Examples with TabixKnownGeneFileReader com.github.lindenb.jvarkit.util.ucsc.TabixKnownGeneFileReader used on opensource projects

Example 1 with TabixKnownGeneFileReader

use of com.github.lindenb.jvarkit.util.ucsc.TabixKnownGeneFileReader in project jvarkit by lindenb.

the class VcfToSvg method doWork.

@Override
public int doWork(final List<String> args) {
    if (this.outputFile != null && !outputFile.getName().contains(SEGMENT)) {
        LOG.error("output file must contain the word " + SEGMENT + " :" + this.outputFile);
        return -1;
    }
    TabixKnownGeneFileReader tabix = null;
    VcfIterator r = null;
    OutputStream outputStream = null;
    XMLStreamWriter w = null;
    PrintWriter manifestW = null;
    try {
        LOG.info("opening knownGene ");
        tabix = new TabixKnownGeneFileReader(knownGeneUri);
        if (manifestFile != null && this.outputFile != null) {
            manifestW = new PrintWriter(manifestFile);
        } else {
            manifestW = new PrintWriter(new NullOuputStream());
        }
        final Set<String> chromosomes = tabix.getChromosomes();
        final XMLOutputFactory xof = XMLOutputFactory.newInstance();
        r = super.openVcfIterator(super.oneFileOrNull(args));
        final VCFHeader header = r.getHeader();
        while (r.hasNext()) {
            final VariantContext ctx = r.next();
            String tabixContig = ctx.getContig();
            if (!chromosomes.contains(tabixContig)) {
                if (tabixContig.startsWith("chr")) {
                    tabixContig = tabixContig.substring(3);
                } else if (!tabixContig.startsWith("chr")) {
                    tabixContig = "chr" + tabixContig;
                }
                if (!chromosomes.contains(tabixContig)) {
                    while (r.hasNext()) {
                        final VariantContext ctx2 = r.peek();
                        if (!ctx2.getContig().equals(ctx.getContig()))
                            break;
                        r.next();
                    }
                    LOG.error("No chromosome " + ctx.getContig() + " in " + knownGeneUri + ". Check the chromosome nomenclature.");
                    continue;
                }
            }
            final List<VariantContext> variants = new ArrayList<>();
            final List<KnownGene> genes = new ArrayList<>();
            variants.add(ctx);
            int chromStart = ctx.getStart() - 1;
            int chromEnd = ctx.getEnd();
            /* walk over know gene, loop until there is no overapping transcript 
			 * over that region */
            for (; ; ) {
                genes.clear();
                /* the max chromEnd, let's see if we can get a bigger */
                int newStart = chromStart;
                int newEnd = chromEnd;
                final Iterator<KnownGene> kgr = tabix.iterator(tabixContig, chromStart, chromEnd);
                while (kgr.hasNext()) {
                    final KnownGene g = kgr.next();
                    if (this.removeNonCoding && g.isNonCoding())
                        continue;
                    genes.add(g);
                    newStart = Math.min(g.getTxStart(), newStart);
                    newEnd = Math.max(g.getTxEnd(), newEnd);
                }
                if (newStart >= chromStart && newEnd <= chromEnd) {
                    break;
                }
                chromStart = newStart;
                chromEnd = newEnd;
            }
            // intergenic, no gene over that variant
            if (genes.isEmpty())
                continue;
            // fill the variant for that region
            while (r.hasNext()) {
                final VariantContext ctx2 = r.peek();
                if (!ctx2.getContig().equals(ctx.getContig()))
                    break;
                if (ctx2.getStart() > chromEnd)
                    break;
                variants.add(r.next());
            }
            if (this.variantsInExonOnly) {
                variants.removeIf(V -> {
                    for (final KnownGene gene : genes) {
                        for (final KnownGene.Exon exon : gene.getExons()) {
                            if (V.getEnd() < exon.getStart() || V.getStart() >= exon.getEnd()) {
                            // rien
                            } else {
                                return false;
                            }
                        }
                    }
                    return true;
                });
            }
            if (this.variantFILTEREDOpacity <= 0) {
                variants.removeIf(V -> V.isFiltered());
            }
            if (this.variantIndelOpacity <= 0) {
                variants.removeIf(V -> V.isIndel());
            }
            if (variants.isEmpty())
                continue;
            LOG.info("Variants (" + variants.size() + ") Transcripts (" + genes.size() + ") " + tabixContig + ":" + chromStart + "-" + chromEnd);
            if (outputFile != null) {
                File fname = new File(outputFile.getParentFile(), outputFile.getName().replaceAll("__SEGMENT__", ctx.getContig() + "_" + chromStart + "_" + chromEnd));
                LOG.info("saving as " + fname);
                outputStream = IOUtils.openFileForWriting(fname);
                w = xof.createXMLStreamWriter(outputStream);
                manifestW.println(ctx.getContig() + "\t" + chromStart + "\t" + chromEnd + "\t" + genes.stream().map(G -> G.getName()).collect(Collectors.joining(",")) + "\t" + genes.size() + "\t" + variants.size() + "\t" + fname);
            } else {
                w = xof.createXMLStreamWriter(stdout());
            }
            double featureHeight = 10;
            double TRANSCRIPT_HEIGHT = featureHeight;
            final int all_genotypes_width = variants.size() * this.genotype_width;
            if (trimToVariants) {
                chromStart = variants.stream().map(V -> V.getStart() - 1).min((A, B) -> A.compareTo(B)).get();
                chromEnd = variants.stream().map(V -> V.getEnd() + 1).max((A, B) -> A.compareTo(B)).get();
            }
            final int drawinAreaWidth = Math.max(all_genotypes_width, 1000);
            final Interval interval = new Interval(ctx.getContig(), chromStart, chromEnd);
            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 + genes.size() * TRANSCRIPT_HEIGHT + interline_weight * featureHeight + header.getSampleNamesInOrder().size() * this.genotype_width));
            title(w, ctx.getContig() + ":" + chromStart + "-" + chromEnd);
            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 Function<Integer, Integer> trim = new Function<Integer, Integer>() {

                @Override
                public Integer apply(final Integer t) {
                    return Math.max(interval.getStart(), Math.min(interval.getEnd(), t));
                }
            };
            final Function<Integer, Double> baseToPixel = new Function<Integer, Double>() {

                @Override
                public Double apply(final Integer t) {
                    return margin_left + drawinAreaWidth * (t - (double) interval.getStart()) / ((double) interval.length());
                }
            };
            final Function<Integer, Double> variantIndexToPixel = new Function<Integer, Double>() {

                @Override
                public Double apply(final Integer 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 KnownGene g : genes) {
                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.getName());
                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.getName());
                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.length();
                    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 (KnownGene.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()) {
                    w.writeEmptyElement("rect");
                    w.writeAttribute("class", "kgcds");
                    w.writeAttribute("x", String.valueOf(baseToPixel.apply(trim.apply(g.getCdsStart()))));
                    w.writeAttribute("y", String.valueOf(midY - cdsHeigh / 4.0));
                    w.writeAttribute("width", String.valueOf(baseToPixel.apply(trim.apply(g.getCdsEnd())) - baseToPixel.apply((trim.apply((g.getCdsStart()))))));
                    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");
            for (final String sample : header.getSampleNamesInOrder()) {
                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", "" + 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);
                w.writeEndElement();
                y += this.genotype_width;
            }
            w.writeCharacters("\n");
            w.writeEndDocument();
            w.writeCharacters("\n");
            w.flush();
            w.close();
            if (outputFile != null) {
                outputStream.flush();
                outputStream.close();
                outputStream = null;
            }
            if (stop_at_first) {
                LOG.info("Stop after first SVG document");
                break;
            }
        }
        r.close();
        manifestW.flush();
        manifestW.close();
        manifestW = null;
        return 0;
    } catch (Throwable err) {
        LOG.error(err);
        return -1;
    } finally {
        CloserUtil.close(r);
        CloserUtil.close(tabix);
        CloserUtil.close(outputStream);
        CloserUtil.close(manifestW);
    }
}

Example 2 with TabixKnownGeneFileReader

use of com.github.lindenb.jvarkit.util.ucsc.TabixKnownGeneFileReader in project jvarkit by lindenb.

the class TView method initialize.

public int initialize() throws IOException {
    if (this.referenceFile != null) {
        this.indexedFastaSequenceFile = new IndexedFastaSequenceFile(this.referenceFile);
    }
    if (this.samRecordFilter == null) {
        this.samRecordFilter = SamFilterParser.ACCEPT_ALL;
    }
    final SamReaderFactory srf = SamReaderFactory.makeDefault().referenceSequence(this.referenceFile).validationStringency(ValidationStringency.LENIENT);
    for (final SamInputResource sir : this.samInputResources) {
        final SamReader samReader = srf.open(sir);
        this.samReaders.add(samReader);
    }
    for (final File vcfFile : IOUtils.unrollFile(this.variantFiles)) {
        final VcfSource vcfSource = new VcfSource();
        LOG.debug("OPEN " + vcfFile);
        vcfSource.vcfFile = vcfFile;
        vcfSource.vcfFileReader = new VCFFileReader(vcfFile, true);
        this.vcfReaders.add(vcfSource);
    }
    if (this.tabixKnownGene != null) {
        this.tabixKnownGene = new TabixKnownGeneFileReader(this.knownGeneUri);
    }
    return 0;
}

Example 3 with TabixKnownGeneFileReader

use of com.github.lindenb.jvarkit.util.ucsc.TabixKnownGeneFileReader in project jvarkit by lindenb.

the class VcfDoest method doWork.

@Override
public int doWork(final List<String> args) {
    if (this.knownGeneURI == null || this.knownGeneURI.trim().isEmpty()) {
        LOG.error("undefined option knownGeneURI");
        return -1;
    }
    if (!this.knownGeneURI.endsWith(".gz")) {
        LOG.error(this.knownGeneURI + " doesn't end with '.gz'");
        return -1;
    }
    LOG.info("reading " + this.knownGeneURI);
    BufferedReader r = null;
    LineIterator li = null;
    PrintWriter pw = null;
    try {
        LOG.info("loading tabix knownGene :" + this.knownGeneURI);
        this.knownGenesTabix = new TabixKnownGeneFileReader(this.knownGeneURI);
        final String inputFile = oneFileOrNull(args);
        pw = super.openFileOrStdoutAsPrintWriter(this.outputFile);
        li = (inputFile == null ? IOUtils.openStreamForLineIterator(stdin()) : IOUtils.openFileForLineIterator(new File(inputFile)));
        run(li, pw);
        CloserUtil.close(li);
        li = null;
        pw.flush();
        pw.close();
        pw = null;
        LOG.info("done");
        return RETURN_OK;
    } catch (Exception e) {
        LOG.error(e);
        return -1;
    } finally {
        CloserUtil.close(r);
        CloserUtil.close(li);
        CloserUtil.close(pw);
        CloserUtil.close(this.knownGenesTabix);
    }
}