/** Uncompress VCF entry having genotypes in "HO,HE,NA" fields */
public VcfEntry uncompressGenotypes() {
// Not compressed? Nothing to do
if (!isCompressedGenotypes()) return this;
// Get 'sparse' matrix entries
String hoStr = getInfo(VCF_INFO_HOMS);
String heStr = getInfo(VCF_INFO_HETS);
String naStr = getInfo(VCF_INFO_NAS);
// Parse 'sparse' entries
List<String> sampleNames = getVcfFileIterator().getVcfHeader().getSampleNames();
if (sampleNames == null)
throw new RuntimeException(
"Cannot find sample names in VCF header. Unable to uncompress genotypes.");
int numSamples = sampleNames.size();
byte gt[] = new byte[numSamples];
parseSparseGt(naStr, gt, -1);
parseSparseGt(heStr, gt, 1);
parseSparseGt(hoStr, gt, 2);
// Remove info fields
if (hoStr != null) rmInfo(VCF_INFO_HOMS);
if (heStr != null) rmInfo(VCF_INFO_HETS);
if (naStr != null) rmInfo(VCF_INFO_NAS);
setFormat("GT");
// Create output string
for (int i = 0; i < gt.length; i++) {
String gtStr;
switch (gt[i]) {
case -1:
gtStr = "./.";
break;
case 0:
gtStr = "0/0";
break;
case 1:
gtStr = "0/1";
break;
case 2:
gtStr = "1/1";
break;
default:
throw new RuntimeException("Unknown code '" + gt[i] + "'");
}
addGenotype(gtStr);
}
return this;
}
/** Parse GENOTPYE entries */
void parseGenotypes() {
if (isCompressedGenotypes()) {
uncompressGenotypes();
} else {
vcfGenotypes = new ArrayList<>();
// No genotype string? => Nothing to do
if (genotypeFieldsStr == null) return;
// Split genotypes and parse them
genotypeFields = genotypeFieldsStr.split("\t");
for (int i = 0; i < genotypeFields.length; i++) {
String gen = genotypeFields[i];
if (gen.equals(VcfFileIterator.MISSING)) gen = "";
addGenotype(gen);
}
}
}
