/**
* Generates tab delimited string containing details about the passed SAMReadGroupRecord
*
* @param readGroupRecord record
* @return tab delimited string containing details about the SAMReadGroupRecord
*/
private String getReadGroupDetails(final SAMReadGroupRecord readGroupRecord) {
final List<String> elements = new ArrayList<>(5);
final String tmp[] =
readGroupRecord
.getPlatformUnit()
.split("\\."); // Expect to look like: D047KACXX110901.1.ACCAACTG
String runBarcode = "?";
String lane = "?";
String molBarcode = "?";
if ((tmp.length == 3) || (tmp.length == 2)) {
runBarcode = tmp[0];
lane = tmp[1];
molBarcode =
(tmp.length == 3)
? tmp[2]
: ""; // In older BAMS there may be no molecular barcode sequence
} else {
log.error("Unexpected format " + readGroupRecord.getPlatformUnit() + " for PU attribute");
}
elements.add(runBarcode);
elements.add(lane);
elements.add(molBarcode);
elements.add(readGroupRecord.getLibrary());
elements.add(readGroupRecord.getSample());
return String.join("\t", elements);
}
@Override
protected void setup(final SAMFileHeader header, final File samFile) {
final String outext =
(null != FILE_EXTENSION) ? FILE_EXTENSION : ""; // Add a file extension if desired
preAdapterSummaryOut =
new File(OUTPUT + SequencingArtifactMetrics.PRE_ADAPTER_SUMMARY_EXT + outext);
preAdapterDetailsOut =
new File(OUTPUT + SequencingArtifactMetrics.PRE_ADAPTER_DETAILS_EXT + outext);
baitBiasSummaryOut =
new File(OUTPUT + SequencingArtifactMetrics.BAIT_BIAS_SUMMARY_EXT + outext);
baitBiasDetailsOut =
new File(OUTPUT + SequencingArtifactMetrics.BAIT_BIAS_DETAILS_EXT + outext);
IOUtil.assertFileIsWritable(preAdapterSummaryOut);
IOUtil.assertFileIsWritable(preAdapterDetailsOut);
IOUtil.assertFileIsWritable(baitBiasSummaryOut);
IOUtil.assertFileIsWritable(baitBiasDetailsOut);
for (final SAMReadGroupRecord rec : header.getReadGroups()) {
samples.add(getOrElse(rec.getSample(), UNKNOWN_SAMPLE));
libraries.add(getOrElse(rec.getLibrary(), UNKNOWN_LIBRARY));
}
if (INTERVALS != null) {
IOUtil.assertFileIsReadable(INTERVALS);
intervalMask =
new IntervalListReferenceSequenceMask(IntervalList.fromFile(INTERVALS).uniqued());
}
if (DB_SNP != null) {
IOUtil.assertFileIsReadable(DB_SNP);
dbSnpMask = new DbSnpBitSetUtil(DB_SNP, header.getSequenceDictionary());
}
// set record-level filters
final List<SamRecordFilter> filters = new ArrayList<SamRecordFilter>();
filters.add(new FailsVendorReadQualityFilter());
filters.add(new NotPrimaryAlignmentFilter());
filters.add(new DuplicateReadFilter());
filters.add(new AlignedFilter(true)); // discard unmapped reads
filters.add(new MappingQualityFilter(MINIMUM_MAPPING_QUALITY));
if (!INCLUDE_UNPAIRED) {
final int effectiveMaxInsertSize =
(MAXIMUM_INSERT_SIZE == 0) ? Integer.MAX_VALUE : MAXIMUM_INSERT_SIZE;
filters.add(new InsertSizeFilter(MINIMUM_INSERT_SIZE, effectiveMaxInsertSize));
}
recordFilter = new AggregateFilter(filters);
// set up the artifact counters
final String sampleAlias = StringUtil.join(",", new ArrayList<String>(samples));
for (final String library : libraries) {
artifactCounters.put(
library, new ArtifactCounter(sampleAlias, library, CONTEXT_SIZE, TANDEM_READS));
}
}
/**
* Method that combines the fingerprint evidence across all the read groups for the same library
* and then produces a matrix of LOD scores for comparing every library with every other library.
*/
private void crossCheckLibraries(
final Map<SAMReadGroupRecord, Fingerprint> fingerprints, final PrintStream out) {
final List<Fingerprint> fixedFps = new ArrayList<>();
for (final SAMReadGroupRecord rg : fingerprints.keySet()) {
final Fingerprint old = fingerprints.get(rg);
final String name = rg.getSample() + "::" + rg.getLibrary();
final Fingerprint newFp = new Fingerprint(name, old.getSource(), old.getInfo());
newFp.putAll(old);
fixedFps.add(newFp);
}
crossCheckSamples(fixedFps, out);
}
/**
* Returns the library associated with the provided read's read group.
*
* @param read read whose library to retrieve
* @param header SAM header containing read groups
* @return the library for the provided read's read group as a String, or null if the read has no
* read group.
*/
public static String getLibrary(final GATKRead read, final SAMFileHeader header) {
final SAMReadGroupRecord readGroup = getSAMReadGroupRecord(read, header);
return readGroup != null ? readGroup.getLibrary() : null;
}
@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);
}
