Examples with Sample com.github.lindenb.jvarkit.pedigree.Sample used on opensource projects

Example 6 with Sample

use of com.github.lindenb.jvarkit.pedigree.Sample in project jvarkit by lindenb.

the class VcfFilterNotInPedigree method doVcfToVcf.

@Override
protected int doVcfToVcf(String inputName, VCFIterator iterin, VariantContextWriter out) {
    final int IGNORE_SINGLETON = -1;
    final Predicate<Genotype> acceptFilteredGenotype = G -> G != null && (this.ignoreFilteredGT == false || !G.isFiltered());
    final Pedigree pedigree;
    final VCFHeader header = iterin.getHeader();
    try {
        pedigree = new PedigreeParser().parse(this.pedigreeFile);
    } catch (final IOException err) {
        LOG.error("Cannot read pedigree in file: " + this.pedigreeFile, err);
        return -1;
    }
    if (pedigree.isEmpty()) {
        throw new JvarkitException.PedigreeError("No pedigree found in header. use VcfInjectPedigree to add it");
    }
    pedigree.checkUniqIds();
    final Set<String> samplesNames = new HashSet<>(header.getSampleNamesInOrder());
    final Set<Sample> individuals = new HashSet<>(pedigree.getSamples());
    final Iterator<Sample> iter = individuals.iterator();
    while (iter.hasNext()) {
        final Sample person = iter.next();
        if (!(samplesNames.contains(person.getId()))) {
            LOG.warn("Ignoring " + person + " because not in VCF header.");
            iter.remove();
        }
    }
    final VCFFilterHeaderLine filter = new VCFFilterHeaderLine(this.filterName, "Will be set for variant where the only genotypes non-homref are NOT in the pedigree ");
    final VCFFilterHeaderLine singletonFilter = new VCFFilterHeaderLine(this.singletonfilterName, "The ALT allele is found in less or equals than " + this.singleton + " individuals in the cases/controls");
    final VCFHeader h2 = new VCFHeader(header);
    JVarkitVersion.getInstance().addMetaData(this, h2);
    if (this.singleton != IGNORE_SINGLETON) {
        h2.addMetaDataLine(singletonFilter);
    }
    out.writeHeader(h2);
    while (iterin.hasNext()) {
        final VariantContext ctx = iterin.next();
        final boolean in_pedigree = individuals.stream().map(P -> ctx.getGenotype(P.getId())).filter(acceptFilteredGenotype).anyMatch(g -> (!(g == null || !g.isCalled() || !g.isAvailable() || g.isNoCall() || g.isHomRef())));
        if (!in_pedigree) {
            if (this.dicardVariant)
                continue;
            final VariantContextBuilder vcb = new VariantContextBuilder(ctx);
            vcb.filter(filter.getID());
            out.add(vcb.make());
        } else {
            boolean is_singleton;
            if (this.singleton != IGNORE_SINGLETON) {
                is_singleton = true;
                for (final Allele alt : ctx.getAlternateAlleles()) {
                    if (individuals.stream().map(P -> ctx.getGenotype(P.getId())).filter(acceptFilteredGenotype).filter(g -> g != null && g.isCalled() && g.getAlleles().contains(alt)).count() > this.singleton) {
                        is_singleton = false;
                        break;
                    }
                }
            } else {
                is_singleton = false;
            }
            if (is_singleton) {
                if (this.dicardVariant)
                    continue;
                final VariantContextBuilder vcb = new VariantContextBuilder(ctx);
                vcb.filter(singletonFilter.getID());
                out.add(vcb.make());
            } else {
                out.add(ctx);
            }
        }
    }
    return 0;
}

Example 7 with Sample

use of com.github.lindenb.jvarkit.pedigree.Sample in project jvarkit by lindenb.

the class VcfToSvg method doWork.

@Override
public int doWork(final List<String> args) {
    VCFReader r = null;
    OutputStream outputStream = null;
    XMLStreamWriter w = null;
    PrintWriter manifestW = null;
    ArchiveFactory archiveFactory = null;
    try {
        r = VCFReaderFactory.makeDefault().open(Paths.get(oneAndOnlyOneFile(args)), true);
        final VCFHeader header = r.getHeader();
        final List<String> samples = new ArrayList<>(header.getSampleNamesInOrder());
        final SAMSequenceDictionary dict = SequenceDictionaryUtils.extractRequired(header);
        intervalListProvider.dictionary(dict);
        /* read gtf if any */
        final IntervalTreeMap<Gene> geneMap = new IntervalTreeMap<>();
        if (this.gtfPath != null) {
            try (GtfReader gtfReader = new GtfReader(this.gtfPath)) {
                gtfReader.setContigNameConverter(ContigNameConverter.fromOneDictionary(dict));
                gtfReader.getAllGenes().stream().filter(G -> !this.removeNonCoding || G.getTranscripts().stream().anyMatch(T -> T.isCoding())).forEach(G -> geneMap.put(new Interval(G), G));
            }
        }
        archiveFactory = ArchiveFactory.open(this.outputPath);
        if (manifestFile != null) {
            manifestW = IOUtils.openPathForPrintWriter(this.manifestFile);
        } else {
            manifestW = new PrintWriter(new NullOuputStream());
        }
        final Pedigree pedigree;
        if (this.pedPath == null) {
            pedigree = PedigreeParser.empty();
        } else {
            pedigree = new PedigreeParser().parse(this.pedPath);
        }
        final Path tmpSvg = Files.createTempFile("vcf.", ".svg");
        final XMLOutputFactory xof = XMLOutputFactory.newInstance();
        for (final Locatable interval : intervalListProvider.dictionary(dict).stream().collect(Collectors.toList())) {
            final List<VariantContext> variants = r.query(interval).stream().filter(V -> this.variantFILTEREDOpacity > 0 || !V.isFiltered()).filter(V -> this.variantIndelOpacity > 0 || !V.isIndel()).collect(Collectors.toCollection(ArrayList::new));
            if (variants.isEmpty())
                continue;
            final List<Transcript> transcripts = geneMap.getOverlapping(interval).stream().flatMap(G -> G.getTranscripts().stream()).filter(T -> !this.removeNonCoding || T.isCoding()).collect(Collectors.toList());
            variants.removeIf(V -> this.gtfPath != null && this.variantsInExonOnly && transcripts.stream().flatMap(T -> T.getExons().stream()).noneMatch(EX -> EX.overlaps(V)));
            if (variants.isEmpty())
                continue;
            final String geneId = transcripts.stream().map(T -> T.getGene().getId()).collect(Collectors.toSet()).stream().collect(HtsCollectors.oneAndOnlyOne()).orElse(null);
            final String geneName = transcripts.stream().map(T -> T.getGene().getGeneName()).collect(Collectors.toSet()).stream().collect(HtsCollectors.oneAndOnlyOne()).orElse(null);
            outputStream = IOUtils.openPathForWriting(tmpSvg);
            w = xof.createXMLStreamWriter(outputStream);
            double featureHeight = 10;
            double TRANSCRIPT_HEIGHT = featureHeight;
            final int all_genotypes_width = variants.size() * this.genotype_width;
            final int drawinAreaWidth = Math.max(all_genotypes_width, 1000);
            final int interline_weight = 6;
            final int margin_top = 10;
            final int margin_bottom = 10;
            final int margin_right = 100;
            final int margin_left = 100;
            w.writeStartDocument("UTF-8", "1.0");
            w.writeStartElement("svg");
            w.writeDefaultNamespace(SVG.NS);
            w.writeNamespace("xlink", XLINK.NS);
            w.writeAttribute("version", "1.1");
            w.writeAttribute("width", String.valueOf(margin_right + margin_right + drawinAreaWidth));
            w.writeAttribute("height", String.valueOf(margin_top + margin_bottom + transcripts.size() * TRANSCRIPT_HEIGHT + interline_weight * featureHeight + samples.size() * this.genotype_width));
            title(w, interval.getContig() + ":" + interval.getStart() + "-" + interval.getEnd());
            w.writeStartElement("desc");
            w.writeCharacters("generated with " + getProgramName() + "\n" + "Author: Pierre Lindenbaum PhD. @yokofakun .");
            w.writeEndElement();
            // defs
            w.writeStartElement("defs");
            // genotypes
            w.writeStartElement("g");
            // 
            w.writeAttribute("id", "g_" + GenotypeType.HOM_REF);
            w.writeEmptyElement("rect");
            w.writeAttribute("style", "fill:lime;stroke;none;");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", "0");
            w.writeAttribute("width", String.valueOf(this.genotype_width));
            w.writeAttribute("height", String.valueOf(this.genotype_width));
            w.writeEndElement();
            w.writeStartElement("g");
            // 
            w.writeAttribute("id", "g_" + GenotypeType.NO_CALL);
            w.writeEmptyElement("rect");
            w.writeAttribute("style", "fill:silver;stroke;gray;");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", "0");
            w.writeAttribute("width", String.valueOf(this.genotype_width));
            w.writeAttribute("height", String.valueOf(this.genotype_width));
            w.writeEndElement();
            w.writeStartElement("g");
            // 
            w.writeAttribute("id", "g_" + GenotypeType.HOM_VAR);
            w.writeEmptyElement("rect");
            w.writeAttribute("style", "fill:crimson;stroke;none;");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", "0");
            w.writeAttribute("width", String.valueOf(this.genotype_width));
            w.writeAttribute("height", String.valueOf(this.genotype_width));
            w.writeEndElement();
            w.writeStartElement("g");
            // 
            w.writeAttribute("id", "g_" + GenotypeType.MIXED);
            w.writeEmptyElement("rect");
            w.writeAttribute("style", "fill:pink;stroke;none;");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", "0");
            w.writeAttribute("width", String.valueOf(this.genotype_width));
            w.writeAttribute("height", String.valueOf(this.genotype_width));
            w.writeEndElement();
            w.writeStartElement("g");
            // 
            w.writeAttribute("id", "g_" + GenotypeType.UNAVAILABLE);
            w.writeEmptyElement("rect");
            w.writeAttribute("style", "fill:gray;stroke;none;");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", "0");
            w.writeAttribute("width", String.valueOf(this.genotype_width));
            w.writeAttribute("height", String.valueOf(this.genotype_width));
            w.writeEndElement();
            w.writeStartElement("g");
            // 
            w.writeAttribute("id", "g_" + GenotypeType.HET);
            w.writeEmptyElement("rect");
            w.writeAttribute("style", "fill:lime;stroke;black;");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", "0");
            w.writeAttribute("width", String.valueOf(genotype_width));
            w.writeAttribute("height", String.valueOf(genotype_width));
            w.writeEmptyElement("polygon");
            w.writeAttribute("style", "fill:crimson;stroke;black;");
            w.writeAttribute("points", "0,0 " + genotype_width + ",0 0," + genotype_width + " 0,0");
            w.writeEndElement();
            // strand
            w.writeEmptyElement("polyline");
            w.writeAttribute("id", "strandF");
            w.writeAttribute("points", "-5,-5 0,0 -5,5");
            w.writeEmptyElement("polyline");
            w.writeAttribute("id", "strandR");
            w.writeAttribute("points", "5,-5 0,0 5,5");
            // gradients
            w.writeStartElement("linearGradient");
            w.writeAttribute("id", "grad01");
            w.writeAttribute("x1", "50%");
            w.writeAttribute("x2", "50%");
            w.writeAttribute("y1", "0%");
            w.writeAttribute("y2", "100%");
            w.writeEmptyElement("stop");
            w.writeAttribute("offset", "0%");
            w.writeAttribute("style", "stop-color:black;stop-opacity:1;");
            w.writeEmptyElement("stop");
            w.writeAttribute("offset", "50%");
            w.writeAttribute("style", "stop-color:white;stop-opacity:1;");
            w.writeEmptyElement("stop");
            w.writeAttribute("offset", "100%");
            w.writeAttribute("style", "stop-color:black;stop-opacity:1;");
            w.writeEndElement();
            // defs
            w.writeEndElement();
            w.writeStartElement("style");
            w.writeCharacters("svg {fill:none; stroke:black;}\n" + "text {fill:black;stroke:none;font-size:" + (featureHeight / 1.5) + "px;}\n" + ".ruler-label { stroke:red;}\n" + ".frame { stroke:black;fill:none;}\n" + ".kgexon {fill:url(#grad01);stroke:black;}\n" + ".gcpercent {fill:url(#grad02);stroke:black;}" + ".coverage {fill:url(#grad03);stroke:black;}" + ".kgcds {fill:yellow;stroke:black;opacity:0.7;}\n" + ".variant{stroke:none;fill:red;opacity:0.2;}\n" + ".xaxis{stroke:gray;fill:none;opacity:0.2;}\n" + ".postick{font-size:9px;stroke:black;stroke-width:1;}");
            // style
            w.writeEndElement();
            final IntFunction<Integer> trim = t -> Math.max(interval.getStart(), Math.min(interval.getEnd(), t));
            final IntFunction<Double> baseToPixel = t -> margin_left + drawinAreaWidth * (t - (double) interval.getStart()) / ((double) interval.getLengthOnReference());
            final IntFunction<Double> variantIndexToPixel = idx -> {
                final double variant_width = drawinAreaWidth / (double) variants.size();
                final double midx = variant_width * idx + variant_width / 2.0;
                return margin_left + midx - genotype_width / 2.0;
            };
            final Function<VariantContext, String> variantTitle = V -> (V.getContig().startsWith("chr") ? V.getContig().substring(3) : V.getContig()) + ":" + V.getStart() + " " + V.getReference().getDisplayString();
            /**
             * title
             */
            double y = 0;
            w.writeStartElement("text");
            w.writeAttribute("x", "0");
            w.writeAttribute("y", String.valueOf(featureHeight));
            w.writeCharacters(interval.toString());
            w.writeEndElement();
            y += featureHeight;
            for (final Transcript g : transcripts) {
                int cdsHeigh = 5;
                double exonHeight = TRANSCRIPT_HEIGHT - 5;
                double midY = TRANSCRIPT_HEIGHT / 2;
                w.writeStartElement("g");
                w.writeAttribute("transform", "translate(0," + y + ")");
                title(w, g.getId());
                w.writeStartElement("text");
                w.writeAttribute("x", String.valueOf(margin_left - 10));
                w.writeAttribute("y", String.valueOf(featureHeight));
                w.writeAttribute("style", "text-anchor:end;");
                w.writeCharacters(g.getId());
                w.writeEndElement();
                /* transcript line */
                w.writeEmptyElement("line");
                w.writeAttribute("class", "kgtr");
                w.writeAttribute("x1", String.valueOf(baseToPixel.apply(trim.apply(g.getTxStart()))));
                w.writeAttribute("y1", String.valueOf(midY));
                w.writeAttribute("x2", String.valueOf(baseToPixel.apply(trim.apply(g.getTxEnd()))));
                w.writeAttribute("y2", String.valueOf(midY));
                /* strand symbols */
                for (double pixX = 0; pixX < drawinAreaWidth; pixX += 30) {
                    double pos0 = interval.getStart() + (pixX / (double) drawinAreaWidth) * interval.getLengthOnReference();
                    if (pos0 + 1 < g.getTxStart())
                        continue;
                    if (pos0 > g.getTxEnd())
                        break;
                    w.writeEmptyElement("use");
                    w.writeAttribute("class", "kgstrand");
                    w.writeAttribute("xlink", XLINK.NS, "href", "#strand" + (g.isPositiveStrand() ? "F" : "R"));
                    w.writeAttribute("x", String.valueOf(margin_left + pixX));
                    w.writeAttribute("y", String.valueOf(midY));
                }
                /* exons */
                for (final Exon exon : g.getExons()) {
                    if (exon.getStart() + 1 >= interval.getEnd())
                        continue;
                    if (exon.getEnd() <= interval.getStart())
                        continue;
                    w.writeStartElement("rect");
                    w.writeAttribute("class", "kgexon");
                    w.writeAttribute("x", String.valueOf(baseToPixel.apply(trim.apply(exon.getStart()))));
                    w.writeAttribute("y", String.valueOf(midY - exonHeight / 2));
                    w.writeAttribute("width", String.valueOf(baseToPixel.apply(trim.apply(exon.getEnd())) - baseToPixel.apply((trim.apply(exon.getStart())))));
                    w.writeAttribute("height", String.valueOf(exonHeight));
                    title(w, exon.getName());
                    w.writeEndElement();
                }
                /* coding line */
                if (!g.isNonCoding() && g.hasCodonStartDefined() && g.hasCodonStopDefined()) {
                    final double codonx1 = baseToPixel.apply(trim.apply(g.getLeftmostCodon().get().getStart()));
                    final double codonx2 = baseToPixel.apply(trim.apply(g.getRightmostCodon().get().getEnd()));
                    w.writeEmptyElement("rect");
                    w.writeAttribute("class", "kgcds");
                    w.writeAttribute("x", String.valueOf(codonx1));
                    w.writeAttribute("y", String.valueOf(midY - cdsHeigh / 4.0));
                    w.writeAttribute("width", String.valueOf(baseToPixel.apply((int) (codonx2 - codonx1))));
                    w.writeAttribute("height", String.valueOf(cdsHeigh / 2.0));
                }
                // String label=String.format("%15s", g.getName());
                // w.writeEmptyElement("path");
                // double fontHeight=Math.min(10,0.8*TRANSCRIPT_HEIGHT);
                // w.writeAttribute("d",this.hershey.svgPath(label,-insets.left,midY-fontHeight/2,insets.left*0.9,fontHeight));
                w.writeEndElement();
                w.writeCharacters("\n");
                y += featureHeight;
            }
            /* draw lines to variants */
            for (int vidx = 0; vidx < variants.size(); ++vidx) {
                final VariantContext vc = variants.get(vidx);
                double x1 = baseToPixel.apply(vc.getStart());
                double x2 = baseToPixel.apply(vc.getEnd());
                final double y2 = y + featureHeight * interline_weight;
                w.writeStartElement("polygon");
                w.writeAttribute("style", "fill:" + (vidx % 2 == 0 ? "ghostwhite" : "lavender") + ";stroke:black;opacity:0.6;stroke-width:0.5;");
                w.writeAttribute("points", "" + x1 + "," + (y - featureHeight / 2.0) + " " + x2 + "," + (y - featureHeight / 2.0) + " " + variantIndexToPixel.apply(vidx) + "," + y2 + " " + (variantIndexToPixel.apply(vidx) + this.genotype_width) + "," + y2);
                title(w, variantTitle.apply(vc));
                w.writeEndElement();
            }
            for (int vidx = 0; vidx < variants.size(); ++vidx) {
                final VariantContext vc = variants.get(vidx);
                final double y2 = y + featureHeight * interline_weight;
                w.writeStartElement("text");
                w.writeAttribute("transform", "translate(" + (String.valueOf(variantIndexToPixel.apply(vidx) + genotype_width / 2.0)) + "," + String.valueOf(y2 - 5) + ") " + "rotate(-45)");
                w.writeAttribute("x", "0");
                w.writeAttribute("y", "0");
                w.writeAttribute("class", "postick");
                w.writeCharacters(variantTitle.apply(vc));
                w.writeEndElement();
                w.writeCharacters("\n");
            }
            y += featureHeight * interline_weight;
            w.writeStartElement("g");
            /* step 0: affected, 1: unaffected, 2: others */
            for (int step = 0; step < 3; ++step) {
                for (final String sample : samples) {
                    final Sample individual = pedigree.getSampleById(sample);
                    if (step == 0 && (individual == null || !individual.isAffected()))
                        continue;
                    if (step == 1 && (individual == null || !individual.isUnaffected()))
                        continue;
                    if (step == 2 && individual != null && individual.isStatusSet())
                        continue;
                    // sample
                    w.writeStartElement("g");
                    switch(step) {
                        case 0:
                            w.writeAttribute("style", "hue-rotate(195deg);");
                            break;
                        case 1:
                            w.writeAttribute("style", "hue-rotate(45deg);");
                            break;
                        default:
                            break;
                    }
                    for (int vidx = 0; vidx < variants.size(); ++vidx) {
                        final VariantContext vc = variants.get(vidx);
                        final Genotype g = vc.getGenotype(sample);
                        double opacity = 1.0;
                        if (vc.isIndel())
                            opacity *= this.variantIndelOpacity;
                        if (vc.isFiltered())
                            opacity *= this.variantFILTEREDOpacity;
                        if (opacity > 1)
                            opacity = 1;
                        if (opacity <= 0)
                            continue;
                        if (opacity < 1) {
                            w.writeStartElement("g");
                            w.writeAttribute("style", "opacity:" + opacity + ";");
                        }
                        w.writeEmptyElement("use");
                        w.writeAttribute("x", String.valueOf(variantIndexToPixel.apply(vidx)));
                        w.writeAttribute("y", String.valueOf(y));
                        w.writeAttribute("xlink", XLINK.NS, "href", "#g_" + g.getType());
                        if (opacity < 1) {
                            w.writeEndElement();
                        }
                    }
                    w.writeCharacters("\n");
                    w.writeStartElement("text");
                    w.writeAttribute("x", String.valueOf(margin_left - 10));
                    w.writeAttribute("y", String.valueOf(y + this.genotype_width / 2.0));
                    w.writeAttribute("style", "text-anchor:end;");
                    w.writeCharacters(sample);
                    // text
                    w.writeEndElement();
                    // g for sample
                    w.writeEndElement();
                    y += this.genotype_width;
                }
            }
            w.writeCharacters("\n");
            w.writeEndDocument();
            w.writeCharacters("\n");
            w.flush();
            w.close();
            final String md5 = StringUtils.md5(interval.getContig() + ":" + interval.getStart() + "-" + interval.getEnd());
            final String filename = md5.substring(0, 2) + File.separatorChar + md5.substring(2) + File.separator + interval.getContig().replaceAll("[/\\:]", "_") + "_" + interval.getStart() + "_" + interval.getEnd() + (StringUtils.isBlank(geneName) ? "" : "." + geneName.replaceAll("[/\\:]", "")) + (StringUtils.isBlank(geneId) ? "" : "." + geneId.replaceAll("[/\\:]", "")) + ".svg";
            OutputStream os = archiveFactory.openOuputStream(filename);
            IOUtils.copyTo(tmpSvg, os);
            os.flush();
            os.close();
            Files.delete(tmpSvg);
            manifestW.print(interval.getContig());
            manifestW.print('\t');
            manifestW.print(interval.getStart() - 1);
            manifestW.print('\t');
            manifestW.print(interval.getEnd());
            manifestW.print('\t');
            manifestW.print(transcripts.stream().map(G -> G.getGene().getId()).collect(Collectors.toSet()).stream().collect(Collectors.joining(";")));
            manifestW.print('\t');
            manifestW.print(transcripts.stream().map(G -> G.getGene().getGeneName()).collect(Collectors.toSet()).stream().collect(Collectors.joining(";")));
            manifestW.print('\t');
            manifestW.print(transcripts.stream().map(G -> G.getId()).collect(Collectors.toSet()).stream().collect(Collectors.joining(";")));
            manifestW.print('\t');
            manifestW.print((archiveFactory.isTarOrZipArchive() ? "" : this.outputPath.toString() + File.separator) + filename);
            manifestW.print('\t');
            manifestW.println(variants.size());
        }
        r.close();
        manifestW.flush();
        manifestW.close();
        manifestW = null;
        archiveFactory.close();
        archiveFactory = null;
        return 0;
    } catch (final Throwable err) {
        LOG.error(err);
        return -1;
    } finally {
        CloserUtil.close(archiveFactory);
        CloserUtil.close(r);
        CloserUtil.close(outputStream);
        CloserUtil.close(manifestW);
    }
}

Example 8 with Sample

use of com.github.lindenb.jvarkit.pedigree.Sample in project jvarkit by lindenb.

the class VCFTrios method doVcfToVcf.

@Override
public int doVcfToVcf(final String inputName, VCFIterator r, final VariantContextWriter w) {
    long count_incompats = 0L;
    final Set<String> sampleNotFoundInVcf = new HashSet<>();
    Pedigree pedigree = null;
    final List<TrioTriple> trios = new ArrayList<>();
    try {
        final DeNovoDetector detector = new DeNovoDetector();
        detector.setConvertingNoCallToHomRef(this.nocall_to_homref);
        final VCFHeader header = r.getHeader();
        final PedigreeParser pedParser = new PedigreeParser();
        pedigree = pedParser.parse(this.pedigreeFile);
        final VCFHeader h2 = new VCFHeader(header);
        final Set<VCFHeaderLine> meta = new HashSet<>();
        meta.add(new VCFInfoHeaderLine(this.attributeName, VCFHeaderLineCount.UNBOUNDED, VCFHeaderLineType.String, "Samples with mendelian incompatibilities." + (this.pedigreeFile == null ? "" : " Pedigree File was : " + this.pedigreeFile)));
        meta.add(VCFStandardHeaderLines.getFormatLine(VCFConstants.GENOTYPE_FILTER_KEY, true));
        if (!StringUtil.isBlank(this.filterAnyIncompat)) {
            meta.add(new VCFFilterHeaderLine(this.filterAnyIncompat, "Variant contains at least one mendelian incompatibilities"));
        }
        if (!StringUtil.isBlank(this.filterNoIncompat)) {
            meta.add(new VCFFilterHeaderLine(this.filterNoIncompat, "Variant does not contain any mendelian incompatibilities"));
        }
        meta.stream().forEach(H -> h2.addMetaDataLine(H));
        JVarkitVersion.getInstance().addMetaData(this, h2);
        for (final Trio pedTrio : pedigree.getTrios()) {
            final TrioTriple trio = new TrioTriple();
            final Sample child = pedTrio.getChild();
            trio.child_id = header.getSampleNameToOffset().getOrDefault(child.getId(), -1);
            if (trio.child_id < 0)
                continue;
            if (pedTrio.hasFather()) {
                final Sample parent = pedTrio.getFather();
                trio.father_id = header.getSampleNameToOffset().getOrDefault(parent.getId(), -1);
            }
            if (pedTrio.hasMother()) {
                final Sample parent = pedTrio.getMother();
                trio.mother_id = header.getSampleNameToOffset().getOrDefault(parent.getId(), -1);
            }
            if (trio.father_id == -1 && trio.mother_id == -1) {
                continue;
            }
            trios.add(trio);
        }
        LOG.info("trios(s) in pedigree: " + trios.size());
        final ProgressFactory.Watcher<VariantContext> progress = ProgressFactory.newInstance().dictionary(header).logger(LOG).build();
        w.writeHeader(h2);
        while (r.hasNext()) {
            final VariantContext ctx = progress.apply(r.next());
            final Set<String> incompatibilities = new HashSet<String>();
            for (final TrioTriple trio : trios) {
                final Genotype gChild = ctx.getGenotype(trio.child_id);
                if (gChild == null)
                    throw new IllegalStateException();
                final Genotype gFather = trio.father_id < 0 ? null : ctx.getGenotype(trio.father_id);
                final Genotype gMother = trio.mother_id < 0 ? null : ctx.getGenotype(trio.mother_id);
                final DeNovoDetector.DeNovoMutation mut = detector.test(ctx, gFather, gMother, gChild);
                if (mut != null) {
                    incompatibilities.add(gChild.getSampleName());
                }
            }
            final VariantContextBuilder vcb = new VariantContextBuilder(ctx);
            if (!incompatibilities.isEmpty()) {
                // set filter for samples that are not a mendelian violation
                if (!StringUtil.isBlank(this.genotypeFilterNameNoIncompat)) {
                    vcb.genotypes(ctx.getGenotypes().stream().map(G -> incompatibilities.contains(G.getSampleName()) ? G : new GenotypeBuilder(G).filters(this.genotypeFilterNameNoIncompat).make()).collect(Collectors.toList()));
                }
                ++count_incompats;
                // set INFO attribute
                vcb.attribute(attributeName, incompatibilities.toArray());
                // set FILTER
                if (!StringUtil.isBlank(this.filterAnyIncompat)) {
                    vcb.filter(this.filterAnyIncompat);
                } else if (!ctx.isFiltered()) {
                    vcb.passFilters();
                }
            } else // No denovo
            {
                // dicard variant
                if (this.discard_variants_without_mendelian_incompat) {
                    continue;
                }
                // set filters
                if (!StringUtil.isBlank(this.filterNoIncompat)) {
                    vcb.filter(this.filterNoIncompat);
                } else if (!ctx.isFiltered()) {
                    vcb.passFilters();
                }
            }
            w.add(vcb.make());
        }
        progress.close();
        LOG.info("incompatibilitie(s) N=" + count_incompats);
        if (!sampleNotFoundInVcf.isEmpty()) {
            LOG.info("SAMPLE(S) not found: " + String.join(" / ", sampleNotFoundInVcf));
        }
        return 0;
    } catch (final Throwable err) {
        LOG.error(err);
        return -1;
    }
}

Example 9 with Sample

use of com.github.lindenb.jvarkit.pedigree.Sample in project jvarkit by lindenb.

the class AbstractVcfBurden method runFisher.

protected FisherResult runFisher(final List<VariantContext> variants) {
    final Map<Sample, SuperVariant> indi2supervariant = new HashMap<>(this.cases.size() + this.controls.size());
    this.cases.stream().forEach(S -> indi2supervariant.put(S, SuperVariant.SV0));
    this.controls.stream().forEach(S -> indi2supervariant.put(S, SuperVariant.SV0));
    for (final VariantContext ctx : variants) {
        final Allele observed_alt = ctx.getAltAlleleWithHighestAlleleCount();
        for (final Sample sample : indi2supervariant.keySet()) {
            if (indi2supervariant.get(sample) == SuperVariant.AT_LEAST_ONE_VARIANT)
                continue;
            final Genotype g = ctx.getGenotype(sample.getId());
            if (g == null || g.isHomRef() || g.isNoCall())
                continue;
            if (g.getAlleles().stream().anyMatch(A -> A.equals(observed_alt))) {
                indi2supervariant.put(sample, SuperVariant.AT_LEAST_ONE_VARIANT);
                break;
            }
        }
    // end for sample
    }
    // end of forVariant
    final FisherResult fisher = new FisherResult();
    for (final Sample sample : indi2supervariant.keySet()) {
        final SuperVariant superVariant = indi2supervariant.get(sample);
        if (superVariant == SuperVariant.SV0) {
            if (sample.isAffected())
                fisher.affected_hom++;
            else
                fisher.unaffected_hom++;
        } else // AT_LEAST_ONE_VARIANT
        {
            if (sample.isAffected())
                fisher.affected_alt++;
            else
                fisher.unaffected_alt++;
        }
    }
    // end of sample
    final FisherExactTest test = FisherExactTest.compute(fisher.affected_alt, fisher.affected_hom, fisher.unaffected_alt, fisher.unaffected_hom);
    fisher.p_value = test.getAsDouble();
    return fisher;
}

Example 10 with Sample

use of com.github.lindenb.jvarkit.pedigree.Sample in project jvarkit by lindenb.

the class CombineVcfFisher method loadGeneGenotypes.

private GeneGenotypes loadGeneGenotypes(final Path path) throws IOException {
    try (VCFReader reader = VCFReaderFactory.makeDefault().open(path, false)) {
        final VCFHeader header = reader.getHeader();
        if (!header.hasGenotypingData()) {
            LOG.error("not genotype in " + path);
            return null;
        }
        /**
         * this is the first VCF header we ever had
         */
        if (this.genotypeSamples == null) {
            /**
             * build the list of sample
             */
            this.genotypeSamples = header.getGenotypeSamples();
            /**
             * build the list of sample status
             */
            this.sample_status = new int[this.genotypeSamples.size()];
            /**
             * fill list of sample status
             */
            for (int i = 0; i < this.genotypeSamples.size(); i++) {
                final String sn = this.genotypeSamples.get(i);
                final Sample sample = this.pedigree.getSampleById(sn);
                int status = STATUS_UNDEFINED;
                if (sample != null) {
                    switch(sample.getStatus()) {
                        case affected:
                            status = STATUS_AFFECTED;
                            break;
                        case unaffected:
                            status = STATUS_UNAFFECTED;
                            break;
                        default:
                            status = STATUS_UNDEFINED;
                            break;
                    }
                }
                this.sample_status[i] = status;
            }
            // check we have affected
            if (IntStream.of(this.sample_status).noneMatch(V -> V == STATUS_AFFECTED)) {
                LOG.warn("No affected sample in " + path);
                return null;
            }
            // check we have unaffected
            if (IntStream.of(this.sample_status).noneMatch(V -> V == STATUS_UNAFFECTED)) {
                LOG.warn("No unaffected sample in " + path);
                return null;
            }
        } else if (!this.genotypeSamples.equals(header.getGenotypeSamples())) {
            throw new IOException("samples are not the same with previous VCF and " + path);
        }
        // loop over variants
        final GeneGenotypes geneGenotypes = new GeneGenotypes(IOUtils.getFilenameWithoutCommonSuffixes(path));
        try (CloseableIterator<VariantContext> iter = reader.iterator()) {
            while (iter.hasNext()) {
                final VariantContext ctx = iter.next();
                if (ctx.isFiltered() || !ctx.isVariant())
                    continue;
                geneGenotypes.nVariants++;
                for (int i = 0; i < this.sample_status.length; i++) {
                    if (this.sample_status[i] == STATUS_UNDEFINED)
                        continue;
                    if (geneGenotypes.genotypes.get(i))
                        continue;
                    final Genotype gt = ctx.getGenotype(i);
                    if (gt.isHomRef() || gt.isNoCall())
                        continue;
                    geneGenotypes.genotypes.set(i);
                }
            }
        }
        if (geneGenotypes.nVariants == 0 || geneGenotypes.genotypes.nextSetBit(0) == -1) {
            LOG.warn("no valid data for " + path);
            return null;
        }
        return geneGenotypes;
    }
}