Skip to content

Commit

Permalink
Merge pull request #2094 from broadinstitute/cn_vqsr
Browse files Browse the repository at this point in the history 
VariantRecalibrator and ApplyVQSR port part 1 (no integration tests).
  • Loading branch information
@cmnbroad
cmnbroad authored Sep 14, 2016
2 parents e500887 + 082305e commit 4ec1b85
Show file tree
Hide file tree
Showing 37 changed files with 4,141 additions and 113 deletions.
10 changes: 10 additions & 0 deletions src/main/java/org/broadinstitute/hellbender/tools/walkers/annotator/AnnotationUtils.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -20,4 +20,14 @@ public static String encodeValueList(final List<Double> valueList, final String
}
return StringUtils.join(outputList, ",");
}

/**
* Helper function to convert a List of Strings to a comma-separated String
* @param stringList the ArrayList with String data
* @return a comma-separated String
*/
public static String encodeStringList( final List<String> stringList) {
return StringUtils.join(stringList, ",");
}

}
91 changes: 9 additions & 82 deletions src/main/java/org/broadinstitute/hellbender/tools/walkers/variantutils/SelectVariants.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,7 @@
import org.broadinstitute.hellbender.utils.samples.PedigreeValidationType;
import org.broadinstitute.hellbender.utils.samples.SampleDB;
import org.broadinstitute.hellbender.utils.samples.SampleDBBuilder;
import org.broadinstitute.hellbender.utils.samples.SampleUtils;
import org.broadinstitute.hellbender.utils.Utils;
import org.broadinstitute.hellbender.utils.text.XReadLines;
import org.broadinstitute.hellbender.utils.variant.*;
Expand Down Expand Up @@ -487,17 +488,20 @@ public void onTraversalStart() {
IDsToKeep = getIDsFromFile(rsIDFile);
IDsToRemove = getIDsFromFile(XLrsIDFile);

//TODO: this should be refactored/consolidated as part of
// https://github.com/broadinstitute/gatk/issues/121 and
// https://github.com/broadinstitute/gatk/issues/1116
Set<VCFHeaderLine> actualLines = null;
SAMSequenceDictionary sequenceDictionary = null;
if (hasReference()) {
File refFile = referenceArguments.getReferenceFile();
sequenceDictionary= this.getReferenceDictionary();
actualLines = withUpdatedContigsAsLines(headerLines, refFile, sequenceDictionary, suppressReferencePath);
actualLines = VcfUtils.updateHeaderContigLines(headerLines, refFile, sequenceDictionary, suppressReferencePath);
}
else {
sequenceDictionary = getHeaderForVariants().getSequenceDictionary();
if (null != sequenceDictionary) {
actualLines = withUpdatedContigsAsLines(headerLines, null, sequenceDictionary, suppressReferencePath);
actualLines = VcfUtils.updateHeaderContigLines(headerLines, null, sequenceDictionary, suppressReferencePath);
}
else {
actualLines = headerLines;
Expand Down Expand Up @@ -646,7 +650,7 @@ protected VariantFilter makeVariantFilter() {
*/
private SortedSet<String> createSampleNameInclusionList(Map<String, VCFHeader> vcfHeaders) {
final SortedSet<String> vcfSamples = VcfUtils.getSortedSampleSet(vcfHeaders, GATKVariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE);
final Collection<String> samplesFromFile = getSamplesFromFiles(sampleFiles);
final Collection<String> samplesFromFile = SampleUtils.getSamplesFromFiles(sampleFiles);
final Collection<String> samplesFromExpressions = matchSamplesExpressions(vcfSamples, sampleExpressions);

// first, check overlap between requested and present samples
Expand Down Expand Up @@ -684,7 +688,7 @@ private SortedSet<String> createSampleNameInclusionList(Map<String, VCFHeader> v
}

// Exclude samples take precedence over include - remove any excluded samples
final Collection<String> XLsamplesFromFile = getSamplesFromFiles(XLsampleFiles);
final Collection<String> XLsamplesFromFile = SampleUtils.getSamplesFromFiles(XLsampleFiles);
final Collection<String> XLsamplesFromExpressions = matchSamplesExpressions(vcfSamples, XLsampleExpressions);
samples.removeAll(XLsamplesFromFile);
samples.removeAll(XLsampleNames);
Expand Down Expand Up @@ -732,7 +736,7 @@ private Set<VariantContext.Type> createSampleTypeInclusionList() {
private Set<VCFHeaderLine> createVCFHeaderLineList(Map<String, VCFHeader> vcfHeaders) {

final Set<VCFHeaderLine> headerLines = VCFUtils.smartMergeHeaders(vcfHeaders.values(), true);
headerLines.add(new VCFHeaderLine("source", "SelectVariants"));
headerLines.add(new VCFHeaderLine("source", this.getClass().getSimpleName()));

if (keepOriginalChrCounts) {
headerLines.add(GATKVCFHeaderLines.getInfoLine(GATKVCFConstants.ORIGINAL_AC_KEY));
Expand All @@ -749,63 +753,6 @@ private Set<VCFHeaderLine> createVCFHeaderLineList(Map<String, VCFHeader> vcfHea
return headerLines;
}

private static Set<VCFHeaderLine> withUpdatedContigsAsLines(
final Set<VCFHeaderLine> oldLines,
final File referenceFile,
final SAMSequenceDictionary refDict,
final boolean referenceNameOnly) {
final Set<VCFHeaderLine> lines = new LinkedHashSet<>(oldLines.size());

for (final VCFHeaderLine line : oldLines) {
if (line instanceof VCFContigHeaderLine) {
continue; // skip old contig lines
}
if (line.getKey().equals(VCFHeader.REFERENCE_KEY)) {
continue; // skip the old reference key
}
lines.add(line);
}

lines.addAll(makeContigHeaderLines(refDict, referenceFile).stream().collect(Collectors.toList()));

if (referenceFile != null) {
final String referenceValue;
if (referenceNameOnly) {
final int extensionStart = referenceFile.getName().lastIndexOf(".");
referenceValue = extensionStart == -1 ? referenceFile.getName() : referenceFile.getName().substring(0, extensionStart);
}
else {
referenceValue = "file://" + referenceFile.getAbsolutePath();
}
lines.add(new VCFHeaderLine(VCFHeader.REFERENCE_KEY, referenceValue));
}
return lines;
}

/**
* Create VCFHeaderLines for each refDict entry, and optionally the assembly if referenceFile != null
* @param refDict reference dictionary
* @param referenceFile for assembly name. May be null
* @return list of vcf contig header lines
*/
private static List<VCFContigHeaderLine> makeContigHeaderLines(final SAMSequenceDictionary refDict,
final File referenceFile) {
final List<VCFContigHeaderLine> lines = new ArrayList<>();
final String assembly = referenceFile != null ? referenceFile.getName() : null;
lines.addAll(refDict.getSequences().stream().map(contig -> makeContigHeaderLine(contig, assembly)).collect(Collectors.toList()));
return lines;
}

private static VCFContigHeaderLine makeContigHeaderLine(final SAMSequenceRecord contig, final String assembly) {
final Map<String, String> map = new LinkedHashMap<>(3);
map.put("ID", contig.getSequenceName());
map.put("length", String.valueOf(contig.getSequenceLength()));
if (assembly != null) {
map.put("assembly", assembly);
}
return new VCFContigHeaderLine(map, contig.getSequenceIndex());
}

/**
* Entry-point function to initialize the samples database from input data
*/
Expand All @@ -831,26 +778,6 @@ private static Collection<String> matchSamplesExpressions (Collection<String> or
return samples;
}

/**
* Given a list of files with sample names it reads all files and creates a list of unique samples from all these files.
* @param files list of files with sample names in
* @return a collection of unique samples from all files
*/
private static Collection<String> getSamplesFromFiles (Collection<File> files) {
final Set<String> samplesFromFiles = new LinkedHashSet<>();
if (files != null) {
for (final File file : files) {
try (XReadLines reader = new XReadLines(file)) {
List<String> lines = reader.readLines();
samplesFromFiles.addAll(lines.stream().collect(Collectors.toList()));
} catch (IOException e) {
throw new UserException.CouldNotReadInputFile(file, e);
}
}
}
return samplesFromFiles;
}

/**
* Get IDs from a file
*
Expand Down
Loading

0 comments on commit 4ec1b85

Please sign in to comment.