Examples with SAMReadGroupRecord htsjdk.samtools.SAMReadGroupRecord used on opensource projects

Example 76 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project jvarkit by lindenb.

the class XContaminations method doWork.

@Override
public int doWork(final List<String> args) {
    long last_save_ms = System.currentTimeMillis();
    if (this.output_as_vcf && !this.use_only_sample_name) {
        LOG.error("cannot write vcf if --sample is not set");
        return -1;
    }
    if (args.size() < 2) {
        LOG.error("Illegal Number of args");
        return -1;
    }
    final Set<File> bamFiles = IOUtils.unrollFiles(args.subList(1, args.size())).stream().map(S -> new File(S)).collect(Collectors.toSet());
    if (bamFiles.isEmpty()) {
        LOG.error("Undefined BAM file(s)");
        return -1;
    }
    SAMRecordIterator iter = null;
    VcfIterator in = null;
    Map<String, SamReader> sample2samReader = new HashMap<>();
    VariantContextWriter vcfw = null;
    try {
        final SamReaderFactory srf = super.createSamReaderFactory();
        if (args.get(0).equals("-")) {
            in = super.openVcfIterator(null);
        } else {
            in = super.openVcfIterator(args.get(0));
        }
        VCFHeader vcfHeader = in.getHeader();
        final SAMSequenceDictionary dict1 = vcfHeader.getSequenceDictionary();
        if (dict1 == null) {
            LOG.error(JvarkitException.VcfDictionaryMissing.getMessage(args.get(0)));
            return -1;
        }
        final Set<String> sampleNames = new HashSet<>(vcfHeader.getSampleNamesInOrder());
        if (sampleNames.isEmpty()) {
            LOG.error("VCF contains no sample");
            return -1;
        }
        for (final File bamFile : bamFiles) {
            LOG.info("Opening " + bamFile);
            final SamReader samReader = srf.open(bamFile);
            final SAMFileHeader samHeader = samReader.getFileHeader();
            final SAMSequenceDictionary dict2 = samHeader.getSequenceDictionary();
            if (dict2 == null) {
                samReader.close();
                LOG.error(JvarkitException.BamDictionaryMissing.getMessage(bamFile.getPath()));
                return -1;
            }
            if (!SequenceUtil.areSequenceDictionariesEqual(dict1, dict2)) {
                samReader.close();
                LOG.error(JvarkitException.DictionariesAreNotTheSame.getMessage(dict1, dict2));
                return -1;
            }
            if (!samReader.hasIndex()) {
                samReader.close();
                LOG.error("sam is not indexed : " + bamFile);
                return -1;
            }
            String sampleName = null;
            for (final SAMReadGroupRecord rgr : samHeader.getReadGroups()) {
                final String s = rgr.getSample();
                if (StringUtil.isBlank(s))
                    continue;
                if (sampleName == null) {
                    sampleName = s;
                } else if (!sampleName.equals(s)) {
                    samReader.close();
                    LOG.error("Cannot handle more than one sample/bam  " + bamFile + " " + sampleName);
                    return -1;
                }
            }
            if (sampleName == null) {
                samReader.close();
                LOG.error("No sample in " + bamFile);
                // skip this bam
                continue;
            }
            if (!sampleNames.contains(sampleName)) {
                samReader.close();
                LOG.error("Not in VCF header: sample " + sampleName + " " + bamFile);
                // skip this bam
                continue;
            }
            if (sample2samReader.containsKey(sampleName)) {
                samReader.close();
                LOG.error("Cannot handle more than one bam/sample: " + bamFile + " " + sampleName);
                return -1;
            }
            sample2samReader.put(sampleName, samReader);
        }
        if (sample2samReader.size() < 2) {
            LOG.error("Not engough BAM/samples. Expected at least two valid BAMs");
            return -1;
        }
        sampleNames.retainAll(sample2samReader.keySet());
        /* create a VCF is VCF output asked */
        final List<SamplePair> sampleListForVcf;
        if (this.output_as_vcf) {
            vcfw = super.openVariantContextWriter(outputFile);
            final Set<VCFHeaderLine> metaData = new HashSet<>();
            metaData.add(new VCFFormatHeaderLine("S1S1", 1, VCFHeaderLineType.Integer, "reads sample 1 supporting sample 1"));
            metaData.add(new VCFFormatHeaderLine("S1S2", 1, VCFHeaderLineType.Integer, "reads sample 1 supporting sample 2"));
            metaData.add(new VCFFormatHeaderLine("S1SO", 1, VCFHeaderLineType.Integer, "reads sample 1 supporting others"));
            metaData.add(new VCFFormatHeaderLine("S2S1", 1, VCFHeaderLineType.Integer, "reads sample 2 supporting sample 1"));
            metaData.add(new VCFFormatHeaderLine("S2S2", 1, VCFHeaderLineType.Integer, "reads sample 2 supporting sample 2"));
            metaData.add(new VCFFormatHeaderLine("S2SO", 1, VCFHeaderLineType.Integer, "reads sample 2 supporting others"));
            metaData.add(new VCFFormatHeaderLine("FR", 1, VCFHeaderLineType.Float, "Fraction. '-1' for unavailable."));
            metaData.add(new VCFFormatHeaderLine("S1A", 1, VCFHeaderLineType.Character, "sample 1 allele"));
            metaData.add(new VCFFormatHeaderLine("S2A", 1, VCFHeaderLineType.Character, "sample 2 allele"));
            metaData.add(new VCFFilterHeaderLine("XCONTAMINATION", "Fraction test is > " + fraction_treshold));
            metaData.add(new VCFFilterHeaderLine("BADSAMPLES", "At least one pair of genotype fails the 'LE' test"));
            metaData.add(new VCFInfoHeaderLine("LE", 1, VCFHeaderLineType.Integer, "number of pair of genotypes having (S1S1<=S1S2 or S2S2<=S2S1)."));
            metaData.add(new VCFInfoHeaderLine("BADSAMPLES", VCFHeaderLineCount.UNBOUNDED, VCFHeaderLineType.String, "Samples founds failing the 'LE' test"));
            sampleListForVcf = new ArrayList<>();
            final List<String> sampleList = new ArrayList<>(sampleNames);
            for (int x = 0; x + 1 < sampleList.size(); ++x) {
                for (int y = x + 1; y < sampleList.size(); ++y) {
                    sampleListForVcf.add(new SamplePair(new SimpleSampleIdenfifier(sampleList.get(x)), new SimpleSampleIdenfifier(sampleList.get(y))));
                }
            }
            final VCFHeader header2 = new VCFHeader(metaData, sampleListForVcf.stream().map(V -> V.getLabel()).sorted().collect(Collectors.toList()));
            header2.setSequenceDictionary(dict1);
            vcfw.writeHeader(header2);
        } else {
            vcfw = null;
            sampleListForVcf = null;
        }
        final Map<SamplePair, SampleAlleles> contaminationTable = new HashMap<>();
        final SAMSequenceDictionaryProgress progress = new SAMSequenceDictionaryProgress(dict1).logger(LOG);
        while (in.hasNext()) {
            final VariantContext ctx = progress.watch(in.next());
            if (!ctx.isSNP() || ctx.isFiltered() || !ctx.isBiallelic() || ctx.isSymbolic() || !this.variantFilter.test(ctx)) {
                continue;
            }
            int count_homref = 0;
            int count_homvar = 0;
            int count_het = 0;
            final Map<String, Genotype> sample2gt = new HashMap<>();
            for (int gidx = 0; gidx < ctx.getNSamples(); ++gidx) {
                final Genotype G = ctx.getGenotype(gidx);
                if (!G.isCalled())
                    continue;
                if (G.isHet()) {
                    // here because in use_singleton we must be sure that there is only one hom_var
                    count_het++;
                    if (this.use_singleton && count_het > 0)
                        break;
                } else if (G.isHomVar()) {
                    // here because in use_singleton we must be sure that there is only one hom_var
                    count_homvar++;
                    if (this.use_singleton && count_homvar > 1)
                        break;
                }
                if (G.isFiltered())
                    continue;
                if (!sample2samReader.containsKey(G.getSampleName()))
                    continue;
                if (!sampleNames.contains(G.getSampleName()))
                    continue;
                if (!this.genotypeFilter.test(ctx, G))
                    continue;
                sample2gt.put(G.getSampleName(), G);
            }
            if (this.use_singleton && count_het > 0)
                continue;
            if (this.use_singleton && count_homvar > 1)
                continue;
            if (sample2gt.size() < 2)
                continue;
            // reset and recount
            count_homref = 0;
            count_homvar = 0;
            count_het = 0;
            for (final String sampleName : sample2gt.keySet()) {
                final Genotype G = ctx.getGenotype(sampleName);
                switch(G.getType()) {
                    case HOM_REF:
                        count_homref++;
                        break;
                    case HOM_VAR:
                        count_homvar++;
                        break;
                    case HET:
                        count_het++;
                        break;
                    default:
                        break;
                }
            }
            // singleton check
            if (this.use_singleton && (count_het > 0 || count_homvar != 1)) {
                continue;
            }
            // at least one HOM_REF and one HOM_VAR
            if (count_homref == 0)
                continue;
            if (count_homvar == 0)
                continue;
            final Map<SampleIdentifier, Counter<Character>> sample_identifier_2allelesCount = new HashMap<>();
            /* scan Reads for those Genotype/Samples */
            for (final String sampleName : sample2gt.keySet()) {
                if (!sample2samReader.containsKey(sampleName))
                    continue;
                // sample name is not in vcf header
                final SamReader samReader = sample2samReader.get(sampleName);
                if (samReader == null)
                    continue;
                final Genotype genotype = sample2gt.get(sampleName);
                if (genotype == null)
                    continue;
                iter = samReader.query(ctx.getContig(), ctx.getStart(), ctx.getEnd(), false);
                while (iter.hasNext()) {
                    final SAMRecord record = iter.next();
                    if (record.getEnd() < ctx.getStart())
                        continue;
                    if (ctx.getEnd() < record.getStart())
                        continue;
                    if (record.getReadUnmappedFlag())
                        continue;
                    if (this.filter.filterOut(record))
                        continue;
                    final SAMReadGroupRecord srgr = record.getReadGroup();
                    // not current sample
                    if (srgr == null)
                        continue;
                    if (!sampleName.equals(srgr.getSample()))
                        continue;
                    final Cigar cigar = record.getCigar();
                    if (cigar == null || cigar.isEmpty())
                        continue;
                    byte[] readSeq = record.getReadBases();
                    if (readSeq == null || readSeq.length == 0)
                        continue;
                    int readPos = record.getReadPositionAtReferencePosition(ctx.getStart());
                    if (readPos < 1)
                        continue;
                    readPos--;
                    if (readPos >= readSeq.length)
                        continue;
                    final char base = Character.toUpperCase((char) readSeq[readPos]);
                    if (base == 'N')
                        continue;
                    final SampleIdentifier sampleIdentifier;
                    if (this.use_only_sample_name) {
                        sampleIdentifier = new SimpleSampleIdenfifier(sampleName);
                    } else {
                        final ShortReadName readName = ShortReadName.parse(record);
                        if (!readName.isValid()) {
                            LOG.info("No a valid read name " + record.getReadName());
                            continue;
                        }
                        sampleIdentifier = new SequencerFlowCellRunLaneSample(readName, sampleName);
                    }
                    Counter<Character> sampleAlleles = sample_identifier_2allelesCount.get(sampleIdentifier);
                    if (sampleAlleles == null) {
                        sampleAlleles = new Counter<Character>();
                        sample_identifier_2allelesCount.put(sampleIdentifier, sampleAlleles);
                    }
                    sampleAlleles.incr(base);
                }
                iter.close();
                iter = null;
            }
            /* end scan reads for this sample */
            /* sum-up data for this SNP */
            final VariantContextBuilder vcb;
            final List<Genotype> genotypeList;
            if (this.output_as_vcf) {
                vcb = new VariantContextBuilder(args.get(0), ctx.getContig(), ctx.getStart(), ctx.getEnd(), ctx.getAlleles());
                if (ctx.hasID())
                    vcb.id(ctx.getID());
                genotypeList = new ArrayList<>();
            } else {
                vcb = null;
                genotypeList = null;
            }
            for (final String sample1 : sample2gt.keySet()) {
                final Genotype g1 = sample2gt.get(sample1);
                final char a1 = g1.getAllele(0).getBaseString().charAt(0);
                for (final String sample2 : sample2gt.keySet()) {
                    if (sample1.compareTo(sample2) >= 0)
                        continue;
                    final Genotype g2 = sample2gt.get(sample2);
                    if (g2.sameGenotype(g1))
                        continue;
                    final char a2 = g2.getAllele(0).getBaseString().charAt(0);
                    for (final SampleIdentifier sfcr1 : sample_identifier_2allelesCount.keySet()) {
                        if (!sfcr1.getSampleName().equals(sample1))
                            continue;
                        final Counter<Character> counter1 = sample_identifier_2allelesCount.get(sfcr1);
                        if (counter1 == null)
                            continue;
                        for (final SampleIdentifier sfcr2 : sample_identifier_2allelesCount.keySet()) {
                            if (!sfcr2.getSampleName().equals(sample2))
                                continue;
                            final SamplePair samplePair = new SamplePair(sfcr1, sfcr2);
                            final Counter<Character> counter2 = sample_identifier_2allelesCount.get(sfcr2);
                            if (counter2 == null)
                                continue;
                            SampleAlleles sampleAlleles = contaminationTable.get(samplePair);
                            if (sampleAlleles == null) {
                                sampleAlleles = new SampleAlleles();
                                contaminationTable.put(samplePair, sampleAlleles);
                                if (!this.output_as_vcf && contaminationTable.size() % 10000 == 0)
                                    LOG.info("n(pairs)=" + contaminationTable.size());
                            }
                            sampleAlleles.number_of_comparaisons++;
                            for (final Character allele : counter1.keySet()) {
                                final long n = counter1.count(allele);
                                if (allele.equals(a1)) {
                                    sampleAlleles.reads_sample1_supporting_sample1 += n;
                                } else if (allele.equals(a2)) {
                                    sampleAlleles.reads_sample1_supporting_sample2 += n;
                                } else {
                                    sampleAlleles.reads_sample1_supporting_other += n;
                                }
                            }
                            for (final Character allele : counter2.keySet()) {
                                final long n = counter2.count(allele);
                                if (allele.equals(a2)) {
                                    sampleAlleles.reads_sample2_supporting_sample2 += n;
                                } else if (allele.equals(a1)) {
                                    sampleAlleles.reads_sample2_supporting_sample1 += n;
                                } else {
                                    sampleAlleles.reads_sample2_supporting_other += n;
                                }
                            }
                        }
                    }
                }
            }
            if (this.output_as_vcf) {
                final Set<String> bad_samples = new TreeSet<>();
                boolean fraction_flag = false;
                int num_lt = 0;
                for (final SamplePair samplepair : sampleListForVcf) {
                    final GenotypeBuilder gb = new GenotypeBuilder(samplepair.getLabel());
                    final SampleAlleles sampleAlleles = contaminationTable.get(samplepair);
                    if (sampleAlleles != null) {
                        gb.attribute("S1S1", sampleAlleles.reads_sample1_supporting_sample1);
                        gb.attribute("S1S2", sampleAlleles.reads_sample1_supporting_sample2);
                        gb.attribute("S1SO", sampleAlleles.reads_sample1_supporting_other);
                        gb.attribute("S2S1", sampleAlleles.reads_sample2_supporting_sample1);
                        gb.attribute("S2S2", sampleAlleles.reads_sample2_supporting_sample2);
                        gb.attribute("S2SO", sampleAlleles.reads_sample2_supporting_other);
                        gb.attribute("S1A", sample2gt.get(samplepair.sample1.getSampleName()).getAllele(0).getDisplayString().charAt(0));
                        gb.attribute("S2A", sample2gt.get(samplepair.sample2.getSampleName()).getAllele(0).getDisplayString().charAt(0));
                        final double fraction = sampleAlleles.getFraction();
                        gb.attribute("FR", fraction);
                        if (!this.passFractionTreshold.test(fraction)) {
                            fraction_flag = true;
                        }
                        boolean bad_lt_flag = false;
                        if (sampleAlleles.reads_sample1_supporting_sample1 <= this.fail_factor * sampleAlleles.reads_sample1_supporting_sample2) {
                            bad_samples.add(samplepair.sample1.getSampleName());
                            bad_lt_flag = true;
                        }
                        if (sampleAlleles.reads_sample2_supporting_sample2 <= this.fail_factor * sampleAlleles.reads_sample2_supporting_sample1) {
                            bad_samples.add(samplepair.sample2.getSampleName());
                            bad_lt_flag = true;
                        }
                        if (bad_lt_flag) {
                            num_lt++;
                        }
                    } else {
                        gb.attribute("S1S1", -1);
                        gb.attribute("S1S2", -1);
                        gb.attribute("S1SO", -1);
                        gb.attribute("S2S1", -1);
                        gb.attribute("S2S2", -1);
                        gb.attribute("S2SO", -1);
                        gb.attribute("S1A", '.');
                        gb.attribute("S2A", '.');
                        gb.attribute("FR", -1f);
                    }
                    genotypeList.add(gb.make());
                }
                if (!bad_samples.isEmpty()) {
                    vcb.attribute("BADSAMPLES", new ArrayList<>(bad_samples));
                }
                vcb.attribute("LE", num_lt);
                if (fraction_flag || !bad_samples.isEmpty()) {
                    if (fraction_flag)
                        vcb.filter("XCONTAMINATION");
                    if (!bad_samples.isEmpty())
                        vcb.filter("BADSAMPLES");
                } else {
                    vcb.passFilters();
                }
                vcb.genotypes(genotypeList);
                vcfw.add(vcb.make());
                contaminationTable.clear();
            } else {
                final long now = System.currentTimeMillis();
                if (this.outputFile != null && this.save_every_sec > -1L && last_save_ms + (this.save_every_sec * 1000L) > now) {
                    saveToFile(contaminationTable);
                    last_save_ms = now;
                }
            }
        }
        progress.finish();
        if (this.output_as_vcf) {
            vcfw.close();
            vcfw = null;
        } else {
            saveToFile(contaminationTable);
        }
        return 0;
    } catch (final Exception e) {
        LOG.error(e);
        return -1;
    } finally {
        CloserUtil.close(vcfw);
        CloserUtil.close(in);
        CloserUtil.close(iter);
        for (SamReader samReader : sample2samReader.values()) CloserUtil.close(samReader);
        sample2samReader.clear();
    }
}

Example 77 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project jvarkit by lindenb.

the class Biostar173114 method doWork.

@Override
public int doWork(final List<String> args) {
    if (keepQualities)
        keepSequence = true;
    SamReader sfr = null;
    SAMFileWriter sfw = null;
    SAMRecordIterator iter = null;
    try {
        sfr = super.openSamReader(oneFileOrNull(args));
        sfw = this.writingBamArgs.openSAMFileWriter(this.outputFile, sfr.getFileHeader(), true);
        iter = sfr.iterator();
        final SAMSequenceDictionaryProgress progress = new SAMSequenceDictionaryProgress(sfr.getFileHeader()).logger(LOG);
        long nReads = 0;
        while (iter.hasNext()) {
            final SAMRecord record = progress.watch(iter.next());
            if (!this.keepAttributes) {
                final SAMReadGroupRecord g = record.getReadGroup();
                record.clearAttributes();
                if (g != null && this.keepReadGroup) {
                    record.setAttribute("RG", g.getId());
                }
            }
            record.setReadName(this.keepName ? record.getReadName() : "R" + Long.toHexString(nReads++));
            if (!this.keepMate && record.getReadPairedFlag()) {
                record.setReadPairedFlag(false);
                record.setMateReferenceIndex(SAMRecord.NO_ALIGNMENT_REFERENCE_INDEX);
                record.setMateAlignmentStart(SAMRecord.NO_ALIGNMENT_START);
                record.setMateUnmappedFlag(false);
                record.setMateNegativeStrandFlag(false);
                record.setInferredInsertSize(0);
                record.setProperPairFlag(false);
            }
            if (!this.keepCigar && !record.getReadUnmappedFlag() && record.getCigar() != null) {
                record.setCigar(new Cigar(record.getCigar().getCigarElements().stream().filter(C -> !C.getOperator().equals(CigarOperator.H)).collect(Collectors.toList())));
            }
            if (!this.keepSequence) {
                record.setReadBases(SAMRecord.NULL_SEQUENCE);
            }
            if (!this.keepQualities) {
                record.setBaseQualities(SAMRecord.NULL_QUALS);
            }
            sfw.addAlignment(record);
        }
        progress.finish();
        sfw.close();
        sfw = null;
        LOG.info("done");
        return RETURN_OK;
    } catch (final Exception err) {
        LOG.error(err);
        return -1;
    } finally {
        CloserUtil.close(iter);
        CloserUtil.close(sfr);
        CloserUtil.close(sfw);
    }
}

Example 78 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project jvarkit by lindenb.

the class Biostar214299 method doWork.

@Override
public int doWork(final List<String> args) {
    if (this.positionFile == null) {
        LOG.error("position File is not defined.");
        return -1;
    }
    final String UNAFFECTED_SAMPLE = "UNAFFECTED";
    final String AMBIGOUS_SAMPLE = "AMBIGOUS";
    final String UNMAPPED = "UNMAPPED";
    SamReader sfr = null;
    SAMFileWriter sfw = null;
    final IntervalTreeMap<Position> positionsTreeMap = new IntervalTreeMap<>();
    final Set<String> samples = new HashSet<>();
    try {
        sfr = openSamReader(oneFileOrNull(args));
        final SAMFileHeader header = sfr.getFileHeader();
        final SAMSequenceDictionary dict = header.getSequenceDictionary();
        if (dict == null) {
            LOG.error("Dictionary missing in input sam");
            return -1;
        }
        try (BufferedReader br = IOUtils.openFileForBufferedReading(this.positionFile)) {
            String line;
            while ((line = br.readLine()) != null) {
                if (line.trim().isEmpty() || line.startsWith("#"))
                    continue;
                final String[] tokens = line.split("[\t]");
                if (tokens.length < 4) {
                    LOG.error("Not enough columns in " + line);
                    return -1;
                }
                final String contig = tokens[0];
                if (dict.getSequence(contig) == null) {
                    LOG.error("No such contig in input's sam dictionary: " + contig);
                    return -1;
                }
                final int refpos = Integer.parseInt(tokens[1]);
                final Interval interval = new Interval(contig, refpos, refpos);
                Position position = positionsTreeMap.get(interval);
                if (position == null) {
                    position = new Position();
                    // position.contig = contig;
                    position.refpos = refpos;
                    positionsTreeMap.put(interval, position);
                }
                final String bases = tokens[2].toUpperCase();
                if (bases.length() != 1 || !bases.matches("[ATGC]")) {
                    LOG.error("in " + line + " bases should be one letter an ATGC");
                    return -1;
                }
                if (position.base2sample.containsKey(bases.charAt(0))) {
                    LOG.error("in " + line + " bases already defined for this position");
                    return -1;
                }
                final String sampleName = tokens[3].trim();
                if (sampleName.isEmpty()) {
                    LOG.error("sample name cannot be empty");
                    return -1;
                }
                samples.add(sampleName);
                position.base2sample.put(bases.charAt(0), sampleName);
            }
        } catch (final IOException err) {
            LOG.error(err);
            return -1;
        }
        if (samples.contains(UNAFFECTED_SAMPLE)) {
            LOG.error("Sample cannot be named " + UNAFFECTED_SAMPLE);
            return -1;
        }
        if (samples.contains(AMBIGOUS_SAMPLE)) {
            LOG.error("Sample cannot be named " + AMBIGOUS_SAMPLE);
            return -1;
        }
        if (samples.contains(UNMAPPED)) {
            LOG.error("Sample cannot be named " + UNMAPPED);
            return -1;
        }
        samples.add(UNAFFECTED_SAMPLE);
        samples.add(AMBIGOUS_SAMPLE);
        samples.add(UNMAPPED);
        final SAMFileHeader newHeader = new SAMFileHeader();
        newHeader.setSortOrder(header.getSortOrder());
        newHeader.setSequenceDictionary(dict);
        newHeader.addComment("generated with " + getProgramName() + " " + getVersion() + " Pierre Lindenbaum : " + getProgramCommandLine());
        /* create groups */
        for (final String sample : samples) {
            final SAMReadGroupRecord rg = new SAMReadGroupRecord(sample);
            rg.setSample(sample);
            rg.setLibrary(sample);
            newHeader.addReadGroup(rg);
        }
        sfw = this.writingBamArgs.openSAMFileWriter(this.outputFile, newHeader, true);
        final SAMSequenceDictionaryProgress progress = new SAMSequenceDictionaryProgress(header).logger(LOG);
        final SAMRecordIterator iter = sfr.iterator();
        while (iter.hasNext()) {
            final SAMRecord rec = progress.watch(iter.next());
            rec.setAttribute("RG", null);
            if (rec.getReadUnmappedFlag()) {
                rec.setAttribute("RG", UNMAPPED);
                sfw.addAlignment(rec);
                continue;
            }
            final Cigar cigar = rec.getCigar();
            final Collection<Position> snps = positionsTreeMap.getContained(new Interval(rec.getContig(), rec.getUnclippedStart(), rec.getUnclippedEnd()));
            if (snps == null || snps.isEmpty()) {
                rec.setAttribute("RG", UNAFFECTED_SAMPLE);
                sfw.addAlignment(rec);
                continue;
            }
            final Map<Integer, Position> index2pos = snps.stream().collect(Collectors.toMap(P -> P.refpos, P -> P));
            final Set<String> selectedSamples = new HashSet<>();
            final byte[] bases = rec.getReadBases();
            if (bases == null || bases.equals(SAMRecord.NULL_SEQUENCE)) {
                LOG.error("Bases missing in read " + rec);
                return -1;
            }
            int refPos1 = rec.getUnclippedStart();
            int readPos0 = 0;
            for (final CigarElement ce : cigar.getCigarElements()) {
                final CigarOperator op = ce.getOperator();
                final boolean consummeReadBaseOrSoftClip = op.consumesReadBases() || op.equals(CigarOperator.S);
                if (op.consumesReferenceBases() && consummeReadBaseOrSoftClip) {
                    for (int i = 0; i < ce.getLength(); ++i) {
                        final int nowRefPos1 = (refPos1 + i);
                        final int nowReadPos0 = (readPos0 + i);
                        final Position position = index2pos.get(nowRefPos1);
                        if (position == null)
                            continue;
                        if (nowReadPos0 >= bases.length)
                            continue;
                        final char base = (char) Character.toUpperCase(bases[nowReadPos0]);
                        final String sample = position.base2sample.get(base);
                        if (sample == null)
                            continue;
                        selectedSamples.add(sample);
                        index2pos.remove(nowRefPos1);
                        if (index2pos.isEmpty())
                            break;
                    }
                }
                if (op.consumesReferenceBases())
                    refPos1 += ce.getLength();
                if (consummeReadBaseOrSoftClip || op.equals(CigarOperator.H)) {
                    readPos0 += ce.getLength();
                }
            }
            if (selectedSamples.isEmpty()) {
                rec.setAttribute("RG", UNAFFECTED_SAMPLE);
            } else if (selectedSamples.size() == 1) {
                rec.setAttribute("RG", selectedSamples.iterator().next());
            } else {
                rec.setAttribute("RG", AMBIGOUS_SAMPLE);
            }
            sfw.addAlignment(rec);
        }
        progress.finish();
        LOG.info("done");
        return RETURN_OK;
    } catch (final Exception err) {
        LOG.error(err);
        return -1;
    } finally {
        CloserUtil.close(sfr);
        CloserUtil.close(sfw);
    }
}

Example 79 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project jvarkit by lindenb.

the class Biostar78400Test method test01.

@Test
public void test01() throws IOException {
    final String flowcell = "HS20001259127";
    final String lane = "1";
    final File in = createTmpFile(".bam");
    SAMFileHeader header = new SAMFileHeader();
    header.setSortOrder(SortOrder.unsorted);
    SAMFileWriter sfw = new SAMFileWriterFactory().makeBAMWriter(header, true, in);
    DefaultSAMRecordFactory recfactory = new DefaultSAMRecordFactory();
    SAMRecord rec = recfactory.createSAMRecord(header);
    rec.setReadName(flowcell + ":" + lane + ":1210:15640:52255");
    rec.setReadString("GAATTC");
    rec.setBaseQualityString("222222");
    SAMUtils.makeReadUnmapped(rec);
    sfw.addAlignment(rec);
    sfw.close();
    assertIsValidBam(in);
    final File xml = createTmpFile(".xml");
    PrintWriter pw = new PrintWriter(xml);
    pw.println("<?xml version=\"1.0\"?><read-groups>" + "<flowcell name=\"" + flowcell + "\"><lane index=\"" + lane + "\">" + "<group ID=\"X1\"><library>L1</library><platform>P1</platform>" + "<sample>S1</sample><platformunit>PU1</platformunit>" + "<center>C1</center><description>blabla</description></group>" + "</lane></flowcell><flowcell name=\"HS20001259128\">" + "<lane index=\"2\"><group ID=\"x2\"><library>L2</library>" + "<platform>P2</platform><sample>S2</sample><platformunit>PU1</platformunit>" + "<center>C1</center><description>blabla</description></group></lane>" + "</flowcell></read-groups>");
    pw.flush();
    pw.close();
    assertIsXml(xml);
    final File out = createTmpFile(".bam");
    Assert.assertEquals(new Biostar78400().instanceMain(newCmd().add("-o").add(out).add("-x").add(xml).add(in).make()), 0);
    SamReader r = SamReaderFactory.makeDefault().validationStringency(ValidationStringency.LENIENT).open(out);
    Assert.assertTrue(r.getFileHeader() != null);
    Assert.assertTrue(r.getFileHeader().getReadGroups() != null);
    Assert.assertFalse(r.getFileHeader().getReadGroups().isEmpty());
    SAMRecordIterator iter = r.iterator();
    Assert.assertTrue(iter.hasNext());
    rec = iter.next();
    SAMReadGroupRecord rg = rec.getReadGroup();
    Assert.assertNotNull(rg);
    Assert.assertEquals(rg.getId(), "X1");
    Assert.assertEquals(rg.getSample(), "S1");
    Assert.assertFalse(iter.hasNext());
    iter.close();
    r.close();
}

Example 80 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project jvarkit by lindenb.

the class BlastToSam method convertIterationToSequenceIteration.

private SequenceIteration convertIterationToSequenceIteration(final List<Iteration> stack, final SAMFileHeader header) throws XMLStreamException, JAXBException {
    final SequenceIteration sequenceIteration = new SequenceIteration();
    if (stack.isEmpty())
        return sequenceIteration;
    final SAMReadGroupRecord rg1 = header.getReadGroup("g1");
    // sequenceIteration.iteration=iter1;
    final SAMRecordFactory samRecordFactory = new DefaultSAMRecordFactory();
    final StringBuilder readContent = new StringBuilder();
    final int iterLength = Integer.parseInt(stack.get(0).getIterationQueryLen());
    for (final Iteration iter1 : stack) {
        for (final Hit hit : iter1.getIterationHits().getHit()) {
            for (final Hsp hsp : hit.getHitHsps().getHsp()) {
                for (final BlastHspAlignment.Align a : new BlastHspAlignment(hsp)) {
                    char c = a.getQueryChar();
                    if (!Character.isLetter(c))
                        continue;
                    final int queryIndex0 = a.getQueryIndex1() - 1;
                    while (readContent.length() <= queryIndex0) readContent.append('N');
                    if (readContent.charAt(queryIndex0) == 'N') {
                        readContent.setCharAt(queryIndex0, c);
                    } else if (readContent.charAt(queryIndex0) != c) {
                        throw new IllegalStateException("Expected character '" + readContent.charAt(queryIndex0) + "' but got '" + c + "' at " + queryIndex0 + "\n" + hsp.getHspQseq() + "\n" + hsp.getHspMidline() + "\n" + hsp.getHspHseq() + "\n" + readContent + "\n");
                    }
                }
            }
        }
    }
    for (Iteration iter1 : stack) {
        for (Hit hit : iter1.getIterationHits().getHit()) {
            for (Hsp hsp : hit.getHitHsps().getHsp()) {
                SAMRecord rec = samRecordFactory.createSAMRecord(header);
                rec.setReadUnmappedFlag(false);
                rec.setReadName(iter1.getIterationQueryDef());
                if (hit.getHitAccession() != null && !hit.getHitAccession().trim().isEmpty() && this.dictionary.getSequence(hit.getHitAccession()) != null) {
                    rec.setReferenceName(hit.getHitAccession());
                } else {
                    rec.setReferenceName(hit.getHitDef());
                }
                final SAMSequenceRecord ssr = this.dictionary.getSequence(hit.getHitDef());
                if (ssr == null) {
                    LOG.warn("Hit is not in SAMDictionary " + hit.getHitDef());
                    rec.setReferenceIndex(-1);
                } else {
                    rec.setReferenceIndex(ssr.getSequenceIndex());
                }
                final BlastHspAlignment blastHspAlignment = new BlastHspAlignment(hsp);
                rec.setReadNegativeStrandFlag(blastHspAlignment.isPlusMinus());
                final List<CigarOperator> cigarL = new ArrayList<CigarOperator>();
                for (BlastHspAlignment.Align a : blastHspAlignment) {
                    // System.err.println("##"+a);
                    if (a.getMidChar() == '|') {
                        cigarL.add(CigarOperator.EQ);
                    } else if (a.getMidChar() == ':') {
                        cigarL.add(CigarOperator.M);
                    } else if (a.getHitChar() == '-') {
                        cigarL.add(CigarOperator.I);
                    } else if (a.getQueryChar() == '-') {
                        cigarL.add(CigarOperator.D);
                    } else {
                        cigarL.add(CigarOperator.X);
                    }
                }
                if (cigarL.size() != hsp.getHspMidline().length()) {
                    throw new IllegalStateException("Boumm");
                }
                Cigar cigarE = new Cigar();
                if (blastHspAlignment.getQueryFrom1() > 1) {
                    cigarE.add(new CigarElement(blastHspAlignment.getQueryFrom1() - 1, CigarOperator.S));
                }
                int x = 0;
                while (x < cigarL.size()) {
                    int y = x + 1;
                    while (y < cigarL.size() && cigarL.get(x) == cigarL.get(y)) {
                        ++y;
                    }
                    cigarE.add(new CigarElement(y - x, cigarL.get(x)));
                    x = y;
                }
                /* soft clip */
                if (blastHspAlignment.getQueryTo1() < readContent.length()) {
                    cigarE.add(new CigarElement((readContent.length() - blastHspAlignment.getQueryTo1()), CigarOperator.S));
                }
                /* hard clip */
                if (readContent.length() < iterLength) {
                    cigarE.add(new CigarElement((iterLength - readContent.length()), CigarOperator.H));
                }
                rec.setCigar(cigarE);
                rec.setMappingQuality(40);
                rec.setAlignmentStart(Math.min(blastHspAlignment.getHitFrom1(), blastHspAlignment.getHitTo1()));
                rec.setAttribute("BB", Float.parseFloat(hsp.getHspBitScore()));
                rec.setAttribute("BE", Float.parseFloat(hsp.getHspEvalue()));
                rec.setAttribute("BS", Float.parseFloat(hsp.getHspScore()));
                rec.setAttribute("NM", Integer.parseInt(hsp.getHspGaps()));
                rec.setAttribute("RG", rg1.getId());
                // setAlignmentEnd not supported in SAM API
                // rec.setAlignmentEnd(Math.max(blastHspAlignment.getHitFrom1(),blastHspAlignment.getHitTo1()));
                sequenceIteration.records.add(rec);
            }
        }
    }
    if (readContent.length() == 0) {
        readContent.append('N');
    }
    byte[] readBases = readContent.toString().getBytes();
    char[] readQuals = new char[readBases.length];
    for (int i = 0; i < readBases.length; ++i) {
        readQuals[i] = (readBases[i] == 'N' ? '#' : 'J');
    }
    if (sequenceIteration.records.isEmpty()) {
        SAMRecord rec = samRecordFactory.createSAMRecord(header);
        rec.setReadName(stack.get(0).getIterationQueryDef());
        rec.setReadUnmappedFlag(true);
        rec.setAttribute("RG", rg1.getId());
        sequenceIteration.records.add(rec);
    }
    for (SAMRecord rec : sequenceIteration.records) {
        rec.setReadString(new String(readBases));
        rec.setReadBases(readBases);
        rec.setBaseQualityString(new String(readQuals, 0, readQuals.length));
        rec.setBaseQualities(htsjdk.samtools.SAMUtils.fastqToPhred(new String(readQuals, 0, readQuals.length)));
    }
    return sequenceIteration;
}