/** Ensure that splitting read groups still works when dealing with a sample-split pileup. */
@Test
public void testSplitBySample() {
SAMReadGroupRecord readGroupOne = new SAMReadGroupRecord("rg1");
readGroupOne.setSample("sample1");
SAMReadGroupRecord readGroupTwo = new SAMReadGroupRecord("rg2");
readGroupTwo.setSample("sample2");
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
header.addReadGroup(readGroupOne);
header.addReadGroup(readGroupTwo);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
read1.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
read2.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read3 = ArtificialSAMUtils.createArtificialRead(header, "read3", 0, 1, 10);
read3.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read4 = ArtificialSAMUtils.createArtificialRead(header, "read4", 0, 1, 10);
read4.setAttribute("RG", readGroupTwo.getId());
ReadBackedPileupImpl sample1Pileup =
new ReadBackedPileupImpl(null, Arrays.asList(read1, read3), Arrays.asList(1, 1));
ReadBackedPileupImpl sample2Pileup =
new ReadBackedPileupImpl(null, Arrays.asList(read2, read4), Arrays.asList(1, 1));
Map<String, ReadBackedPileupImpl> sampleToPileupMap =
new HashMap<String, ReadBackedPileupImpl>();
sampleToPileupMap.put(readGroupOne.getSample(), sample1Pileup);
sampleToPileupMap.put(readGroupTwo.getSample(), sample2Pileup);
ReadBackedPileup compositePileup = new ReadBackedPileupImpl(null, sampleToPileupMap);
ReadBackedPileup rg1Pileup = compositePileup.getPileupForReadGroup("rg1");
List<GATKSAMRecord> rg1Reads = rg1Pileup.getReads();
Assert.assertEquals(rg1Reads.size(), 2, "Wrong number of reads in read group rg1");
Assert.assertEquals(
rg1Reads.get(0), read1, "Read " + read1.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(1), read3, "Read " + read3.getReadName() + " should be in rg1 but isn't");
ReadBackedPileup rg2Pileup = compositePileup.getPileupForReadGroup("rg2");
List<GATKSAMRecord> rg2Reads = rg2Pileup.getReads();
Assert.assertEquals(rg1Reads.size(), 2, "Wrong number of reads in read group rg2");
Assert.assertEquals(
rg2Reads.get(0), read2, "Read " + read2.getReadName() + " should be in rg2 but isn't");
Assert.assertEquals(
rg2Reads.get(1), read4, "Read " + read4.getReadName() + " should be in rg2 but isn't");
}
@Test
public void testGetPileupForSample() {
String sample1 = "sample1";
String sample2 = "sample2";
SAMReadGroupRecord readGroupOne = new SAMReadGroupRecord("rg1");
readGroupOne.setSample(sample1);
SAMReadGroupRecord readGroupTwo = new SAMReadGroupRecord("rg2");
readGroupTwo.setSample(sample2);
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
header.addReadGroup(readGroupOne);
header.addReadGroup(readGroupTwo);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
read1.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
read2.setAttribute("RG", readGroupTwo.getId());
Map<String, ReadBackedPileupImpl> sampleToPileupMap =
new HashMap<String, ReadBackedPileupImpl>();
sampleToPileupMap.put(
sample1, new ReadBackedPileupImpl(null, Collections.singletonList(read1), 0));
sampleToPileupMap.put(
sample2, new ReadBackedPileupImpl(null, Collections.singletonList(read2), 0));
ReadBackedPileup pileup = new ReadBackedPileupImpl(null, sampleToPileupMap);
ReadBackedPileup sample2Pileup = pileup.getPileupForSample(sample2);
Assert.assertEquals(
sample2Pileup.getNumberOfElements(), 1, "Sample 2 pileup has wrong number of elements");
Assert.assertEquals(
sample2Pileup.getReads().get(0), read2, "Sample 2 pileup has incorrect read");
ReadBackedPileup missingSamplePileup = pileup.getPileupForSample("missing");
Assert.assertNull(missingSamplePileup, "Pileup for sample 'missing' should be null but isn't");
missingSamplePileup = pileup.getPileupForSample("not here");
Assert.assertNull(missingSamplePileup, "Pileup for sample 'not here' should be null but isn't");
}
/** Ensure that basic read group splitting works. */
@Test
public void testSplitByReadGroup() {
SAMReadGroupRecord readGroupOne = new SAMReadGroupRecord("rg1");
SAMReadGroupRecord readGroupTwo = new SAMReadGroupRecord("rg2");
SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(1, 1, 1000);
header.addReadGroup(readGroupOne);
header.addReadGroup(readGroupTwo);
GATKSAMRecord read1 = ArtificialSAMUtils.createArtificialRead(header, "read1", 0, 1, 10);
read1.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read2 = ArtificialSAMUtils.createArtificialRead(header, "read2", 0, 1, 10);
read2.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read3 = ArtificialSAMUtils.createArtificialRead(header, "read3", 0, 1, 10);
read3.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read4 = ArtificialSAMUtils.createArtificialRead(header, "read4", 0, 1, 10);
read4.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read5 = ArtificialSAMUtils.createArtificialRead(header, "read5", 0, 1, 10);
read5.setAttribute("RG", readGroupTwo.getId());
GATKSAMRecord read6 = ArtificialSAMUtils.createArtificialRead(header, "read6", 0, 1, 10);
read6.setAttribute("RG", readGroupOne.getId());
GATKSAMRecord read7 = ArtificialSAMUtils.createArtificialRead(header, "read7", 0, 1, 10);
read7.setAttribute("RG", readGroupOne.getId());
ReadBackedPileup pileup =
new ReadBackedPileupImpl(
null,
Arrays.asList(read1, read2, read3, read4, read5, read6, read7),
Arrays.asList(1, 1, 1, 1, 1, 1, 1));
ReadBackedPileup rg1Pileup = pileup.getPileupForReadGroup("rg1");
List<GATKSAMRecord> rg1Reads = rg1Pileup.getReads();
Assert.assertEquals(rg1Reads.size(), 4, "Wrong number of reads in read group rg1");
Assert.assertEquals(
rg1Reads.get(0), read1, "Read " + read1.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(1), read3, "Read " + read3.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(2), read6, "Read " + read6.getReadName() + " should be in rg1 but isn't");
Assert.assertEquals(
rg1Reads.get(3), read7, "Read " + read7.getReadName() + " should be in rg1 but isn't");
ReadBackedPileup rg2Pileup = pileup.getPileupForReadGroup("rg2");
List<GATKSAMRecord> rg2Reads = rg2Pileup.getReads();
Assert.assertEquals(rg2Reads.size(), 3, "Wrong number of reads in read group rg2");
Assert.assertEquals(
rg2Reads.get(0), read2, "Read " + read2.getReadName() + " should be in rg2 but isn't");
Assert.assertEquals(
rg2Reads.get(1), read4, "Read " + read4.getReadName() + " should be in rg2 but isn't");
Assert.assertEquals(
rg2Reads.get(2), read5, "Read " + read5.getReadName() + " should be in rg2 but isn't");
}
/** Get the ID of the readgroup. */
public static String getID(final SAMReadGroupRecord rg) {
final String pu = rg.getPlatformUnit();
return pu == null ? rg.getId() : pu;
}
@Override
public void onTraversalDone(Integer sum) {
final GATKReport report = new GATKReport();
report.addTable(TABLE_NAME, "Table of read group properties", 12);
GATKReportTable table = report.getTable(TABLE_NAME);
DateFormat dateFormatter = DateFormat.getDateInstance(DateFormat.SHORT);
table.addColumn("readgroup");
// * Emits a GATKReport containing read group, sample, library, platform, center, median insert
// size and
// * median read length for each read group in every BAM file.
table.addColumn("sample", "%s");
table.addColumn("library", "%s");
table.addColumn("platform", "%s");
table.addColumn("center", "%s");
table.addColumn("date", "%s");
table.addColumn("has.any.reads");
table.addColumn("is.paired.end");
table.addColumn("n.reads.analyzed", "%d");
table.addColumn("simple.read.type", "%s");
table.addColumn("median.read.length");
table.addColumn("median.insert.size");
for (final SAMReadGroupRecord rg : getToolkit().getSAMFileHeader().getReadGroups()) {
final String rgID = rg.getId();
table.addRowID(rgID, true);
PerReadGroupInfo info = readGroupInfo.get(rgID);
// we are paired if > 25% of reads are paired
final boolean isPaired = info.nReadsPaired / (1.0 * (info.nReadsSeen + 1)) > 0.25;
final boolean hasAnyReads = info.nReadsSeen > 0;
final int readLength = info.readLength.getMedian(0);
setTableValue(table, rgID, "sample", rg.getSample());
setTableValue(table, rgID, "library", rg.getLibrary());
setTableValue(table, rgID, "platform", rg.getPlatform());
setTableValue(table, rgID, "center", rg.getSequencingCenter());
try {
setTableValue(
table,
rgID,
"date",
rg.getRunDate() != null ? dateFormatter.format(rg.getRunDate()) : "NA");
} catch (NullPointerException e) {
// TODO: remove me when bug in Picard is fixed that causes NPE when date isn't present
setTableValue(table, rgID, "date", "NA");
}
setTableValue(table, rgID, "has.any.reads", hasAnyReads);
setTableValue(table, rgID, "is.paired.end", isPaired);
setTableValue(table, rgID, "n.reads.analyzed", info.nReadsSeen);
setTableValue(
table,
rgID,
"simple.read.type",
hasAnyReads ? String.format("%dx%d", isPaired ? 2 : 1, readLength) : "NA");
setTableValue(table, rgID, "median.read.length", hasAnyReads ? readLength : "NA");
setTableValue(
table,
rgID,
"median.insert.size",
hasAnyReads && isPaired ? info.insertSize.getMedian(0) : "NA");
}
report.print(out);
}
@Override
public void initialize() {
for (final SAMReadGroupRecord rg : getToolkit().getSAMFileHeader().getReadGroups()) {
readGroupInfo.put(rg.getId(), new PerReadGroupInfo());
}
}