zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merged bug fix from Stable into Unstable

This commit is contained in:
Matt Hanna 2011-09-12 21:54:11 -04:00
parent 4e116760f4 e63d9d8f8e
commit 64707c33bb
3 changed files with 45 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
public/java/src/org/broadinstitute/sting/gatk/datasources/reads/BAMScheduler.java
View File

@ -26,6 +26,7 @@ package org.broadinstitute.sting.gatk.datasources.reads;
import net.sf.picard.util.PeekableIterator;
import net.sf.samtools.GATKBAMFileSpan;
import net.sf.samtools.GATKChunk;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.GenomeLocSortedSet;
@ -84,7 +85,7 @@ public class BAMScheduler implements Iterator<FilePointer> {
if(currentLocus == GenomeLoc.UNMAPPED) {
nextFilePointer = new FilePointer(GenomeLoc.UNMAPPED);
for(SAMReaderID id: dataSource.getReaderIDs())
nextFilePointer.addFileSpans(id,new GATKBAMFileSpan());
nextFilePointer.addFileSpans(id,new GATKBAMFileSpan(new GATKChunk(indexFiles.get(id).getStartOfLastLinearBin(),Long.MAX_VALUE)));
currentLocus = null;
continue;
}

39
public/java/src/org/broadinstitute/sting/gatk/datasources/reads/GATKBAMIndex.java
View File

@ -215,6 +215,45 @@ public class GATKBAMIndex {
return (new GATKBin(bin).getBinNumber()-levelStart+1)*(BIN_GENOMIC_SPAN /levelSize);
}
/**
* Use to get close to the unmapped reads at the end of a BAM file.
* @return The file offset of the first record in the last linear bin, or -1
* if there are no elements in linear bins (i.e. no mapped reads).
*/
public long getStartOfLastLinearBin() {
openIndexFile();
seek(4);
final int sequenceCount = readInteger();
// Because no reads may align to the last sequence in the sequence dictionary,
// grab the last element of the linear index for each sequence, and return
// the last one from the last sequence that has one.
long lastLinearIndexPointer = -1;
for (int i = 0; i < sequenceCount; i++) {
// System.out.println("# Sequence TID: " + i);
final int nBins = readInteger();
// System.out.println("# nBins: " + nBins);
for (int j1 = 0; j1 < nBins; j1++) {
// Skip bin #
skipBytes(4);
final int nChunks = readInteger();
// Skip chunks
skipBytes(16 * nChunks);
}
final int nLinearBins = readInteger();
if (nLinearBins > 0) {
// Skip to last element of list of linear bins
skipBytes(8 * (nLinearBins - 1));
lastLinearIndexPointer = readLongs(1)[0];
}
}
closeIndexFile();
return lastLinearIndexPointer;
}
/**
* Gets the possible number of bins for a given reference sequence.
* @return How many bins could possibly be used according to this indexing scheme to index a single contig.

21
public/java/src/org/broadinstitute/sting/gatk/datasources/reads/ReadShardStrategy.java
View File

@ -134,24 +134,11 @@ public class ReadShardStrategy implements ShardStrategy {
Map<SAMReaderID,SAMFileSpan> selectedReaders = new HashMap<SAMReaderID,SAMFileSpan>();
while(selectedReaders.size() == 0 && currentFilePointer != null) {
shardPosition = currentFilePointer.fileSpans;
for(SAMReaderID id: shardPosition.keySet()) {
// If the region contains location information (in other words, it is not at
// the start of the unmapped region), add the region.
if(currentFilePointer.isRegionUnmapped) {
// If the region is unmapped and no location data exists, add a null as an indicator to
// start at the next unmapped region.
if(!isIntoUnmappedRegion) {
selectedReaders.put(id,null);
isIntoUnmappedRegion = true;
}
else
selectedReaders.put(id,position.get(id));
}
else {
SAMFileSpan fileSpan = shardPosition.get(id).removeContentsBefore(position.get(id));
if(!fileSpan.isEmpty())
selectedReaders.put(id,fileSpan);
}
SAMFileSpan fileSpan = shardPosition.get(id).removeContentsBefore(position.get(id));
if(!fileSpan.isEmpty())
selectedReaders.put(id,fileSpan);
}
if(selectedReaders.size() > 0) {