@Test
public void testEncodeDecode() throws Exception {
String testFile = TestUtils.DATA_DIR + "sam/NA12878.muc1.test.sam";
SAMReader reader = new SAMReader(testFile);
Iterator<PicardAlignment> inputAlignmentIterator = reader.iterator();
ArrayList<PicardAlignment> inputAlignmentList = new ArrayList<PicardAlignment>();
while (inputAlignmentIterator.hasNext()) {
PicardAlignment al = inputAlignmentIterator.next();
inputAlignmentList.add(al);
}
ByteArrayOutputStream bos = new ByteArrayOutputStream();
FeatureEncoder<PicardAlignment> alignmentEncoder = new SamAlignmentEncoder();
alignmentEncoder.encodeAll(bos, reader.iterator());
ByteArrayInputStream bis = new ByteArrayInputStream(bos.toByteArray());
FeatureDecoder<PicardAlignment> alignmentDecoder = new AlignmentDecoder();
Iterator<PicardAlignment> decodedAlignments = alignmentDecoder.decodeAll(bis, false);
int ind = 0;
while (decodedAlignments.hasNext()) {
Alignment act = decodedAlignments.next();
Alignment exp = inputAlignmentList.get(ind++);
TestUtils.assertFeaturesEqual(exp, act);
assertEquals(exp.getCigarString(), act.getCigarString());
}
assertEquals("Different number of alignments read in as out", inputAlignmentList.size(), ind);
}
/**
* 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();
}
}