "GBSX v1.3";
+    public final static String VERSION = "GBSX demultiplex v2";
     public final static String LICENCE = "GPLv3";
 
     /*

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

@@ -376,7 +376,6 @@ 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) {
@@ -482,7 +481,6 @@ private void processSingleReadFastqFiles(){
                     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());
@@ -663,7 +661,6 @@ private ProcessedFragment parseFastqRead(FastqRead read1, FastqRead read2) throw
                             read2optimalSequence = read2optimalSequence.substring(0, read1optimalSequence.length());
                             read2optimalQuality = read2optimalQuality.substring(0, read1optimalSequence.length());
                             trimedR2 = true;
-                            //sequenceError = "CORRECTED R2\n";
                             this.correctionLog.addCorrecterR2Corrected(sample);
                         }else{
                             if (read1optimalSequence.length() + sample.getBarcode().length() + this.parameters.getAdaptorCompareSize() >= read1.getSequence().length()){
@@ -675,7 +672,6 @@ private ProcessedFragment parseFastqRead(FastqRead read1, FastqRead read2) throw
                                 this.correctionLog.addCorrecterR1Corrected(sample);
                             }else{
                                 this.correctionLog.addCorrecterR2NotCorrected(sample);
-                                //sequenceError = "NOTCORRECT R2\n";
                             }
                         }
                     }else{
@@ -686,7 +682,6 @@ private ProcessedFragment parseFastqRead(FastqRead read1, FastqRead read2) throw
                             read2optimalQuality = read2optimalQuality.substring(0, read1optimalSequence.length());
                             trimedR2 = true;
                             this.correctionLog.addCorrecterR2Corrected(sample);
-                            //sequenceError = "CORRECTED R2\n";
                         }else{
                             if (read1optimalSequence.length() + sample.getBarcode().length() + this.parameters.getAdaptorCompareSize() + sample.getPossibleEnzymeCutSiteLength() + sample.getPossibleEnzymeCutSiteLength() >= read1.getSequence().length()){
                                 //read1 is only checked on cutsite, not on adaptor, not corrected so wrong
@@ -697,7 +692,6 @@ private ProcessedFragment parseFastqRead(FastqRead read1, FastqRead read2) throw
                                 this.correctionLog.addCorrecterR1Corrected(sample);
                             }else{
                                 this.correctionLog.addCorrecterR2NotCorrected(sample);
-                                //sequenceError = "NOTCORRECT R2\n";
                             }
                         }
                     }
@@ -714,10 +708,8 @@ private ProcessedFragment parseFastqRead(FastqRead read1, FastqRead read2) throw
                             read1optimalQuality = read1optimalQuality.substring(0, read2optimalSequence.length());
                             trimedR1 = true;
                             this.correctionLog.addCorrecterR1Corrected(sample);
-                            //sequenceError = "CORRECTED R1\n";
                         }else{
                             this.correctionLog.addCorrecterR1NotCorrected(sample);
-                            //sequenceError = "NOTCORRECT R1\n";
                         }
                     }else{
                         //not keep cutsites
@@ -726,10 +718,8 @@ private ProcessedFragment parseFastqRead(FastqRead read1, FastqRead read2) throw
                             read1optimalQuality = read1optimalQuality.substring(0, read2optimalSequence.length());
                             trimedR1 = true;
                             this.correctionLog.addCorrecterR1Corrected(sample);
-                            //sequenceError = "CORRECTED R1\n";
                         }else{
                             this.correctionLog.addCorrecterR1NotCorrected(sample);
-                            //sequenceError = "NOTCORRECT R1\n";
                         }
                     }
                 }
@@ -845,11 +835,8 @@ private ProcessedFragment parseFastqRead(FastqRead read1) throws InvalidReadExce
         //find the next enzyme site (if there is any)
         int read1EndLocation = -1;
         int[] read1EndLocationLength = {-1, 0};
-//        EnzymeComparator enzymeComparator = new EnzymeComparator();
-//        if (enzymeComparator.compare(sample.getEnzyme(), this.parameters.getNeutralEnzyme()) != 0){
-            read1EndLocationLength = this.findRead1EnzymeLocation(read1modifiedSequence, sample);
-            read1EndLocation = read1EndLocationLength[0];
-//        }
+        read1EndLocationLength = this.findRead1EnzymeLocation(read1modifiedSequence, sample);
+        read1EndLocation = read1EndLocationLength[0];
         String read1optimalSequence = read1modifiedSequence;
         String read1optimalQuality = read1modifiedQuality;
         if (read1EndLocation != -1){
@@ -989,7 +976,7 @@ private int[] findRead2EnzymeLocation(String sequence, Sample sample, String fou
             if (extraAdaptorSearch > this.parameters.getCommonAdaptor().length()){
                 extraAdaptorSearch = this.parameters.getCommonAdaptor().length();
             }
-//            int extraAdaptorSearch = 4;
+            
             int[] place = {-1, 0};
             boolean searchMore = true;
             int posloc = -1;
@@ -1001,9 +988,7 @@ private int[] findRead2EnzymeLocation(String sequence, Sample sample, String fou
                 if (posplace[0] == -1){
                     //no index found
                     searchMore = false;
-                }else if (posloc + 1 + posplace[0] < sequence.length() - complementFoundEnzyme.length() - sample.getComplementBarcode().length()
-//                        && posloc + 1 + posplace[0] + complementFoundEnzyme.length() + sample.getComplementBarcode().length() + extraAdaptorSearch < sequence.length()
-                        ){
+                }else if (posloc + 1 + posplace[0] < sequence.length() - complementFoundEnzyme.length() - sample.getComplementBarcode().length()){
                     //if new found possible location is smaller then the maximum location
                     //and the the new found location + enzyme + barcode + extraAdaptor search is smaller than the length of the sequence
                     extraAdaptorSearch = this.parameters.getAdaptorCompareSize();
@@ -1040,21 +1025,12 @@ private int[] findRead2EnzymeLocation(String sequence, Sample sample, String fou
             place[1] = complementFoundEnzyme.length();
             if (place[0] == -1){
                 //find with no barcode
-//                int[] newplace = this.findingDistanceAlgorithm.indexOf(sequence.substring(sequence.length() - complementFoundEnzyme.length() - sample.getBarcode().length() - extraAdaptorSearch), complementFoundEnzyme, this.parameters.getAllowedMismatchesEnzyme());
                 int[] newplace = this.findingDistanceAlgorithm.indexOf(sequence.substring(sequence.length() - complementFoundEnzyme.length() - sample.getBarcode().length()), complementFoundEnzyme, this.parameters.getAllowedMismatchesEnzyme());
 
                 if (newplace[0] != -1){
-//                    place[0] = (sequence.length() - complementFoundEnzyme.length() - sample.getBarcode().length() - extraAdaptorSearch + newplace[0]);
-//                    if(this.findingDistanceAlgorithm.isEquivalent(sequence.substring(sequence.length() - sample.getBarcode().length() + newplace[0]), sample.getComplementBarcode().substring(0, sample.getBarcode().length() + complementFoundEnzyme.length() - newplace[0]), this.parameters.getAllowedMismatchesBarcode(sample))){
-//                        place[0] = (sequence.length() - complementFoundEnzyme.length() - sample.getBarcode().length() + newplace[0]);
-//                    }else{
-//                        newplace[0] = -1;
-//                    }
                     place[0] = (sequence.length() - complementFoundEnzyme.length() - sample.getBarcode().length() + newplace[0]);
                 }
                 else{
-//                if (newplace[0] == -1){
-                    //find with part enzyme
                     for (int ei=0; ei < complementFoundEnzyme.length(); ei++){
                         newplace = this.findingDistanceAlgorithm.calculateEquivalentDistance(sequence.substring(sequence.length() - complementFoundEnzyme.length() + ei), complementFoundEnzyme.substring(0, complementFoundEnzyme.length() - ei), 0);
                         if (newplace[0] != -1){

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

@@ -74,11 +74,6 @@ public enum DemultiplexArguments implements Arguments{
      * Keep the enzyme cut sites
      */
     KEEP_ENZYME_CUTSITES ("-kc"),
-    /**
-     * The allowed mismatches in a barcode, enzyme or combination
-     * Optional
-     */
-    ALLOWED_MISMATCHES ("-m"),
     /**
      * The used algorithm to search for mismatches
      */

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

@@ -44,7 +44,8 @@ public DemultiplexParameters(){
         this.arguments = new HashMap();
         //standard parameters
         this.arguments.put(DemultiplexArguments.OUTPUT_DIRECTORY, System.getProperty("user.dir"));
-        this.arguments.put(DemultiplexArguments.ALLOWED_MISMATCHES, "1");
+        this.arguments.put(DemultiplexArguments.ALLOWED_MISMATCHES_BARCODE, "1");
+        this.arguments.put(DemultiplexArguments.ALLOWED_MISMATCHES_ENZYME, "1");
         this.arguments.put(DemultiplexArguments.START_BARCODE_DISTANCE, "0");
         this.arguments.put(DemultiplexArguments.COMPLETE_CHECK, "true");
         this.arguments.put(DemultiplexArguments.QUALITY_SCORE, FastqScores.ILLUMINA_1_8.getType());
@@ -163,10 +164,7 @@ public String getErrorRequiredParametersSet(){
      * @return int | the allowed mismatches for the barcode
      */
     public int getAllowedMismatchesBarcode(){
-        if (this.arguments.containsKey(DemultiplexArguments.ALLOWED_MISMATCHES_BARCODE)){
-            return Integer.parseInt(this.arguments.get(DemultiplexArguments.ALLOWED_MISMATCHES_BARCODE));
-        }
-        return Integer.parseInt(this.arguments.get(DemultiplexArguments.ALLOWED_MISMATCHES));
+        return Integer.parseInt(this.arguments.get(DemultiplexArguments.ALLOWED_MISMATCHES_BARCODE));
     }
 
     /**
@@ -240,10 +238,7 @@ public boolean useLongFileNames(){
      * @return int | the allowed mismatches for the enzyme
      */
     public int getAllowedMismatchesEnzyme(){
-        if (this.arguments.containsKey(DemultiplexArguments.ALLOWED_MISMATCHES_ENZYME)){
-            return Integer.parseInt(this.arguments.get(DemultiplexArguments.ALLOWED_MISMATCHES_ENZYME));
-        }
-        return Integer.parseInt(this.arguments.get(DemultiplexArguments.ALLOWED_MISMATCHES));
+        return Integer.parseInt(this.arguments.get(DemultiplexArguments.ALLOWED_MISMATCHES_ENZYME));
     }
 
     /**
@@ -556,7 +551,6 @@ public String getParametersHelp(){
         toHelp += "\t -lf \t use long file names (standard false) filename is standard the sample name, long file names is sample name _ barcode _ enzyme" + "\n";
         toHelp += "\t -rad \t if the data is rad data or not (-rad true for RAD data, -rad false for GBS data) standard false (GBS)" + "\n";
         toHelp += "\t -gzip \t the input and output are/must be gziped (.gz) (standard false: input and output are .fastq, if true this is .fastq.gz)" + "\n";
-        toHelp += "\t -m \t the allowed mismatches in the barcodes + enzymes (standard this value is 1)" + "\n";
         toHelp += "\t -mb \t the allowed mismatches in the barcodes (overrides the option -m)" + "\n";
         toHelp += "\t -me \t the allowed mismatches in the enzymes (overrides the option -m)" + "\n";
         toHelp += "\t -minsl \t the minimum allowed length for the sequences (standard 0, rejected sequences are found in the stats for each sample in the rejected.count column. The sequences self are found untrimmed in the undetermined file.)" + "\n";

src/be/uzleuven/gc/logistics/GBSX/utils/fasta/infrastructure/FastaBufferedWriter.java

@@ -31,7 +31,7 @@ public class FastaBufferedWriter {
     /**
      * creates a new FastaBufferedWriter of the given file.
      * @param file | the file that must be writen
-     * @param  zipped | true if the file must be zipped
+     * @param  ziped | true if the file must be zipped
      * @throws FileNotFoundException | if the given file isn't found
      * @throws IOException | if an error occures while opening the file
      */
@@ -46,7 +46,7 @@ public FastaBufferedWriter(File file, boolean ziped) throws FileNotFoundExceptio
     /**
      * creates a new FastaBufferedWriter of the given file.
      * @param file | the file that must be writen
-     * @param  zipped | true if the file must be zipped
+     * @param  ziped | true if the file must be zipped
      * @param append | true if append to existing file
      * @throws FileNotFoundException | if the given file isn't found
      * @throws IOException | if an error occures while opening the file
@@ -62,7 +62,7 @@ public FastaBufferedWriter(File file, boolean ziped, boolean append) throws File
     /**
      * creates a new FastaBufferedReader of the given URL.
      * @param url | the URL of the file that must be writen
-     * @param  zipped | true if the file must be zipped
+     * @param  ziped | true if the file must be zipped
      * @throws FileNotFoundException | if the given file isn't found
      * @throws IOException | if an error occures while opening the file
      */
@@ -76,7 +76,7 @@ public FastaBufferedWriter(URL url, boolean ziped) throws FileNotFoundException,
     /**
      * creates a new FastaBufferedReader of the given URL.
      * @param url | the URL of the file that must be writen
-     * @param  zipped | true if the file must be zipped
+     * @param  ziped | true if the file must be zipped
      * @param append | true if append to existing file
      * @throws FileNotFoundException | if the given file isn't found
      * @throws IOException | if an error occures while opening the file

src/be/uzleuven/gc/logistics/GBSX/utils/fasta/infrastructure/FastaPairBufferedReader.java

@@ -51,19 +51,6 @@ public FastaPairBufferedReader(File fileRead1, File fileRead2, boolean ziped) th
         this.fastaBufferedReader2 = new FastaBufferedReader(fileRead2, ziped);
     }
 
-    /**
-     * creates 2 new FastqBufferedReader of the given files.
-     * @param fileRead1 | the URL for read 1
-     * @param fileRead2 | the URL for read 2
-     * @param ziped | boolean if the fasta file is ziped (gz)
-     * @throws FileNotFoundException | if the given file isn't found
-     * @throws IOException | if an error occures while opening the file
-     */
-//    public FastqPairBufferedReader(URL fileRead1, URL fileRead2, boolean ziped) throws FileNotFoundException, IOException{
-//        this.fastaBufferedReader1 = new FastaBufferedReader(fileRead1, ziped);
-//        this.fastaBufferedReader2 = new FastaBufferedReader(fileRead2, ziped);
-//    }
-    
     /**
      * reads the next lines in the fasta file. and returns it as an fastaread
      * @return null if there are no fasta files anymore, else a FastaRead

src/be/uzleuven/gc/logistics/GBSX/utils/fasta/infrastructure/FastaPairBufferedWriter.java

@@ -49,19 +49,6 @@ public FastaPairBufferedWriter(File fileRead1, File fileRead2, boolean ziped) th
         this.fastaBufferedWriter2 = new FastaBufferedWriter(fileRead2, ziped);
     }
 
-    /**
-     * creates 2 new FastaBufferedReader of the given files.
-     * @param fileRead1 | the URL for read 1
-     * @param fileRead2 | the URL for read 2
-     * @param ziped | boolean if the fasta file is ziped (gz)
-     * @throws FileNotFoundException | if the given file isn't found
-     * @throws IOException | if an error occures while opening the file
-     */
-//    public FastaPairBufferedReader(URL fileRead1, URL fileRead2, boolean ziped) throws FileNotFoundException, IOException{
-//        this.fastaBufferedReader1 = new FastaBufferedReader(fileRead1, ziped);
-//        this.fastaBufferedReader2 = new FastaBufferedReader(fileRead2, ziped);
-//    }
-    
     /**
      * reads the next lines in the fasta file. and returns it as an fastaread
      * @return null if there are no fasta files anymore, else a FastqRead