// necessary to not overload equals for genotypes
private void assertGenotypesAreMostlyEqual(GenotypesContext actual, GenotypesContext expected) {
if (actual == expected) {
return;
}
if (actual == null || expected == null) {
Assert.fail("Maps not equal: expected: " + expected + " and actual: " + actual);
}
if (actual.size() != expected.size()) {
Assert.fail("Maps do not have the same size:" + actual.size() + " != " + expected.size());
}
for (Genotype value : actual) {
Genotype expectedValue = expected.get(value.getSampleName());
Assert.assertEquals(
value.getAlleles(), expectedValue.getAlleles(), "Alleles in Genotype aren't equal");
Assert.assertEquals(value.getGQ(), expectedValue.getGQ(), "GQ values aren't equal");
Assert.assertEquals(
value.hasLikelihoods(),
expectedValue.hasLikelihoods(),
"Either both have likelihoods or both not");
if (value.hasLikelihoods())
Assert.assertEquals(
value.getLikelihoods().getAsVector(),
expectedValue.getLikelihoods().getAsVector(),
"Genotype likelihoods aren't equal");
}
}
private static void mergeGenotypes(
GenotypesContext mergedGenotypes,
VariantContext oneVC,
AlleleMapper alleleMapping,
boolean uniqifySamples) {
for (Genotype g : oneVC.getGenotypes()) {
String name = mergedSampleName(oneVC.getSource(), g.getSampleName(), uniqifySamples);
if (!mergedGenotypes.containsSample(name)) {
// only add if the name is new
Genotype newG = g;
if (uniqifySamples || alleleMapping.needsRemapping()) {
final List<Allele> alleles =
alleleMapping.needsRemapping() ? alleleMapping.remap(g.getAlleles()) : g.getAlleles();
newG =
new Genotype(
name,
alleles,
g.getLog10PError(),
g.getFilters(),
g.getAttributes(),
g.isPhased());
}
mergedGenotypes.add(newG);
}
}
}
/**
* Returns a context identical to this with the REF and ALT alleles reverse complemented.
*
* @param vc variant context
* @return new vc
*/
public static VariantContext reverseComplement(VariantContext vc) {
// create a mapping from original allele to reverse complemented allele
HashMap<Allele, Allele> alleleMap = new HashMap<Allele, Allele>(vc.getAlleles().size());
for (Allele originalAllele : vc.getAlleles()) {
Allele newAllele;
if (originalAllele.isNoCall() || originalAllele.isNull()) newAllele = originalAllele;
else
newAllele =
Allele.create(
BaseUtils.simpleReverseComplement(originalAllele.getBases()),
originalAllele.isReference());
alleleMap.put(originalAllele, newAllele);
}
// create new Genotype objects
GenotypesContext newGenotypes = GenotypesContext.create(vc.getNSamples());
for (final Genotype genotype : vc.getGenotypes()) {
List<Allele> newAlleles = new ArrayList<Allele>();
for (Allele allele : genotype.getAlleles()) {
Allele newAllele = alleleMap.get(allele);
if (newAllele == null) newAllele = Allele.NO_CALL;
newAlleles.add(newAllele);
}
newGenotypes.add(Genotype.modifyAlleles(genotype, newAlleles));
}
return new VariantContextBuilder(vc).alleles(alleleMap.values()).genotypes(newGenotypes).make();
}
public static GenotypesContext stripPLs(GenotypesContext genotypes) {
GenotypesContext newGs = GenotypesContext.create(genotypes.size());
for (final Genotype g : genotypes) {
newGs.add(g.hasLikelihoods() ? removePLs(g) : g);
}
return newGs;
}
public static VariantContext purgeUnallowedGenotypeAttributes(
VariantContext vc, Set<String> allowedAttributes) {
if (allowedAttributes == null) return vc;
GenotypesContext newGenotypes = GenotypesContext.create(vc.getNSamples());
for (final Genotype genotype : vc.getGenotypes()) {
Map<String, Object> attrs = new HashMap<String, Object>();
for (Map.Entry<String, Object> attr : genotype.getAttributes().entrySet()) {
if (allowedAttributes.contains(attr.getKey())) attrs.put(attr.getKey(), attr.getValue());
}
newGenotypes.add(Genotype.modifyAttributes(genotype, attrs));
}
return new VariantContextBuilder(vc).genotypes(newGenotypes).make();
}
public static VariantContextBuilder pruneVariantContext(
final VariantContextBuilder builder, Collection<String> keysToPreserve) {
final VariantContext vc = builder.make();
if (keysToPreserve == null) keysToPreserve = Collections.emptyList();
// VC info
final Map<String, Object> attributes = subsetAttributes(vc.commonInfo, keysToPreserve);
// Genotypes
final GenotypesContext genotypes = GenotypesContext.create(vc.getNSamples());
for (final Genotype g : vc.getGenotypes()) {
Map<String, Object> genotypeAttributes = subsetAttributes(g.commonInfo, keysToPreserve);
genotypes.add(
new Genotype(
g.getSampleName(),
g.getAlleles(),
g.getLog10PError(),
g.getFilters(),
genotypeAttributes,
g.isPhased()));
}
return builder.genotypes(genotypes).attributes(attributes);
}
/** @return true if the context has associated genotypes */
public boolean hasGenotypes() {
return !genotypes.isEmpty();
}
/** @return the number of samples in the context */
public int getNSamples() {
return genotypes.size();
}
public VariantContext subContextFromSamples(Set<String> sampleNames) {
VariantContextBuilder builder = new VariantContextBuilder(this);
GenotypesContext newGenotypes = genotypes.subsetToSamples(sampleNames);
return builder.genotypes(newGenotypes).alleles(allelesOfGenotypes(newGenotypes)).make();
}
public VariantContext subContextFromSamples(Set<String> sampleNames, Collection<Allele> alleles) {
VariantContextBuilder builder = new VariantContextBuilder(this);
return builder.genotypes(genotypes.subsetToSamples(sampleNames)).alleles(alleles).make();
}
public Iterable<Genotype> getGenotypesOrderedBy(Iterable<String> sampleOrdering) {
return genotypes.iterateInSampleNameOrder(sampleOrdering);
}
public static VariantContext createVariantContextWithTrimmedAlleles(VariantContext inputVC) {
// see if we need to trim common reference base from all alleles
boolean trimVC;
// We need to trim common reference base from all alleles in all genotypes if a ref base is
// common to all alleles
Allele refAllele = inputVC.getReference();
if (!inputVC.isVariant()) trimVC = false;
else if (refAllele.isNull()) trimVC = false;
else {
trimVC =
(AbstractVCFCodec.computeForwardClipping(
new ArrayList<Allele>(inputVC.getAlternateAlleles()),
inputVC.getReference().getDisplayString())
> 0);
}
// nothing to do if we don't need to trim bases
if (trimVC) {
List<Allele> alleles = new ArrayList<Allele>();
GenotypesContext genotypes = GenotypesContext.create();
// set the reference base for indels in the attributes
Map<String, Object> attributes = new TreeMap<String, Object>(inputVC.getAttributes());
Map<Allele, Allele> originalToTrimmedAlleleMap = new HashMap<Allele, Allele>();
for (Allele a : inputVC.getAlleles()) {
if (a.isSymbolic()) {
alleles.add(a);
originalToTrimmedAlleleMap.put(a, a);
} else {
// get bases for current allele and create a new one with trimmed bases
byte[] newBases = Arrays.copyOfRange(a.getBases(), 1, a.length());
Allele trimmedAllele = Allele.create(newBases, a.isReference());
alleles.add(trimmedAllele);
originalToTrimmedAlleleMap.put(a, trimmedAllele);
}
}
// detect case where we're trimming bases but resulting vc doesn't have any null allele. In
// that case, we keep original representation
// example: mixed records such as {TA*,TGA,TG}
boolean hasNullAlleles = false;
for (Allele a : originalToTrimmedAlleleMap.values()) {
if (a.isNull()) hasNullAlleles = true;
if (a.isReference()) refAllele = a;
}
if (!hasNullAlleles) return inputVC;
// now we can recreate new genotypes with trimmed alleles
for (final Genotype genotype : inputVC.getGenotypes()) {
List<Allele> originalAlleles = genotype.getAlleles();
List<Allele> trimmedAlleles = new ArrayList<Allele>();
for (Allele a : originalAlleles) {
if (a.isCalled()) trimmedAlleles.add(originalToTrimmedAlleleMap.get(a));
else trimmedAlleles.add(Allele.NO_CALL);
}
genotypes.add(Genotype.modifyAlleles(genotype, trimmedAlleles));
}
final VariantContextBuilder builder = new VariantContextBuilder(inputVC);
return builder
.alleles(alleles)
.genotypes(genotypes)
.attributes(attributes)
.referenceBaseForIndel(new Byte(inputVC.getReference().getBases()[0]))
.make();
}
return inputVC;
}
/**
* Merges VariantContexts into a single hybrid. Takes genotypes for common samples in priority
* order, if provided. If uniqifySamples is true, the priority order is ignored and names are
* created by concatenating the VC name with the sample name
*
* @param genomeLocParser loc parser
* @param unsortedVCs collection of unsorted VCs
* @param priorityListOfVCs priority list detailing the order in which we should grab the VCs
* @param filteredRecordMergeType merge type for filtered records
* @param genotypeMergeOptions merge option for genotypes
* @param annotateOrigin should we annotate the set it came from?
* @param printMessages should we print messages?
* @param setKey the key name of the set
* @param filteredAreUncalled are filtered records uncalled?
* @param mergeInfoWithMaxAC should we merge in info from the VC with maximum allele count?
* @return new VariantContext representing the merge of unsortedVCs
*/
public static VariantContext simpleMerge(
final GenomeLocParser genomeLocParser,
final Collection<VariantContext> unsortedVCs,
final List<String> priorityListOfVCs,
final FilteredRecordMergeType filteredRecordMergeType,
final GenotypeMergeType genotypeMergeOptions,
final boolean annotateOrigin,
final boolean printMessages,
final String setKey,
final boolean filteredAreUncalled,
final boolean mergeInfoWithMaxAC) {
if (unsortedVCs == null || unsortedVCs.size() == 0) return null;
if (annotateOrigin && priorityListOfVCs == null)
throw new IllegalArgumentException(
"Cannot merge calls and annotate their origins without a complete priority list of VariantContexts");
if (genotypeMergeOptions == GenotypeMergeType.REQUIRE_UNIQUE)
verifyUniqueSampleNames(unsortedVCs);
List<VariantContext> prepaddedVCs =
sortVariantContextsByPriority(unsortedVCs, priorityListOfVCs, genotypeMergeOptions);
// Make sure all variant contexts are padded with reference base in case of indels if necessary
List<VariantContext> VCs = new ArrayList<VariantContext>();
for (VariantContext vc : prepaddedVCs) {
// also a reasonable place to remove filtered calls, if needed
if (!filteredAreUncalled || vc.isNotFiltered())
VCs.add(createVariantContextWithPaddedAlleles(vc, false));
}
if (VCs.size() == 0) // everything is filtered out and we're filteredAreUncalled
return null;
// establish the baseline info from the first VC
final VariantContext first = VCs.get(0);
final String name = first.getSource();
final Allele refAllele = determineReferenceAllele(VCs);
final Set<Allele> alleles = new LinkedHashSet<Allele>();
final Set<String> filters = new TreeSet<String>();
final Map<String, Object> attributes = new TreeMap<String, Object>();
final Set<String> inconsistentAttributes = new HashSet<String>();
final Set<String> variantSources =
new HashSet<
String>(); // contains the set of sources we found in our set of VCs that are variant
final Set<String> rsIDs = new LinkedHashSet<String>(1); // most of the time there's one id
GenomeLoc loc = getLocation(genomeLocParser, first);
int depth = 0;
int maxAC = -1;
final Map<String, Object> attributesWithMaxAC = new TreeMap<String, Object>();
double log10PError = 1;
VariantContext vcWithMaxAC = null;
GenotypesContext genotypes = GenotypesContext.create();
// counting the number of filtered and variant VCs
int nFiltered = 0;
boolean remapped = false;
// cycle through and add info from the other VCs, making sure the loc/reference matches
for (VariantContext vc : VCs) {
if (loc.getStart() != vc.getStart()) // || !first.getReference().equals(vc.getReference()) )
throw new ReviewedStingException(
"BUG: attempting to merge VariantContexts with different start sites: first="
+ first.toString()
+ " second="
+ vc.toString());
if (getLocation(genomeLocParser, vc).size() > loc.size())
loc = getLocation(genomeLocParser, vc); // get the longest location
nFiltered += vc.isFiltered() ? 1 : 0;
if (vc.isVariant()) variantSources.add(vc.getSource());
AlleleMapper alleleMapping = resolveIncompatibleAlleles(refAllele, vc, alleles);
remapped = remapped || alleleMapping.needsRemapping();
alleles.addAll(alleleMapping.values());
mergeGenotypes(
genotypes, vc, alleleMapping, genotypeMergeOptions == GenotypeMergeType.UNIQUIFY);
log10PError = Math.min(log10PError, vc.isVariant() ? vc.getLog10PError() : 1);
filters.addAll(vc.getFilters());
//
// add attributes
//
// special case DP (add it up) and ID (just preserve it)
//
if (vc.hasAttribute(VCFConstants.DEPTH_KEY))
depth += vc.getAttributeAsInt(VCFConstants.DEPTH_KEY, 0);
if (vc.hasID()) rsIDs.add(vc.getID());
if (mergeInfoWithMaxAC && vc.hasAttribute(VCFConstants.ALLELE_COUNT_KEY)) {
String rawAlleleCounts = vc.getAttributeAsString(VCFConstants.ALLELE_COUNT_KEY, null);
// lets see if the string contains a , separator
if (rawAlleleCounts.contains(VCFConstants.INFO_FIELD_ARRAY_SEPARATOR)) {
List<String> alleleCountArray =
Arrays.asList(
rawAlleleCounts
.substring(1, rawAlleleCounts.length() - 1)
.split(VCFConstants.INFO_FIELD_ARRAY_SEPARATOR));
for (String alleleCount : alleleCountArray) {
final int ac = Integer.valueOf(alleleCount.trim());
if (ac > maxAC) {
maxAC = ac;
vcWithMaxAC = vc;
}
}
} else {
final int ac = Integer.valueOf(rawAlleleCounts);
if (ac > maxAC) {
maxAC = ac;
vcWithMaxAC = vc;
}
}
}
for (Map.Entry<String, Object> p : vc.getAttributes().entrySet()) {
String key = p.getKey();
// if we don't like the key already, don't go anywhere
if (!inconsistentAttributes.contains(key)) {
boolean alreadyFound = attributes.containsKey(key);
Object boundValue = attributes.get(key);
boolean boundIsMissingValue =
alreadyFound && boundValue.equals(VCFConstants.MISSING_VALUE_v4);
if (alreadyFound && !boundValue.equals(p.getValue()) && !boundIsMissingValue) {
// we found the value but we're inconsistent, put it in the exclude list
// System.out.printf("Inconsistent INFO values: %s => %s and %s%n", key, boundValue,
// p.getValue());
inconsistentAttributes.add(key);
attributes.remove(key);
} else if (!alreadyFound || boundIsMissingValue) { // no value
// if ( vc != first ) System.out.printf("Adding key %s => %s%n", p.getKey(),
// p.getValue());
attributes.put(key, p.getValue());
}
}
}
}
// if we have more alternate alleles in the merged VC than in one or more of the
// original VCs, we need to strip out the GL/PLs (because they are no longer accurate), as well
// as allele-dependent attributes like AC,AF
for (VariantContext vc : VCs) {
if (vc.alleles.size() == 1) continue;
if (hasPLIncompatibleAlleles(alleles, vc.alleles)) {
if (!genotypes.isEmpty())
logger.warn(
String.format(
"Stripping PLs at %s due incompatible alleles merged=%s vs. single=%s",
genomeLocParser.createGenomeLoc(vc), alleles, vc.alleles));
genotypes = stripPLs(genotypes);
// this will remove stale AC,AF attributed from vc
calculateChromosomeCounts(vc, attributes, true);
break;
}
}
// take the VC with the maxAC and pull the attributes into a modifiable map
if (mergeInfoWithMaxAC && vcWithMaxAC != null) {
attributesWithMaxAC.putAll(vcWithMaxAC.getAttributes());
}
// if at least one record was unfiltered and we want a union, clear all of the filters
if ((filteredRecordMergeType == FilteredRecordMergeType.KEEP_IF_ANY_UNFILTERED
&& nFiltered != VCs.size())
|| filteredRecordMergeType == FilteredRecordMergeType.KEEP_UNCONDITIONAL) filters.clear();
if (annotateOrigin) { // we care about where the call came from
String setValue;
if (nFiltered == 0
&& variantSources.size() == priorityListOfVCs.size()) // nothing was unfiltered
setValue = MERGE_INTERSECTION;
else if (nFiltered == VCs.size()) // everything was filtered out
setValue = MERGE_FILTER_IN_ALL;
else if (variantSources.isEmpty()) // everyone was reference
setValue = MERGE_REF_IN_ALL;
else {
LinkedHashSet<String> s = new LinkedHashSet<String>();
for (VariantContext vc : VCs)
if (vc.isVariant())
s.add(vc.isFiltered() ? MERGE_FILTER_PREFIX + vc.getSource() : vc.getSource());
setValue = Utils.join("-", s);
}
if (setKey != null) {
attributes.put(setKey, setValue);
if (mergeInfoWithMaxAC && vcWithMaxAC != null) {
attributesWithMaxAC.put(setKey, vcWithMaxAC.getSource());
}
}
}
if (depth > 0) attributes.put(VCFConstants.DEPTH_KEY, String.valueOf(depth));
final String ID = rsIDs.isEmpty() ? VCFConstants.EMPTY_ID_FIELD : Utils.join(",", rsIDs);
final VariantContextBuilder builder = new VariantContextBuilder().source(name).id(ID);
builder.loc(loc.getContig(), loc.getStart(), loc.getStop());
builder.alleles(alleles);
builder.genotypes(genotypes);
builder.log10PError(log10PError);
builder.filters(filters).attributes(mergeInfoWithMaxAC ? attributesWithMaxAC : attributes);
// Trim the padded bases of all alleles if necessary
VariantContext merged = createVariantContextWithTrimmedAlleles(builder.make());
if (printMessages && remapped) System.out.printf("Remapped => %s%n", merged);
return merged;
}
public static VariantContext createVariantContextWithPaddedAlleles(
VariantContext inputVC, boolean refBaseShouldBeAppliedToEndOfAlleles) {
// see if we need to pad common reference base from all alleles
boolean padVC;
// We need to pad a VC with a common base if the length of the reference allele is less than the
// length of the VariantContext.
// This happens because the position of e.g. an indel is always one before the actual event (as
// per VCF convention).
long locLength = (inputVC.getEnd() - inputVC.getStart()) + 1;
if (inputVC.hasSymbolicAlleles()) padVC = true;
else if (inputVC.getReference().length() == locLength) padVC = false;
else if (inputVC.getReference().length() == locLength - 1) padVC = true;
else
throw new IllegalArgumentException(
"Badly formed variant context at location "
+ String.valueOf(inputVC.getStart())
+ " in contig "
+ inputVC.getChr()
+ ". Reference length must be at most one base shorter than location size");
// nothing to do if we don't need to pad bases
if (padVC) {
if (!inputVC.hasReferenceBaseForIndel())
throw new ReviewedStingException(
"Badly formed variant context at location "
+ inputVC.getChr()
+ ":"
+ inputVC.getStart()
+ "; no padded reference base is available.");
Byte refByte = inputVC.getReferenceBaseForIndel();
List<Allele> alleles = new ArrayList<Allele>();
for (Allele a : inputVC.getAlleles()) {
// get bases for current allele and create a new one with trimmed bases
if (a.isSymbolic()) {
alleles.add(a);
} else {
String newBases;
if (refBaseShouldBeAppliedToEndOfAlleles)
newBases = a.getBaseString() + new String(new byte[] {refByte});
else newBases = new String(new byte[] {refByte}) + a.getBaseString();
alleles.add(Allele.create(newBases, a.isReference()));
}
}
// now we can recreate new genotypes with trimmed alleles
GenotypesContext genotypes = GenotypesContext.create(inputVC.getNSamples());
for (final Genotype g : inputVC.getGenotypes()) {
List<Allele> inAlleles = g.getAlleles();
List<Allele> newGenotypeAlleles = new ArrayList<Allele>(g.getAlleles().size());
for (Allele a : inAlleles) {
if (a.isCalled()) {
if (a.isSymbolic()) {
newGenotypeAlleles.add(a);
} else {
String newBases;
if (refBaseShouldBeAppliedToEndOfAlleles)
newBases = a.getBaseString() + new String(new byte[] {refByte});
else newBases = new String(new byte[] {refByte}) + a.getBaseString();
newGenotypeAlleles.add(Allele.create(newBases, a.isReference()));
}
} else {
// add no-call allele
newGenotypeAlleles.add(Allele.NO_CALL);
}
}
genotypes.add(
new Genotype(
g.getSampleName(),
newGenotypeAlleles,
g.getLog10PError(),
g.getFilters(),
g.getAttributes(),
g.isPhased()));
}
return new VariantContextBuilder(inputVC).alleles(alleles).genotypes(genotypes).make();
} else return inputVC;
}
public boolean hasGenotypes(Collection<String> sampleNames) {
return genotypes.containsSamples(sampleNames);
}
public Iterable<Genotype> getGenotypesOrderedByName() {
return genotypes.iterateInSampleNameOrder();
}
@Test
public void testFixReverseComplementedGenotypes() {
final Allele refA = Allele.create("A", true);
final Allele altC = Allele.create("C", false);
final GenotypesContext originalGenotypes = GenotypesContext.create(3);
originalGenotypes.add(new GenotypeBuilder("homref").alleles(Arrays.asList(refA, refA)).make());
originalGenotypes.add(new GenotypeBuilder("het").alleles(Arrays.asList(refA, altC)).make());
originalGenotypes.add(new GenotypeBuilder("homvar").alleles(Arrays.asList(altC, altC)).make());
final Allele refT = Allele.create("T", true);
final Allele altG = Allele.create("G", false);
final GenotypesContext expectedGenotypes = GenotypesContext.create(3);
expectedGenotypes.add(new GenotypeBuilder("homref").alleles(Arrays.asList(refT, refT)).make());
expectedGenotypes.add(new GenotypeBuilder("het").alleles(Arrays.asList(refT, altG)).make());
expectedGenotypes.add(new GenotypeBuilder("homvar").alleles(Arrays.asList(altG, altG)).make());
final Map<Allele, Allele> reverseComplementAlleleMap = new HashMap<Allele, Allele>(2);
reverseComplementAlleleMap.put(refA, refT);
reverseComplementAlleleMap.put(altC, altG);
final GenotypesContext actualGenotypes =
LiftoverVcf.fixGenotypes(originalGenotypes, reverseComplementAlleleMap);
for (final String sample : Arrays.asList("homref", "het", "homvar")) {
final List<Allele> expected = expectedGenotypes.get(sample).getAlleles();
final List<Allele> actual = actualGenotypes.get(sample).getAlleles();
Assert.assertEquals(expected.get(0), actual.get(0));
Assert.assertEquals(expected.get(1), actual.get(1));
}
}
public Genotype getGenotype(int ith) {
return genotypes.get(ith);
}
/**
* the actual constructor. Private access only
*
* @param source source
* @param contig the contig
* @param start the start base (one based)
* @param stop the stop reference base (one based)
* @param alleles alleles
* @param genotypes genotypes map
* @param log10PError qual
* @param filters filters: use null for unfiltered and empty set for passes filters
* @param attributes attributes
* @param referenceBaseForIndel padded reference base
* @param validationToPerform set of validation steps to take
*/
protected VariantContext(
String source,
String ID,
String contig,
long start,
long stop,
Collection<Allele> alleles,
GenotypesContext genotypes,
double log10PError,
Set<String> filters,
Map<String, Object> attributes,
Byte referenceBaseForIndel,
EnumSet<Validation> validationToPerform) {
if (contig == null) {
throw new IllegalArgumentException("Contig cannot be null");
}
this.contig = contig;
this.start = start;
this.stop = stop;
// intern for efficiency. equals calls will generate NPE if ID is inappropriately passed in as
// null
if (ID == null || ID.equals(""))
throw new IllegalArgumentException("ID field cannot be the null or the empty string");
this.ID = ID.equals(VCFConstants.EMPTY_ID_FIELD) ? VCFConstants.EMPTY_ID_FIELD : ID;
this.commonInfo = new CommonInfo(source, log10PError, filters, attributes);
REFERENCE_BASE_FOR_INDEL = referenceBaseForIndel;
// todo -- remove me when this check is no longer necessary
if (this.commonInfo.hasAttribute(ID_KEY))
throw new IllegalArgumentException(
"Trying to create a VariantContext with a ID key. Please use provided constructor argument ID");
if (alleles == null) {
throw new IllegalArgumentException("Alleles cannot be null");
}
// we need to make this a LinkedHashSet in case the user prefers a given ordering of alleles
this.alleles = makeAlleles(alleles);
if (genotypes == null || genotypes == NO_GENOTYPES) {
this.genotypes = NO_GENOTYPES;
} else {
this.genotypes = genotypes.immutable();
}
// cache the REF and ALT alleles
int nAlleles = alleles.size();
for (Allele a : alleles) {
if (a.isReference()) {
REF = a;
} else if (nAlleles == 2) { // only cache ALT when biallelic
ALT = a;
}
}
if (!validationToPerform.isEmpty()) {
validate(validationToPerform);
}
}
public void writeBeagleOutput(
VariantContext preferredVC, VariantContext otherVC, boolean isValidationSite, double prior) {
GenomeLoc currentLoc =
VariantContextUtils.getLocation(getToolkit().getGenomeLocParser(), preferredVC);
StringBuffer beagleOut = new StringBuffer();
String marker = String.format("%s:%d ", currentLoc.getContig(), currentLoc.getStart());
beagleOut.append(marker);
if (markers != null)
markers.append(marker).append("\t").append(Integer.toString(markerCounter++)).append("\t");
for (Allele allele : preferredVC.getAlleles()) {
String bglPrintString;
if (allele.isNoCall() || allele.isNull()) bglPrintString = "-";
else bglPrintString = allele.getBaseString(); // get rid of * in case of reference allele
beagleOut.append(String.format("%s ", bglPrintString));
if (markers != null) markers.append(bglPrintString).append("\t");
}
if (markers != null) markers.append("\n");
GenotypesContext preferredGenotypes = preferredVC.getGenotypes();
GenotypesContext otherGenotypes = goodSite(otherVC) ? otherVC.getGenotypes() : null;
for (String sample : samples) {
boolean isMaleOnChrX = CHECK_IS_MALE_ON_CHR_X && getSample(sample).getGender() == Gender.MALE;
Genotype genotype;
boolean isValidation;
// use sample as key into genotypes structure
if (preferredGenotypes.containsSample(sample)) {
genotype = preferredGenotypes.get(sample);
isValidation = isValidationSite;
} else if (otherGenotypes != null && otherGenotypes.containsSample(sample)) {
genotype = otherGenotypes.get(sample);
isValidation = !isValidationSite;
} else {
// there is magically no genotype for this sample.
throw new StingException(
"Sample "
+ sample
+ " arose with no genotype in variant or validation VCF. This should never happen.");
}
/*
* Use likelihoods if: is validation, prior is negative; or: is not validation, has genotype key
*/
double[] log10Likelihoods = null;
if ((isValidation && prior < 0.0) || genotype.hasLikelihoods()) {
log10Likelihoods = genotype.getLikelihoods().getAsVector();
// see if we need to randomly mask out genotype in this position.
if (GenomeAnalysisEngine.getRandomGenerator().nextDouble() <= insertedNoCallRate) {
// we are masking out this genotype
log10Likelihoods =
isMaleOnChrX ? HAPLOID_FLAT_LOG10_LIKELIHOODS : DIPLOID_FLAT_LOG10_LIKELIHOODS;
}
if (isMaleOnChrX) {
log10Likelihoods[1] = -255; // todo -- warning this is dangerous for multi-allele case
}
}
/** otherwise, use the prior uniformly */
else if (!isValidation && genotype.isCalled() && !genotype.hasLikelihoods()) {
// hack to deal with input VCFs with no genotype likelihoods. Just assume the called
// genotype
// is confident. This is useful for Hapmap and 1KG release VCFs.
double AA = (1.0 - prior) / 2.0;
double AB = (1.0 - prior) / 2.0;
double BB = (1.0 - prior) / 2.0;
if (genotype.isHomRef()) {
AA = prior;
} else if (genotype.isHet()) {
AB = prior;
} else if (genotype.isHomVar()) {
BB = prior;
}
log10Likelihoods = MathUtils.toLog10(new double[] {AA, isMaleOnChrX ? 0.0 : AB, BB});
} else {
log10Likelihoods =
isMaleOnChrX ? HAPLOID_FLAT_LOG10_LIKELIHOODS : DIPLOID_FLAT_LOG10_LIKELIHOODS;
}
writeSampleLikelihoods(beagleOut, preferredVC, log10Likelihoods);
}
beagleWriter.println(beagleOut.toString());
}
static VariantContext reallyMergeIntoMNP(
VariantContext vc1, VariantContext vc2, ReferenceSequenceFile referenceFile) {
int startInter = vc1.getEnd() + 1;
int endInter = vc2.getStart() - 1;
byte[] intermediateBases = null;
if (startInter <= endInter) {
intermediateBases =
referenceFile.getSubsequenceAt(vc1.getChr(), startInter, endInter).getBases();
StringUtil.toUpperCase(intermediateBases);
}
MergedAllelesData mergeData =
new MergedAllelesData(
intermediateBases, vc1, vc2); // ensures that the reference allele is added
GenotypesContext mergedGenotypes = GenotypesContext.create();
for (final Genotype gt1 : vc1.getGenotypes()) {
Genotype gt2 = vc2.getGenotype(gt1.getSampleName());
List<Allele> site1Alleles = gt1.getAlleles();
List<Allele> site2Alleles = gt2.getAlleles();
List<Allele> mergedAllelesForSample = new LinkedList<Allele>();
/* NOTE: Since merged alleles are added to mergedAllelesForSample in the SAME order as in the input VC records,
we preserve phase information (if any) relative to whatever precedes vc1:
*/
Iterator<Allele> all2It = site2Alleles.iterator();
for (Allele all1 : site1Alleles) {
Allele all2 = all2It.next(); // this is OK, since allSamplesAreMergeable()
Allele mergedAllele = mergeData.ensureMergedAllele(all1, all2);
mergedAllelesForSample.add(mergedAllele);
}
double mergedGQ = Math.max(gt1.getLog10PError(), gt2.getLog10PError());
Set<String> mergedGtFilters =
new HashSet<
String>(); // Since gt1 and gt2 were unfiltered, the Genotype remains unfiltered
Map<String, Object> mergedGtAttribs = new HashMap<String, Object>();
PhaseAndQuality phaseQual = calcPhaseForMergedGenotypes(gt1, gt2);
if (phaseQual.PQ != null) mergedGtAttribs.put(ReadBackedPhasingWalker.PQ_KEY, phaseQual.PQ);
Genotype mergedGt =
new Genotype(
gt1.getSampleName(),
mergedAllelesForSample,
mergedGQ,
mergedGtFilters,
mergedGtAttribs,
phaseQual.isPhased);
mergedGenotypes.add(mergedGt);
}
String mergedName = mergeVariantContextNames(vc1.getSource(), vc2.getSource());
double mergedLog10PError = Math.min(vc1.getLog10PError(), vc2.getLog10PError());
Set<String> mergedFilters =
new HashSet<
String>(); // Since vc1 and vc2 were unfiltered, the merged record remains unfiltered
Map<String, Object> mergedAttribs = mergeVariantContextAttributes(vc1, vc2);
// ids
List<String> mergedIDs = new ArrayList<String>();
if (vc1.hasID()) mergedIDs.add(vc1.getID());
if (vc2.hasID()) mergedIDs.add(vc2.getID());
String mergedID =
mergedIDs.isEmpty()
? VCFConstants.EMPTY_ID_FIELD
: Utils.join(VCFConstants.ID_FIELD_SEPARATOR, mergedIDs);
VariantContextBuilder mergedBuilder =
new VariantContextBuilder(
mergedName,
vc1.getChr(),
vc1.getStart(),
vc2.getEnd(),
mergeData.getAllMergedAlleles())
.id(mergedID)
.genotypes(mergedGenotypes)
.log10PError(mergedLog10PError)
.filters(mergedFilters)
.attributes(mergedAttribs);
VariantContextUtils.calculateChromosomeCounts(mergedBuilder, true);
return mergedBuilder.make();
}
