/**
* Convert a BAM file containing paried-end tags to the ascii "pair" format used for HiC.
*
* @param inputBam
* @param outputFile
* @throws IOException
*/
public static void filterBam(String inputBam, String outputFile, List<Chromosome> chromosomes)
throws IOException {
CloseableIterator<Alignment> iter = null;
AlignmentReader reader = null;
PrintWriter pw = null;
HashSet allChroms = new HashSet(chromosomes);
try {
pw = new PrintWriter(new FileWriter(outputFile));
reader = AlignmentReaderFactory.getReader(inputBam, false);
iter = reader.iterator();
while (iter.hasNext()) {
Alignment alignment = iter.next();
ReadMate mate = alignment.getMate();
// Filter unpaired and "normal" pairs. Only interested in abnormals
if (alignment.isPaired()
&& alignment.isMapped()
&& alignment.getMappingQuality() > 10
&& mate != null
&& mate.isMapped()
&& allChroms.contains(alignment.getChr())
&& allChroms.contains(mate.getChr())
&& (!alignment.getChr().equals(mate.getChr())
|| alignment.getInferredInsertSize() > 1000)) {
// Each pair is represented twice in the file, keep the record with the "leftmost"
// coordinate
if (alignment.getStart() < mate.getStart()) {
String strand = alignment.isNegativeStrand() ? "-" : "+";
String mateStrand = mate.isNegativeStrand() ? "-" : "+";
pw.println(
alignment.getReadName()
+ "\t"
+ alignment.getChr()
+ "\t"
+ alignment.getStart()
+ "\t"
+ strand
+ "\t.\t"
+ mate.getChr()
+ "\t"
+ mate.getStart()
+ "\t"
+ mateStrand);
}
}
}
} finally {
pw.close();
iter.close();
reader.close();
}
}
