private static List<String> calcVCFGenotypeKeys(VariantContext vc) {
Set<String> keys = new HashSet<String>();
boolean sawGoodGT = false;
boolean sawGoodQual = false;
boolean sawGenotypeFilter = false;
for (Genotype g : vc.getGenotypes().values()) {
keys.addAll(g.getAttributes().keySet());
if (g.isAvailable()) sawGoodGT = true;
if (g.hasNegLog10PError()) sawGoodQual = true;
if (g.isFiltered() && g.isCalled()) sawGenotypeFilter = true;
}
if (sawGoodQual) keys.add(VCFConstants.GENOTYPE_QUALITY_KEY);
if (sawGenotypeFilter) keys.add(VCFConstants.GENOTYPE_FILTER_KEY);
List<String> sortedList = ParsingUtils.sortList(new ArrayList<String>(keys));
// make sure the GT is first
if (sawGoodGT) {
List<String> newList = new ArrayList<String>(sortedList.size() + 1);
newList.add(VCFConstants.GENOTYPE_KEY);
newList.addAll(sortedList);
sortedList = newList;
}
return sortedList;
}
/**
* add the genotype data
*
* @param vc the variant context
* @param genotypeFormatKeys Genotype formatting string
* @param alleleMap alleles for this context
* @throws IOException for writer
*/
private void addGenotypeData(
VariantContext vc, Map<Allele, String> alleleMap, List<String> genotypeFormatKeys)
throws IOException {
for (String sample : mHeader.getGenotypeSamples()) {
mWriter.write(VCFConstants.FIELD_SEPARATOR);
Genotype g = vc.getGenotype(sample);
if (g == null) {
// TODO -- The VariantContext needs to know what the general ploidy is of the samples
// TODO -- We shouldn't be assuming diploid genotypes here!
mWriter.write(VCFConstants.EMPTY_GENOTYPE);
continue;
}
List<String> attrs = new ArrayList<String>(genotypeFormatKeys.size());
for (String key : genotypeFormatKeys) {
if (key.equals(VCFConstants.GENOTYPE_KEY)) {
if (!g.isAvailable()) {
throw new ReviewedStingException(
"GTs cannot be missing for some samples if they are available for others in the record");
}
writeAllele(g.getAllele(0), alleleMap);
for (int i = 1; i < g.getPloidy(); i++) {
mWriter.write(g.isPhased() ? VCFConstants.PHASED : VCFConstants.UNPHASED);
writeAllele(g.getAllele(i), alleleMap);
}
continue;
}
Object val = g.hasAttribute(key) ? g.getAttribute(key) : VCFConstants.MISSING_VALUE_v4;
// some exceptions
if (key.equals(VCFConstants.GENOTYPE_QUALITY_KEY)) {
if (Math.abs(g.getNegLog10PError() - Genotype.NO_NEG_LOG_10PERROR) < 1e-6)
val = VCFConstants.MISSING_VALUE_v4;
else {
val = getQualValue(Math.min(g.getPhredScaledQual(), VCFConstants.MAX_GENOTYPE_QUAL));
}
} else if (key.equals(VCFConstants.GENOTYPE_FILTER_KEY)) {
val =
g.isFiltered()
? ParsingUtils.join(";", ParsingUtils.sortList(g.getFilters()))
: (g.filtersWereApplied()
? VCFConstants.PASSES_FILTERS_v4
: VCFConstants.UNFILTERED);
}
VCFFormatHeaderLine metaData = mHeader.getFormatHeaderLine(key);
if (metaData != null) {
int numInFormatField = metaData.getCount(vc.getAlternateAlleles().size());
if (numInFormatField > 1 && val.equals(VCFConstants.MISSING_VALUE_v4)) {
// If we have a missing field but multiple values are expected, we need to construct a
// new string with all fields.
// For example, if Number=2, the string has to be ".,."
StringBuilder sb = new StringBuilder(VCFConstants.MISSING_VALUE_v4);
for (int i = 1; i < numInFormatField; i++) {
sb.append(",");
sb.append(VCFConstants.MISSING_VALUE_v4);
}
val = sb.toString();
}
}
// assume that if key is absent, then the given string encoding suffices
String outputValue = formatVCFField(val);
if (outputValue != null) attrs.add(outputValue);
}
// strip off trailing missing values
for (int i = attrs.size() - 1; i >= 0; i--) {
if (isMissingValue(attrs.get(i))) attrs.remove(i);
else break;
}
for (int i = 0; i < attrs.size(); i++) {
if (i > 0 || genotypeFormatKeys.contains(VCFConstants.GENOTYPE_KEY))
mWriter.write(VCFConstants.GENOTYPE_FIELD_SEPARATOR);
mWriter.write(attrs.get(i));
}
}
}