Changes passing anchor reference from using index to text

jibsch · Oct 21, 2013 · d601b9c · d601b9c
1 parent fdead12
commit d601b9c
Show file tree

Hide file tree

Showing 38 changed files with 103 additions and 58 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,6 +1,6 @@
 # Test data
 MS0465*
-results_*
+130423*
 results_Socrates_*
 sample_*
 do_all.sh

diff --git a/Socrates b/Socrates
diff --git a/bin/net/wehi/socrates/AnnotatePairedClusters.class b/bin/net/wehi/socrates/AnnotatePairedClusters.class
diff --git a/bin/net/wehi/socrates/BAMStratifier$AlignmentInfo.class b/bin/net/wehi/socrates/BAMStratifier$AlignmentInfo.class
diff --git a/bin/net/wehi/socrates/BAMStratifier$MergeWorker.class b/bin/net/wehi/socrates/BAMStratifier$MergeWorker.class
diff --git a/bin/net/wehi/socrates/BAMStratifier$OutputFiles.class b/bin/net/wehi/socrates/BAMStratifier$OutputFiles.class
diff --git a/bin/net/wehi/socrates/BAMStratifier$StratifyWorker.class b/bin/net/wehi/socrates/BAMStratifier$StratifyWorker.class
diff --git a/bin/net/wehi/socrates/BAMStratifier.class b/bin/net/wehi/socrates/BAMStratifier.class
diff --git a/bin/net/wehi/socrates/Cluster.class b/bin/net/wehi/socrates/Cluster.class
diff --git a/bin/net/wehi/socrates/PairedCluster.class b/bin/net/wehi/socrates/PairedCluster.class
diff --git a/bin/net/wehi/socrates/RealignmentBAM$Info.class b/bin/net/wehi/socrates/RealignmentBAM$Info.class
diff --git a/bin/net/wehi/socrates/RealignmentBAM.class b/bin/net/wehi/socrates/RealignmentBAM.class
diff --git a/bin/net/wehi/socrates/RealignmentClustering$ClusteringWorker.class b/bin/net/wehi/socrates/RealignmentClustering$ClusteringWorker.class
diff --git a/bin/net/wehi/socrates/RealignmentClustering$ResultCounter.class b/bin/net/wehi/socrates/RealignmentClustering$ResultCounter.class
diff --git a/bin/net/wehi/socrates/RealignmentClustering.class b/bin/net/wehi/socrates/RealignmentClustering.class
diff --git a/bin/net/wehi/socrates/SOCRATES.class b/bin/net/wehi/socrates/SOCRATES.class
diff --git a/bin/net/wehi/socrates/util/ConsensusMaker.class b/bin/net/wehi/socrates/util/ConsensusMaker.class
diff --git a/bin/net/wehi/socrates/util/FindSeqByName.class b/bin/net/wehi/socrates/util/FindSeqByName.class
diff --git a/bin/net/wehi/socrates/util/GenomicIndexInterval.class b/bin/net/wehi/socrates/util/GenomicIndexInterval.class
diff --git a/bin/net/wehi/socrates/util/GenomicIntervalList.class b/bin/net/wehi/socrates/util/GenomicIntervalList.class
diff --git a/bin/net/wehi/socrates/util/GenomicIntervalListMap.class b/bin/net/wehi/socrates/util/GenomicIntervalListMap.class
diff --git a/bin/net/wehi/socrates/util/MemoryMappedFile.class b/bin/net/wehi/socrates/util/MemoryMappedFile.class
diff --git a/bin/net/wehi/socrates/util/ProgressVerbose.class b/bin/net/wehi/socrates/util/ProgressVerbose.class
diff --git a/bin/net/wehi/socrates/util/RealignmentCluster.class b/bin/net/wehi/socrates/util/RealignmentCluster.class
diff --git a/bin/net/wehi/socrates/util/RealignmentRecordSummary.class b/bin/net/wehi/socrates/util/RealignmentRecordSummary.class
diff --git a/bin/net/wehi/socrates/util/SAMFileInfo.class b/bin/net/wehi/socrates/util/SAMFileInfo.class
diff --git a/bin/net/wehi/socrates/util/SAMRecordSummary.class b/bin/net/wehi/socrates/util/SAMRecordSummary.class
diff --git a/bin/net/wehi/socrates/util/SeekableMemoryStream.class b/bin/net/wehi/socrates/util/SeekableMemoryStream.class
diff --git a/bin/net/wehi/socrates/util/TabixReader$Iterator.class b/bin/net/wehi/socrates/util/TabixReader$Iterator.class
diff --git a/bin/net/wehi/socrates/util/TabixReader$TIndex.class b/bin/net/wehi/socrates/util/TabixReader$TIndex.class
diff --git a/bin/net/wehi/socrates/util/TabixReader$TIntv.class b/bin/net/wehi/socrates/util/TabixReader$TIntv.class
diff --git a/bin/net/wehi/socrates/util/TabixReader$TPair64.class b/bin/net/wehi/socrates/util/TabixReader$TPair64.class
diff --git a/bin/net/wehi/socrates/util/TabixReader.class b/bin/net/wehi/socrates/util/TabixReader.class
diff --git a/bin/net/wehi/socrates/util/Utilities.class b/bin/net/wehi/socrates/util/Utilities.class
diff --git a/bin/net/wehi/socrates/util/Values.class b/bin/net/wehi/socrates/util/Values.class
diff --git a/src/net/wehi/socrates/BAMStratifier.java b/src/net/wehi/socrates/BAMStratifier.java
@@ -471,7 +471,7 @@ private void flush(int keepInBuffer) {
 						boolean clip3p = (aln.getReadNegativeStrandFlag()) ? true : false;
 						boolean ideal = ((proper&&!clip3p) || (!proper&&clip3p));
 
-						String header = RealignmentRecordSummary.makeFastqHeader(aln, summary, aln.getReferenceIndex(), summary.getHeadClipPos(), ideal,
+						String header = RealignmentRecordSummary.makeFastqHeader(aln, summary, sourceBAMInfo.getSequenceName(aln.getReferenceIndex()), summary.getHeadClipPos(), ideal,
 								Utilities.sequenceByteToString(summary.getHeadClipAlignedSequence(), false), '-');
 						String clipSeq = Utilities.sequenceByteToString(head, false); // make breakpoint at 3' end
 						String clipQual = Utilities.qualityByteToString(summary.getHeadClipQuality(), false);
@@ -484,7 +484,7 @@ private void flush(int keepInBuffer) {
 						boolean clip3p = (aln.getReadNegativeStrandFlag()) ? false : true;
 						boolean ideal = ((proper&&!clip3p) || (!proper&&clip3p));
 
-						String header = RealignmentRecordSummary.makeFastqHeader(aln, summary, aln.getReferenceIndex(), summary.getTailClipPos(), ideal,
+						String header = RealignmentRecordSummary.makeFastqHeader(aln, summary, sourceBAMInfo.getSequenceName(aln.getReferenceIndex()), summary.getTailClipPos(), ideal,
 								Utilities.sequenceByteToString(summary.getTailClipAlignedSequence(), false), '+');
 						String clipSeq = Utilities.sequenceByteToString(tail, false); // make breakpoint at 3' end
 						String clipQual = Utilities.qualityByteToString(summary.getTailClipQuality(), false);
@@ -664,20 +664,20 @@ public void delete_empty() {
 	private class AlignmentInfo {
 		public int refIdx, pos5p, mrefIdx, mpos5p;
 		public String strands;
-		public boolean singleEnd;
+		public boolean paired;
 
 		public AlignmentInfo(SAMRecord aln) {
 			refIdx = aln.getReferenceIndex();
 			pos5p = aln.getReadNegativeStrandFlag() ? aln.getUnclippedEnd() : aln.getUnclippedStart();
 		    String strand = aln.getReadNegativeStrandFlag() ? "-" : "+";
 
 			if (aln.getReadPairedFlag()) {
-				singleEnd = true;
+				paired = true;
 			    mrefIdx = aln.getMateReferenceIndex();
 			    mpos5p = aln.getMateNegativeStrandFlag() ? aln.getMateAlignmentStart()+aln.getReadLength() : aln.getMateAlignmentStart();
 			    strands = strand + (aln.getMateNegativeStrandFlag() ? "-" : "+");
 			} else {
-				singleEnd = false;
+				paired = false;
 				mrefIdx = -1;
 				mpos5p = -1;
 				strands = strand;
@@ -686,7 +686,7 @@ public AlignmentInfo(SAMRecord aln) {
 
 		@Override
 		public int hashCode() {
-			if (this.singleEnd) {
+			if (! this.paired) {
 				return new HashCodeBuilder(17,31).
 						append( refIdx ).
 						append( pos5p ).
@@ -709,7 +709,7 @@ public boolean equals(Object other) {
 			if (other==this) return true;
 
 			AlignmentInfo oinfo = (AlignmentInfo)other;
-			if (singleEnd) {
+			if (!paired) {
 				return new EqualsBuilder().
 						append( refIdx, oinfo.refIdx ).
 						append( pos5p, oinfo.pos5p ).

diff --git a/src/net/wehi/socrates/RealignmentBAM.java b/src/net/wehi/socrates/RealignmentBAM.java
@@ -22,6 +22,7 @@
 import net.sf.samtools.util.SortingCollection;
 import net.sf.samtools.util.CloseableIterator;
 import net.wehi.socrates.util.ProgressVerbose;
+import net.wehi.socrates.util.SAMFileInfo;
 //import net.wehi.socrates.util.Utilities;
 
 /**
@@ -41,6 +42,7 @@ public static void makeRealignmentBAM(String inputBAMFilename, String outputBamF
 		try {
 			SortingCollection<SAMRecord> buffer = SortingCollection.newInstance(SAMRecord.class, new BAMRecordCodec(sam.getFileHeader()), 
 																				new SAMRecordCoordinateComparator(), 500000, new File("."));
+			SAMFileInfo info = new SAMFileInfo(sam);
 
 			SAMRecordIterator iter = sam.iterator();
 			while (iter.hasNext()) {
@@ -53,11 +55,19 @@ public static void makeRealignmentBAM(String inputBAMFilename, String outputBamF
 				String[] tokens = readName.split("&");
 				assert tokens.length == 5;
 
+				int mateRefId = info.getSequenceIndex(tokens[1]);
+				if (mateRefId < 0) {
+					System.err.println("Read: " + tokens[0]);
+					System.err.println("\tAnchor reference sequence " + tokens[1] + " does not exist in realignment BAM file.");
+					System.err.println("\tSkipped");
+					continue;
+				}
+
 				aln.setReadName( tokens[0] );
 				aln.setFirstOfPairFlag(true);
 				aln.setProperPairFlag(false);
 				aln.setReadPairedFlag(true);
-				aln.setMateReferenceIndex( Integer.parseInt(tokens[1]) );
+				aln.setMateReferenceIndex( mateRefId );
 				aln.setMateAlignmentStart( Integer.parseInt(tokens[2]) );
 				aln.setMateNegativeStrandFlag( tokens[3].equals("-") );
 				aln.setMateUnmappedFlag( !(tokens[4].equals("1")) );

diff --git a/src/net/wehi/socrates/util/RealignmentRecordSummary.java b/src/net/wehi/socrates/util/RealignmentRecordSummary.java
@@ -59,7 +59,7 @@ public RealignmentRecordSummary(SAMRecord realign) {
 	}
 
 	public static String makeFastqHeader(SAMRecord aln, SAMRecordSummary summary, 
-			int bpChr, int bpPos, boolean isIdeal, String alignSeq, char clipStrand) {
+			String bpChr, int bpPos, boolean isIdeal, String alignSeq, char clipStrand) {
 		StringBuilder name;
 		if (aln.getReadPairedFlag())
 			name = new StringBuilder(aln.getReadName() + (aln.getFirstOfPairFlag()?"/1":"/2"));