use of htsjdk.variant.vcf.VCFEncoder in project jvarkit by lindenb.
the class LumpySort method doWork.
@Override
public int doWork(final List<String> args) {
VariantContextWriter vcw = null;
LineIterator vcfIn = null;
Environment environment = null;
Database variantsDb1 = null;
final List<File> inputs = IOUtil.unrollFiles(args.stream().map(S -> new File(S)).collect(Collectors.toList()), ".vcf", ".vcf.gz");
if (inputs.isEmpty()) {
LOG.error("empty vcf list");
return -1;
}
try {
IOUtil.assertDirectoryIsWritable(this.bdbHomeDir);
final Set<VCFHeaderLine> metaData = new HashSet<>();
final Set<String> sampleNames = new TreeSet<>();
final IntervalTreeMap<Boolean> intervalTreeMapBed;
if (this.bedFile != null) {
intervalTreeMapBed = new IntervalTreeMap<>();
final BedLineCodec bedLineCodec = new BedLineCodec();
final BufferedReader br = IOUtils.openFileForBufferedReading(this.bedFile);
br.lines().map(L -> bedLineCodec.decode(L)).filter(L -> L != null).forEach(B -> intervalTreeMapBed.put(B.toInterval(), true));
br.close();
} else {
intervalTreeMapBed = null;
}
for (int idx = 0; idx < inputs.size(); ++idx) {
final File vcfFile = inputs.get(idx);
LOG.info("Read header " + (idx + 1) + "/" + inputs.size());
final VCFFileReader r = new VCFFileReader(vcfFile, false);
final VCFHeader header = r.getFileHeader();
if (!LumpyConstants.isLumpyHeader(header)) {
LOG.error("doesn't look like a Lumpy-SV vcf header " + vcfFile);
r.close();
return -1;
}
if (!header.hasGenotypingData()) {
LOG.error("No sample in " + vcfFile);
r.close();
return -1;
}
for (final String sampleName : header.getSampleNamesInOrder()) {
if (sampleNames.contains(sampleName)) {
LOG.error("Sample found twice " + sampleName + " in " + vcfFile);
r.close();
return -1;
}
sampleNames.add(sampleName);
}
metaData.addAll(header.getMetaDataInInputOrder().stream().filter(H -> !H.getKey().equals("fileDate")).collect(Collectors.toSet()));
r.close();
}
final VCFInfoHeaderLine nSampleInfoHeaderLine = new VCFInfoHeaderLine("NSAMPLES", 1, VCFHeaderLineType.Integer, "Number of affected samples.");
metaData.add(nSampleInfoHeaderLine);
final VCFFormatHeaderLine chromStartFormatHeaderLine = new VCFFormatHeaderLine("CB", 1, VCFHeaderLineType.Integer, "Original Variant POS");
metaData.add(chromStartFormatHeaderLine);
final VCFFormatHeaderLine chromEndFormatHeaderLine = new VCFFormatHeaderLine("CE", 1, VCFHeaderLineType.Integer, "Original Variant END");
metaData.add(chromEndFormatHeaderLine);
final VCFHeader outHeader = new VCFHeader(metaData, sampleNames);
final VCFHeaderVersion[] versions = VCFHeaderVersion.values();
this.vcfEncoder = new VCFEncoder(outHeader, false, true);
this.vcfCodec.setVCFHeader(outHeader, versions[versions.length - 1]);
/* open BDB env */
final Transaction txn = null;
environment = new Environment(this.bdbHomeDir, new EnvironmentConfig().setAllowCreate(true).setReadOnly(false));
variantsDb1 = environment.openDatabase(txn, "variants1", new DatabaseConfig().setBtreeComparator(KeySorterComparator.class).setAllowCreate(true).setReadOnly(false).setTemporary(true));
long total_variants = 0L;
final LumpyVarBinding lumpVarBinding = new LumpyVarBinding();
final KeySorterBinding keySorterBinding = new KeySorterBinding();
for (int idx = 0; idx < inputs.size(); ++idx) {
final long millisecstart = System.currentTimeMillis();
final File vcfFile = inputs.get(idx);
int nVariant = 0;
final VCFFileReader r = new VCFFileReader(vcfFile, false);
final List<Genotype> missing = new ArrayList<>(sampleNames.size());
for (final String sn : sampleNames) {
if (r.getFileHeader().getSampleNamesInOrder().contains(sn))
continue;
missing.add(GenotypeBuilder.createMissing(sn, 2));
}
final CloseableIterator<VariantContext> iter = r.iterator();
while (iter.hasNext()) {
VariantContext ctx = iter.next();
if (!this.keep_secondary) {
if (ctx.hasAttribute("SECONDARY"))
continue;
}
if (!this.variantFilter.test(ctx))
continue;
if (intervalTreeMapBed != null && !intervalTreeMapBed.containsOverlapping(ctx))
continue;
final List<Genotype> gtList = new ArrayList<>(ctx.getGenotypes());
for (int gi = 0; gi < gtList.size(); gi++) {
Genotype g = gtList.get(gi);
final GenotypeBuilder gb;
if (this.do_genotype && isAvailableGenotype(g)) {
gb = new GenotypeBuilder(g.getSampleName(), ctx.getAlternateAlleles());
gb.attributes(g.getExtendedAttributes());
} else {
gb = new GenotypeBuilder(g);
}
gb.attribute(chromStartFormatHeaderLine.getID(), ctx.getStart());
gb.attribute(chromEndFormatHeaderLine.getID(), ctx.getEnd());
gtList.set(gi, gb.make());
}
gtList.addAll(missing);
ctx = new VariantContextBuilder(ctx).genotypes(gtList).rmAttribute("PRPOS").make();
final LumpyVar lvar = new LumpyVar(ctx, total_variants);
final DatabaseEntry key = new DatabaseEntry();
final DatabaseEntry data = new DatabaseEntry();
lumpVarBinding.objectToEntry(lvar, data);
keySorterBinding.objectToEntry(lvar.getSortKey(), key);
if (variantsDb1.put(txn, key, data) != OperationStatus.SUCCESS) {
r.close();
LOG.error("insertion failed");
return -1;
}
nVariant++;
total_variants++;
}
iter.close();
r.close();
LOG.info("Read " + (idx + 1) + "/" + inputs.size() + " variants of " + vcfFile + " N=" + nVariant + " Total:" + total_variants + " That took: " + Duration.ofMillis(System.currentTimeMillis() - millisecstart));
System.gc();
}
if (intervalTreeMapBed != null)
intervalTreeMapBed.clear();
System.gc();
LOG.info("Writing output");
final List<Allele> ALLELES_NO_CALLS = this.do_genotype ? Collections.singletonList(Allele.NO_CALL) : Arrays.asList(Allele.NO_CALL, Allele.NO_CALL);
final Cursor cursor = variantsDb1.openCursor(txn, null);
vcw = super.openVariantContextWriter(this.outputFile);
vcw.writeHeader(outHeader);
for (; ; ) {
final DatabaseEntry key = new DatabaseEntry();
final DatabaseEntry data = new DatabaseEntry();
OperationStatus status = cursor.getNext(key, data, LockMode.DEFAULT);
if (!status.equals(OperationStatus.SUCCESS))
break;
final LumpyVar first = lumpVarBinding.entryToObject(data);
if (this.do_not_merge_ctx) {
vcw.add(first.ctx);
continue;
}
final KeySorter keySorter1 = keySorterBinding.entryToObject(key);
final List<LumpyVar> buffer = new ArrayList<>();
buffer.add(first);
final DatabaseEntry key2 = new DatabaseEntry();
final DatabaseEntry data2 = new DatabaseEntry();
final Cursor cursor2 = cursor.dup(true);
for (; ; ) {
status = cursor2.getNext(key2, data2, LockMode.DEFAULT);
if (!status.equals(OperationStatus.SUCCESS))
break;
final KeySorter keySorter2 = keySorterBinding.entryToObject(key2);
if (keySorter1.compare1(keySorter2) != 0) {
break;
}
final LumpyVar lv = lumpVarBinding.entryToObject(data2);
if (lv.ctx.getStart() > first.ctx.getEnd()) {
break;
}
if (first.canMerge(lv)) {
buffer.add(lv);
cursor2.delete();
}
}
cursor2.close();
// delete 'first'
cursor.delete();
final int variantStartA = buffer.stream().mapToInt(V -> V.ctx.getStart()).min().getAsInt();
final int variantStartB = (int) buffer.stream().mapToInt(V -> V.ctx.getStart()).average().getAsDouble();
final int variantStartC = buffer.stream().mapToInt(V -> V.ctx.getStart()).max().getAsInt();
final int variantEndA = buffer.stream().mapToInt(V -> V.ctx.getEnd()).min().getAsInt();
final int variantEndB = (int) buffer.stream().mapToInt(V -> V.ctx.getEnd()).average().getAsDouble();
final int variantEndC = buffer.stream().mapToInt(V -> V.ctx.getEnd()).max().getAsInt();
final VariantContextBuilder vcb = new VariantContextBuilder("lumpymerge", first.ctx.getContig(), variantStartB, variantEndB, first.ctx.getAlleles());
vcb.attribute("END", variantEndB);
vcb.attribute("SVTYPE", first.ctx.getAttribute("SVTYPE"));
vcb.attribute("SVLEN", (int) Percentile.median().evaluate(buffer.stream().mapToInt(V -> V.ctx.getEnd() - V.ctx.getStart())));
vcb.attribute("CIPOS", Arrays.asList(variantStartB - variantStartA, variantStartC - variantStartB));
vcb.attribute("CIEND", Arrays.asList(variantEndB - variantEndA, variantEndC - variantEndB));
vcb.attribute("SU", buffer.stream().flatMap(V -> V.ctx.getGenotypes().stream()).mapToInt(G -> G.getAttributeAsInt("SU", 0)).sum());
vcb.attribute("SR", buffer.stream().flatMap(V -> V.ctx.getGenotypes().stream()).mapToInt(G -> G.getAttributeAsInt("SR", 0)).sum());
vcb.attribute("PE", buffer.stream().flatMap(V -> V.ctx.getGenotypes().stream()).mapToInt(G -> G.getAttributeAsInt("PE", 0)).sum());
final Map<String, Genotype> sample2genotype = new HashMap<>(sampleNames.size());
buffer.stream().flatMap(V -> V.ctx.getGenotypes().stream()).filter(G -> isAvailableGenotype(G)).forEach(G -> {
sample2genotype.put(G.getSampleName(), G);
});
vcb.attribute(nSampleInfoHeaderLine.getID(), sample2genotype.size());
for (final String sn : sampleNames) {
if (!sample2genotype.containsKey(sn)) {
sample2genotype.put(sn, new GenotypeBuilder(sn, ALLELES_NO_CALLS).attribute("SU", 0).attribute("SR", 0).attribute("PE", 0).make());
}
}
vcb.genotypes(sample2genotype.values());
vcw.add(vcb.make());
}
cursor.close();
vcw.close();
vcw = null;
variantsDb1.close();
variantsDb1 = null;
environment.close();
environment = null;
return 0;
} catch (final Exception err) {
LOG.error(err);
return -1;
} finally {
CloserUtil.close(vcfIn);
CloserUtil.close(vcw);
CloserUtil.close(variantsDb1);
CloserUtil.close(environment);
}
}
Aggregations