Ejemplo n.º 1
0
  private VariantCallContext generateEmptyContext(
      RefMetaDataTracker tracker,
      ReferenceContext ref,
      Map<String, AlignmentContext> stratifiedContexts,
      AlignmentContext rawContext) {
    VariantContext vc;
    if (UAC.GenotypingMode
        == GenotypeLikelihoodsCalculationModel.GENOTYPING_MODE.GENOTYPE_GIVEN_ALLELES) {
      VariantContext vcInput =
          UnifiedGenotyperEngine.getVCFromAllelesRod(
              tracker, ref, rawContext.getLocation(), false, logger, UAC.alleles);
      if (vcInput == null) return null;
      vc =
          new VariantContextBuilder(
                  "UG_call",
                  ref.getLocus().getContig(),
                  vcInput.getStart(),
                  vcInput.getEnd(),
                  vcInput.getAlleles())
              .make();
    } else {
      // deal with bad/non-standard reference bases
      if (!Allele.acceptableAlleleBases(new byte[] {ref.getBase()})) return null;

      Set<Allele> alleles = new HashSet<Allele>();
      alleles.add(Allele.create(ref.getBase(), true));
      vc =
          new VariantContextBuilder(
                  "UG_call",
                  ref.getLocus().getContig(),
                  ref.getLocus().getStart(),
                  ref.getLocus().getStart(),
                  alleles)
              .make();
    }

    if (annotationEngine != null) {
      // Note: we want to use the *unfiltered* and *unBAQed* context for the annotations
      final ReadBackedPileup pileup = rawContext.getBasePileup();
      stratifiedContexts = AlignmentContextUtils.splitContextBySampleName(pileup);

      vc = annotationEngine.annotateContext(tracker, ref, stratifiedContexts, vc);
    }

    return new VariantCallContext(vc, false);
  }
  private ArrayList<Allele> computeConsensusAlleles(
      ReferenceContext ref,
      Map<String, AlignmentContext> contexts,
      AlignmentContextUtils.ReadOrientation contextType) {
    Allele refAllele = null, altAllele = null;
    GenomeLoc loc = ref.getLocus();
    ArrayList<Allele> aList = new ArrayList<Allele>();

    HashMap<String, Integer> consensusIndelStrings = new HashMap<String, Integer>();

    int insCount = 0, delCount = 0;
    // quick check of total number of indels in pileup
    for (Map.Entry<String, AlignmentContext> sample : contexts.entrySet()) {
      AlignmentContext context = AlignmentContextUtils.stratify(sample.getValue(), contextType);

      final ReadBackedExtendedEventPileup indelPileup = context.getExtendedEventPileup();
      insCount += indelPileup.getNumberOfInsertions();
      delCount += indelPileup.getNumberOfDeletions();
    }

    if (insCount < minIndelCountForGenotyping && delCount < minIndelCountForGenotyping)
      return aList;

    for (Map.Entry<String, AlignmentContext> sample : contexts.entrySet()) {
      // todo -- warning, can be duplicating expensive partition here
      AlignmentContext context = AlignmentContextUtils.stratify(sample.getValue(), contextType);

      final ReadBackedExtendedEventPileup indelPileup = context.getExtendedEventPileup();

      for (ExtendedEventPileupElement p : indelPileup.toExtendedIterable()) {
        // SAMRecord read = p.getRead();
        GATKSAMRecord read = ReadUtils.hardClipAdaptorSequence(p.getRead());
        if (read == null) continue;
        if (ReadUtils.is454Read(read)) {
          continue;
        }

        /*                if (DEBUG && p.isIndel()) {
                         System.out.format("Read: %s, cigar: %s, aln start: %d, aln end: %d, p.len:%d, Type:%s, EventBases:%s\n",
                                 read.getReadName(),read.getCigar().toString(),read.getAlignmentStart(),read.getAlignmentEnd(),
                                 p.getEventLength(),p.getType().toString(), p.getEventBases());
                     }
        */

        String indelString = p.getEventBases();
        if (p.isInsertion()) {
          boolean foundKey = false;
          if (read.getAlignmentEnd() == loc.getStart()) {
            // first corner condition: a read has an insertion at the end, and we're right at the
            // insertion.
            // In this case, the read could have any of the inserted bases and we need to build a
            // consensus
            for (String s : consensusIndelStrings.keySet()) {
              int cnt = consensusIndelStrings.get(s);
              if (s.startsWith(indelString)) {
                // case 1: current insertion is prefix of indel in hash map
                consensusIndelStrings.put(s, cnt + 1);
                foundKey = true;
                break;
              } else if (indelString.startsWith(s)) {
                // case 2: indel stored in hash table is prefix of current insertion
                // In this case, new bases are new key.
                consensusIndelStrings.remove(s);
                consensusIndelStrings.put(indelString, cnt + 1);
                foundKey = true;
                break;
              }
            }
            if (!foundKey)
              // none of the above: event bases not supported by previous table, so add new key
              consensusIndelStrings.put(indelString, 1);

          } else if (read.getAlignmentStart() == loc.getStart() + 1) {
            // opposite corner condition: read will start at current locus with an insertion
            for (String s : consensusIndelStrings.keySet()) {
              int cnt = consensusIndelStrings.get(s);
              if (s.endsWith(indelString)) {
                // case 1: current insertion is suffix of indel in hash map
                consensusIndelStrings.put(s, cnt + 1);
                foundKey = true;
                break;
              } else if (indelString.endsWith(s)) {
                // case 2: indel stored in hash table is suffix of current insertion
                // In this case, new bases are new key.

                consensusIndelStrings.remove(s);
                consensusIndelStrings.put(indelString, cnt + 1);
                foundKey = true;
                break;
              }
            }
            if (!foundKey)
              // none of the above: event bases not supported by previous table, so add new key
              consensusIndelStrings.put(indelString, 1);

          } else {
            // normal case: insertion somewhere in the middle of a read: add count to hash map
            int cnt =
                consensusIndelStrings.containsKey(indelString)
                    ? consensusIndelStrings.get(indelString)
                    : 0;
            consensusIndelStrings.put(indelString, cnt + 1);
          }

        } else if (p.isDeletion()) {
          indelString = String.format("D%d", p.getEventLength());
          int cnt =
              consensusIndelStrings.containsKey(indelString)
                  ? consensusIndelStrings.get(indelString)
                  : 0;
          consensusIndelStrings.put(indelString, cnt + 1);
        }
      }

      /*            if (DEBUG) {
          int icount = indelPileup.getNumberOfInsertions();
          int dcount = indelPileup.getNumberOfDeletions();
          if (icount + dcount > 0)
          {
              List<Pair<String,Integer>> eventStrings = indelPileup.getEventStringsWithCounts(ref.getBases());
              System.out.format("#ins: %d, #del:%d\n", insCount, delCount);

              for (int i=0 ; i < eventStrings.size() ; i++ ) {
                  System.out.format("%s:%d,",eventStrings.get(i).first,eventStrings.get(i).second);
                  //                int k=0;
              }
              System.out.println();
          }
      }             */
    }

    int maxAlleleCnt = 0;
    String bestAltAllele = "";
    for (String s : consensusIndelStrings.keySet()) {
      int curCnt = consensusIndelStrings.get(s);
      if (curCnt > maxAlleleCnt) {
        maxAlleleCnt = curCnt;
        bestAltAllele = s;
      }
      //            if (DEBUG)
      //                System.out.format("Key:%s, number: %d\n",s,consensusIndelStrings.get(s)  );
    } // gdebug-

    if (maxAlleleCnt < minIndelCountForGenotyping) return aList;

    if (bestAltAllele.startsWith("D")) {
      // get deletion length
      int dLen = Integer.valueOf(bestAltAllele.substring(1));
      // get ref bases of accurate deletion
      int startIdxInReference = (int) (1 + loc.getStart() - ref.getWindow().getStart());

      // System.out.println(new String(ref.getBases()));
      byte[] refBases =
          Arrays.copyOfRange(ref.getBases(), startIdxInReference, startIdxInReference + dLen);

      if (Allele.acceptableAlleleBases(refBases)) {
        refAllele = Allele.create(refBases, true);
        altAllele = Allele.create(Allele.NULL_ALLELE_STRING, false);
      }
    } else {
      // insertion case
      if (Allele.acceptableAlleleBases(bestAltAllele)) {
        refAllele = Allele.create(Allele.NULL_ALLELE_STRING, true);
        altAllele = Allele.create(bestAltAllele, false);
      }
    }
    if (refAllele != null && altAllele != null) {
      aList.add(0, refAllele);
      aList.add(1, altAllele);
    }
    return aList;
  }