public static Map<VariantContext.Type, List<VariantContext>> separateVariantContextsByType(
Collection<VariantContext> VCs) {
HashMap<VariantContext.Type, List<VariantContext>> mappedVCs =
new HashMap<VariantContext.Type, List<VariantContext>>();
for (VariantContext vc : VCs) {
// look at previous variant contexts of different type. If:
// a) otherVC has alleles which are subset of vc, remove otherVC from its list and add otherVC
// to vc's list
// b) vc has alleles which are subset of otherVC. Then, add vc to otherVC's type list (rather,
// do nothing since vc will be added automatically to its list)
// c) neither: do nothing, just add vc to its own list
boolean addtoOwnList = true;
for (VariantContext.Type type : VariantContext.Type.values()) {
if (type.equals(vc.getType())) continue;
if (!mappedVCs.containsKey(type)) continue;
List<VariantContext> vcList = mappedVCs.get(type);
for (int k = 0; k < vcList.size(); k++) {
VariantContext otherVC = vcList.get(k);
if (allelesAreSubset(otherVC, vc)) {
// otherVC has a type different than vc and its alleles are a subset of vc: remove
// otherVC from its list and add it to vc's type list
vcList.remove(k);
// avoid having empty lists
if (vcList.size() == 0) mappedVCs.remove(vcList);
if (!mappedVCs.containsKey(vc.getType()))
mappedVCs.put(vc.getType(), new ArrayList<VariantContext>());
mappedVCs.get(vc.getType()).add(otherVC);
break;
} else if (allelesAreSubset(vc, otherVC)) {
// vc has a type different than otherVC and its alleles are a subset of VC: add vc to
// otherVC's type list and don't add to its own
mappedVCs.get(type).add(vc);
addtoOwnList = false;
break;
}
}
}
if (addtoOwnList) {
if (!mappedVCs.containsKey(vc.getType()))
mappedVCs.put(vc.getType(), new ArrayList<VariantContext>());
mappedVCs.get(vc.getType()).add(vc);
}
}
return mappedVCs;
}
/** Set up the VCF writer, the sample expressions and regexs, and the JEXL matcher */
public void initialize() {
// Get list of samples to include in the output
List<String> rodNames = Arrays.asList(variantCollection.variants.getName());
Map<String, VCFHeader> vcfRods = VCFUtils.getVCFHeadersFromRods(getToolkit(), rodNames);
TreeSet<String> vcfSamples =
new TreeSet<String>(
SampleUtils.getSampleList(
vcfRods, VariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE));
Collection<String> samplesFromFile = SampleUtils.getSamplesFromFiles(sampleFiles);
Collection<String> samplesFromExpressions =
SampleUtils.matchSamplesExpressions(vcfSamples, sampleExpressions);
// first, add any requested samples
samples.addAll(samplesFromFile);
samples.addAll(samplesFromExpressions);
samples.addAll(sampleNames);
// if none were requested, we want all of them
if (samples.isEmpty()) {
samples.addAll(vcfSamples);
NO_SAMPLES_SPECIFIED = true;
}
// now, exclude any requested samples
Collection<String> XLsamplesFromFile = SampleUtils.getSamplesFromFiles(XLsampleFiles);
samples.removeAll(XLsamplesFromFile);
samples.removeAll(XLsampleNames);
if (samples.size() == 0 && !NO_SAMPLES_SPECIFIED)
throw new UserException(
"All samples requested to be included were also requested to be excluded.");
for (String sample : samples) logger.info("Including sample '" + sample + "'");
// if user specified types to include, add these, otherwise, add all possible variant context
// types to list of vc types to include
if (TYPES_TO_INCLUDE.isEmpty()) {
for (VariantContext.Type t : VariantContext.Type.values()) selectedTypes.add(t);
} else {
for (VariantContext.Type t : TYPES_TO_INCLUDE) selectedTypes.add(t);
}
// Initialize VCF header
Set<VCFHeaderLine> headerLines = VCFUtils.smartMergeHeaders(vcfRods.values(), logger);
headerLines.add(new VCFHeaderLine("source", "SelectVariants"));
if (KEEP_ORIGINAL_CHR_COUNTS) {
headerLines.add(
new VCFFormatHeaderLine("AC_Orig", 1, VCFHeaderLineType.Integer, "Original AC"));
headerLines.add(
new VCFFormatHeaderLine("AF_Orig", 1, VCFHeaderLineType.Float, "Original AF"));
headerLines.add(
new VCFFormatHeaderLine("AN_Orig", 1, VCFHeaderLineType.Integer, "Original AN"));
}
vcfWriter.writeHeader(new VCFHeader(headerLines, samples));
for (int i = 0; i < SELECT_EXPRESSIONS.size(); i++) {
// It's not necessary that the user supply select names for the JEXL expressions, since those
// expressions will only be needed for omitting records. Make up the select names here.
selectNames.add(String.format("select-%d", i));
}
jexls = VariantContextUtils.initializeMatchExps(selectNames, SELECT_EXPRESSIONS);
// Look at the parameters to decide which analysis to perform
DISCORDANCE_ONLY = discordanceTrack.isBound();
if (DISCORDANCE_ONLY)
logger.info(
"Selecting only variants discordant with the track: " + discordanceTrack.getName());
CONCORDANCE_ONLY = concordanceTrack.isBound();
if (CONCORDANCE_ONLY)
logger.info(
"Selecting only variants concordant with the track: " + concordanceTrack.getName());
if (MENDELIAN_VIOLATIONS) {
if (FAMILY_STRUCTURE_FILE != null) {
try {
for (final String line : new XReadLines(FAMILY_STRUCTURE_FILE)) {
MendelianViolation mv =
new MendelianViolation(line, MENDELIAN_VIOLATION_QUAL_THRESHOLD);
if (samples.contains(mv.getSampleChild())
&& samples.contains(mv.getSampleDad())
&& samples.contains(mv.getSampleMom())) mvSet.add(mv);
}
} catch (FileNotFoundException e) {
throw new UserException.CouldNotReadInputFile(FAMILY_STRUCTURE_FILE, e);
}
if (outMVFile != null)
try {
outMVFileStream = new PrintStream(outMVFile);
} catch (FileNotFoundException e) {
throw new UserException.CouldNotCreateOutputFile(
outMVFile, "Can't open output file", e);
}
} else
mvSet.add(new MendelianViolation(FAMILY_STRUCTURE, MENDELIAN_VIOLATION_QUAL_THRESHOLD));
} else if (!FAMILY_STRUCTURE.isEmpty()) {
mvSet.add(new MendelianViolation(FAMILY_STRUCTURE, MENDELIAN_VIOLATION_QUAL_THRESHOLD));
MENDELIAN_VIOLATIONS = true;
}
SELECT_RANDOM_NUMBER = numRandom > 0;
if (SELECT_RANDOM_NUMBER) {
logger.info("Selecting " + numRandom + " variants at random from the variant track");
variantArray = new RandomVariantStructure[numRandom];
}
SELECT_RANDOM_FRACTION = fractionRandom > 0;
if (SELECT_RANDOM_FRACTION)
logger.info(
"Selecting approximately "
+ 100.0 * fractionRandom
+ "% of the variants at random from the variant track");
}