@Override
public void initialize(
AnnotatorCompatible walker, GenomeAnalysisEngine toolkit, Set<VCFHeaderLine> headerLines) {
// Make sure that we actually have a valid SnpEff rod binding (just in case the user specified
// -A SnpEff
// without providing a SnpEff rod via --snpEffFile):
if (!isValidRodBinding(walker.getSnpEffRodBinding())) {
canAnnotate = false;
return;
}
RodBinding<VariantContext> snpEffRodBinding = walker.getSnpEffRodBinding();
// Make sure that the SnpEff version number and command-line header lines are present in the VCF
// header of
// the SnpEff rod, and that the file was generated by a supported version of SnpEff:
VCFHeader snpEffVCFHeader =
GATKVCFUtils.getVCFHeadersFromRods(toolkit, Arrays.asList(snpEffRodBinding.getName()))
.get(snpEffRodBinding.getName());
VCFHeaderLine snpEffVersionLine =
snpEffVCFHeader.getOtherHeaderLine(SNPEFF_VCF_HEADER_VERSION_LINE_KEY);
VCFHeaderLine snpEffCommandLine =
snpEffVCFHeader.getOtherHeaderLine(SNPEFF_VCF_HEADER_COMMAND_LINE_KEY);
if (!isValidSnpEffVersionAndCommandLine(snpEffVersionLine, snpEffCommandLine)) {
canAnnotate = false;
return;
}
// If everything looks ok, add the SnpEff version number and command-line header lines to the
// header of the VCF output file, changing the key names so that our output file won't be
// mistaken in the future for a SnpEff output file:
headerLines.add(
new VCFHeaderLine(OUTPUT_VCF_HEADER_VERSION_LINE_KEY, snpEffVersionLine.getValue()));
headerLines.add(
new VCFHeaderLine(OUTPUT_VCF_HEADER_COMMAND_LINE_KEY, snpEffCommandLine.getValue()));
// Can only be called from VariantAnnotator
if (!(walker instanceof VariantAnnotator)) {
if (walker != null)
logger.warn(
"Annotation will not be calculated, must be called from VariantAnnotator, not "
+ walker.getClass().getName());
else logger.warn("Annotation will not be calculated, must be called from VariantAnnotator");
canAnnotate = false;
return;
}
}
public void writeHeader(VCFHeader header) {
mHeader = doNotWriteGenotypes ? new VCFHeader(header.getMetaData()) : header;
try {
// the file format field needs to be written first
mWriter.write(
VCFHeader.METADATA_INDICATOR
+ VCFHeaderVersion.VCF4_1.getFormatString()
+ "="
+ VCFHeaderVersion.VCF4_1.getVersionString()
+ "\n");
for (VCFHeaderLine line : mHeader.getMetaData()) {
if (VCFHeaderVersion.isFormatString(line.getKey())) continue;
// are the records filtered (so we know what to put in the FILTER column of passing records)
// ?
if (line instanceof VCFFilterHeaderLine) filtersWereAppliedToContext = true;
mWriter.write(VCFHeader.METADATA_INDICATOR);
mWriter.write(line.toString());
mWriter.write("\n");
}
// write out the column line
mWriter.write(VCFHeader.HEADER_INDICATOR);
for (VCFHeader.HEADER_FIELDS field : mHeader.getHeaderFields()) {
mWriter.write(field.toString());
mWriter.write(VCFConstants.FIELD_SEPARATOR);
}
if (mHeader.hasGenotypingData()) {
mWriter.write("FORMAT");
for (String sample : mHeader.getGenotypeSamples()) {
mWriter.write(VCFConstants.FIELD_SEPARATOR);
mWriter.write(sample);
}
}
mWriter.write("\n");
mWriter.flush(); // necessary so that writing to an output stream will work
} catch (IOException e) {
throw new TribbleException("IOException writing the VCF header to " + locationString(), e);
}
}
private boolean isValidSnpEffVersionAndCommandLine(
final VCFHeaderLine snpEffVersionLine, final VCFHeaderLine snpEffCommandLine) {
if (snpEffVersionLine == null
|| snpEffVersionLine.getValue() == null
|| snpEffVersionLine.getValue().trim().length() == 0) {
logger.warn(
String.format(
"Could not find a %s entry in the VCF header for the SnpEff input file, "
+ "and so could not verify that the file was generated by a supported version of SnpEff (%s)",
SNPEFF_VCF_HEADER_VERSION_LINE_KEY, supportedSnpEffVersionsString()));
return false;
}
if (snpEffCommandLine == null
|| snpEffCommandLine.getValue() == null
|| snpEffCommandLine.getValue().trim().length() == 0) {
logger.warn(
String.format(
"Could not find a %s entry in the VCF header for the SnpEff input file, "
+ "which should be added by all supported versions of SnpEff (%s)",
SNPEFF_VCF_HEADER_COMMAND_LINE_KEY, supportedSnpEffVersionsString()));
return false;
}
String snpEffVersionString = snpEffVersionLine.getValue().replaceAll("\"", "").split(" ")[0];
if (!isSupportedSnpEffVersion(snpEffVersionString, snpEffCommandLine.getValue())) {
logger.warn(
String.format(
"The version of SnpEff used to generate the SnpEff input file (%s) "
+ "is not currently supported by the GATK, and was not run in GATK "
+ "compatibility mode. Supported versions are: %s",
snpEffVersionString, supportedSnpEffVersionsString()));
return false;
}
return true;
}
