// 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");
}
}
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 void assertEquals(final Genotype actual, final Genotype expected) {
Assert.assertEquals(actual.getSampleName(), expected.getSampleName(), "Genotype names");
Assert.assertEquals(actual.getAlleles(), expected.getAlleles(), "Genotype alleles");
Assert.assertEquals(
actual.getGenotypeString(), expected.getGenotypeString(), "Genotype string");
Assert.assertEquals(actual.getType(), expected.getType(), "Genotype type");
// filters are the same
Assert.assertEquals(actual.getFilters(), expected.getFilters(), "Genotype fields");
Assert.assertEquals(actual.isFiltered(), expected.isFiltered(), "Genotype isFiltered");
// inline attributes
Assert.assertEquals(actual.getDP(), expected.getDP(), "Genotype dp");
Assert.assertTrue(Arrays.equals(actual.getAD(), expected.getAD()));
Assert.assertEquals(actual.getGQ(), expected.getGQ(), "Genotype gq");
Assert.assertEquals(actual.hasPL(), expected.hasPL(), "Genotype hasPL");
Assert.assertEquals(actual.hasAD(), expected.hasAD(), "Genotype hasAD");
Assert.assertEquals(actual.hasGQ(), expected.hasGQ(), "Genotype hasGQ");
Assert.assertEquals(actual.hasDP(), expected.hasDP(), "Genotype hasDP");
Assert.assertEquals(
actual.hasLikelihoods(), expected.hasLikelihoods(), "Genotype haslikelihoods");
Assert.assertEquals(
actual.getLikelihoodsString(),
expected.getLikelihoodsString(),
"Genotype getlikelihoodsString");
Assert.assertEquals(
actual.getLikelihoods(), expected.getLikelihoods(), "Genotype getLikelihoods");
Assert.assertTrue(Arrays.equals(actual.getPL(), expected.getPL()));
Assert.assertEquals(
actual.getPhredScaledQual(), expected.getPhredScaledQual(), "Genotype phredScaledQual");
assertAttributesEquals(actual.getExtendedAttributes(), expected.getExtendedAttributes());
Assert.assertEquals(actual.isPhased(), expected.isPhased(), "Genotype isPhased");
Assert.assertEquals(actual.getPloidy(), expected.getPloidy(), "Genotype getPloidy");
}
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());
}
