@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;
}
}
@Override
public Map<String, Object> annotate(
final RefMetaDataTracker tracker,
final AnnotatorCompatible walker,
final ReferenceContext ref,
final Map<String, AlignmentContext> stratifiedContexts,
final VariantContext vc,
final Map<String, PerReadAlleleLikelihoodMap> stratifiedPerReadAlleleLikelihoodMap) {
// Can only call from UnifiedGenotyper
if (!(walker instanceof UnifiedGenotyper)) {
if (!walkerIdentityCheckWarningLogged) {
if (walker != null)
logger.warn(
"Annotation will not be calculated, must be called from UnifiedGenotyper, not "
+ walker.getClass().getName());
else logger.warn("Annotation will not be calculated, must be called from UnifiedGenotyper");
walkerIdentityCheckWarningLogged = true;
}
return null;
}
if (stratifiedContexts.isEmpty()) return null;
// not meaningful when we're at an indel location: deletions that start at location N are by
// definition called at the position N-1, and at position N-1
// there are no informative deletions in the pileup
if (!vc.isSNP()) return null;
int deletions = 0;
int depth = 0;
for (Map.Entry<String, AlignmentContext> sample : stratifiedContexts.entrySet()) {
for (final PileupElement p : sample.getValue().getBasePileup()) {
depth++;
if (p.isDeletion()) deletions++;
}
}
Map<String, Object> map = new HashMap<>();
map.put(
getKeyNames().get(0),
String.format("%.2f", depth == 0 ? 0.0 : (double) deletions / (double) depth));
return map;
}
@Override
public Map<String, Object> annotate(
final RefMetaDataTracker tracker,
final AnnotatorCompatible walker,
final ReferenceContext ref,
final Map<String, AlignmentContext> stratifiedContexts,
final VariantContext vc,
final Map<String, PerReadAlleleLikelihoodMap> stratifiedPerReadAlleleLikelihoodMap) {
// Can not annotate if failed initialization conditions
if (!canAnnotate) return null;
RodBinding<VariantContext> snpEffRodBinding = walker.getSnpEffRodBinding();
// Get only SnpEff records that start at this locus, not merely span it:
List<VariantContext> snpEffRecords = tracker.getValues(snpEffRodBinding, ref.getLocus());
// Within this set, look for a SnpEff record whose ref/alt alleles match the record to annotate.
// If there is more than one such record, we only need to pick the first one, since the
// biological
// effects will be the same across all such records:
VariantContext matchingRecord = getMatchingSnpEffRecord(snpEffRecords, vc);
if (matchingRecord == null) {
return null;
}
// Parse the SnpEff INFO field annotation from the matching record into individual effect
// objects:
List<SnpEffEffect> effects = parseSnpEffRecord(matchingRecord);
if (effects.isEmpty()) {
return null;
}
// Add only annotations for one of the most biologically-significant effects from this set:
SnpEffEffect mostSignificantEffect = getMostSignificantEffect(effects);
return mostSignificantEffect.getAnnotations();
}
/**
* Checks if the input data is appropriate
*
* @param annotation the input genotype annotation key name(s)
* @param walker input walker
* @param map input map for each read, holds underlying alleles represented by an aligned read,
* and corresponding relative likelihood.
* @param g input genotype
* @param warningsLogged array that enforces the warning is logged once for each caller
* @param logger logger specific for each caller
* @return true if the walker is a HaplotypeCaller, the likelihood map is non-null and the
* genotype is non-null and called, false otherwise
* @throws IllegalArgumentException if annotation, walker, g, warningsLogged, or logger are null.
* @throws ReviewedGATKException if the size of warningsLogged is less than 3.
*/
public static boolean isAppropriateInput(
final String annotation,
final AnnotatorCompatible walker,
final PerReadAlleleLikelihoodMap map,
final Genotype g,
final boolean[] warningsLogged,
final Logger logger) {
if (annotation == null) {
throw new IllegalArgumentException("The input annotation cannot be null");
}
if (walker == null) {
throw new IllegalArgumentException("The input walker cannot be null");
}
if (g == null) {
throw new IllegalArgumentException("The input genotype cannot be null");
}
if (warningsLogged == null) {
throw new IllegalArgumentException("The input warnings logged cannot be null");
}
if (logger == null) {
throw new IllegalArgumentException("The input logger cannot be null");
}
if (warningsLogged.length < WARNINGS_LOGGED_SIZE) {
throw new ReviewedGATKException(
"Warnings logged array must have at least "
+ WARNINGS_LOGGED_SIZE
+ " elements, but has "
+ warningsLogged.length);
}
if (!(walker instanceof HaplotypeCaller) && !(walker instanceof MuTect2)) {
if (!warningsLogged[0]) {
logger.warn(
annotation + ANNOTATION_HC_WARN_MSG + ", not " + walker.getClass().getSimpleName());
warningsLogged[0] = true;
}
return false;
}
if (map == null) {
if (!warningsLogged[1]) {
logger.warn(
"Annotation will not be calculated, can only be used with likelihood based annotations in the HaplotypeCaller");
warningsLogged[1] = true;
}
return false;
}
if (!g.isCalled()) {
if (!warningsLogged[2]) {
logger.warn("Annotation will not be calculated, genotype is not called");
warningsLogged[2] = true;
}
return false;
}
return true;
}