C++ bam_record类代码示例

static
void
addReadToDepthEst(
    const bam_record& bamRead,
    const pos_t beginPos,
    std::vector<unsigned>& depth)
{
    using namespace ALIGNPATH;

    const pos_t endPos(beginPos+depth.size());

    // get cigar:
    path_t apath;
    bam_cigar_to_apath(bamRead.raw_cigar(), bamRead.n_cigar(), apath);

    pos_t refPos(bamRead.pos()-1);
    BOOST_FOREACH(const path_segment& ps, apath)
    {
        if (refPos>=endPos) return;

        if (MATCH == ps.type)
        {
            for (pos_t pos(refPos); pos < (refPos+static_cast<pos_t>(ps.length)); ++pos)
            {
                if (pos>=beginPos)
                {
                    if (pos>=endPos) return;
                    depth[pos-beginPos]++;
                }
            }
        }
        if (is_segment_type_ref_length(ps.type)) refPos += ps.length;
    }
}

void
SVScorePairRefProcessor::
processClearedRecord(
    const bam_record& bamRead)
{
    using namespace illumina::common;

    assert(bamParams.isSet);

    const pos_t refPos(bamRead.pos()-1);
    if (! bamParams.interval.range.is_pos_intersect(refPos)) return;

    const bool isLargeInsert(isLargeInsertSV(sv));

#ifdef DEBUG_MEGAPAIR
    log_os << __FUNCTION__ << ": read: " << bamRead << "\n";
#endif

    /// check if fragment is too big or too small:
    const int templateSize(std::abs(bamRead.template_size()));
    if (templateSize < bamParams.minFrag) return;
    if (templateSize > bamParams.maxFrag) return;

    // count only from the down stream reads
    const bool isFirstBamRead(isFirstRead(bamRead));

    // get fragment range:
    pos_t fragBeginRefPos(refPos);
    if (! isFirstBamRead)
    {
        fragBeginRefPos=bamRead.mate_pos()-1;
    }

    const pos_t fragEndRefPos(fragBeginRefPos+templateSize);

    if (fragBeginRefPos > fragEndRefPos)
    {
        std::ostringstream oss;
        oss << "ERROR: Failed to parse fragment range from bam record. Frag begin,end: " << fragBeginRefPos << " " << fragEndRefPos << " bamRecord: " << bamRead << "\n";
        BOOST_THROW_EXCEPTION(LogicException(oss.str()));
    }

    {
        const pos_t fragOverlap(std::min((1+svParams.centerPos-fragBeginRefPos), (fragEndRefPos-svParams.centerPos)));
#ifdef DEBUG_MEGAPAIR
        log_os << __FUNCTION__ << ": frag begin/end/overlap: " << fragBeginRefPos << " " << fragEndRefPos << " " << fragOverlap << "\n";
#endif
        if (fragOverlap < pairOpt.minFragSupport) return;
    }

    SVFragmentEvidence& fragment(evidence.getSampleEvidence(bamParams.bamIndex)[bamRead.qname()]);

    static const bool isShadow(false);

    SVFragmentEvidenceRead& evRead(fragment.getRead(bamRead.is_first()));
    setReadEvidence(svParams.minMapQ, svParams.minTier2MapQ, bamRead, isShadow, evRead);

    setAlleleFrag(*bamParams.fragDistroPtr, templateSize, fragment.ref.getBp(isBp1),isLargeInsert);
}

bool
isMateInsertionEvidenceCandidate2(
    const bam_record& bamRead,
    const bool isSearchForLeftOpen,
    const bool isSearchForRightOpen)
{
    if ((! isSearchForLeftOpen) && (! bamRead.is_fwd_strand())) return false;
    if ((! isSearchForRightOpen) && bamRead.is_fwd_strand()) return false;
    return true;
}

    void
    addRead(
        const bam_record& bamRead)
    {
        if (_isRegionInit)
        {
            if (bamRead.pos() > _endPos + 1000)
            {
                _maxPos=_endPos;
                setNewRegion();
            }
        }

        if (! _isRegionInit)
        {
            _minPos=bamRead.pos();
            _maxPos=bamRead.pos();
            _endPos=bamRead.pos() + bamRead.read_size();
            _isRegionInit=true;
        }
        else
        {
            if (bamRead.pos() > _maxPos)
            {
                _maxPos = bamRead.pos();
                _endPos=bamRead.pos() + bamRead.read_size();
            }
        }

        _count++;
        _totalReadLength += bamRead.read_size();
    }

FragmentSizeType::index_t
SVLocusScanner::
_getFragmentSizeType(
    const bam_record& bamRead,
    const unsigned defaultReadGroupIndex) const
{
    using namespace FragmentSizeType;
    if (bamRead.target_id() != bamRead.mate_target_id()) return DISTANT;
    const int32_t fragmentSize(std::abs(bamRead.template_size()));
    return classifySize(_stats[defaultReadGroupIndex], fragmentSize);
}

static
void
getSingleReadSVCandidates(
    const ReadScannerOptions& opt,
    const ReadScannerDerivOptions& dopt,
    const bam_record& localRead,
    const SimpleAlignment& localAlign,
    const chromMap_t& chromToIndex,
    const reference_contig_segment& refSeq,
    std::vector<SVObservation>& candidates)
{
    using namespace illumina::common;

    const bool isRead2(localRead.is_paired() && (localRead.read_no() == 2));
    const FRAGSOURCE::index_t fragSource(isRead2 ? FRAGSOURCE::READ2 : FRAGSOURCE::READ1);

    // - process any large indels in the localRead:
    getSVCandidatesFromReadIndels(opt, dopt, localAlign, fragSource, candidates);
#ifdef DEBUG_SCANNER
    log_os << __FUNCTION__ << ": post-indels candidate_size: " << candidates.size() << "\n";
#endif

    // a read can provide SA split evidence or semi-aligned/soft-clip, but not both.
    // this prevents split reads from triggering spurious local assembles. It is
    // possible for a read to genuinely contain evidence of both, but this should
    // be very rare.
    if (localRead.isSASplit())
    {
        getSACandidatesFromRead(opt, dopt, localRead, localAlign, fragSource, chromToIndex,
                                candidates);
#ifdef DEBUG_SCANNER
        log_os << __FUNCTION__ << ": post-split read candidate_size: " << candidates.size() << "\n";
#endif
    }
    else
    {
        if (dopt.isSmallCandidates)
        {
            getSVCandidatesFromSemiAligned(opt, dopt, localRead, localAlign, fragSource, refSeq,
                                           candidates);
        }
#ifdef DEBUG_SCANNER
        log_os << __FUNCTION__ << ": post-semialigned candidate_size: " << candidates.size() << "\n";
#endif
    }
}

void
addSupplementaryAlignmentEvidence(
    bam_record& bamRead,
    const std::string& svStr)
{
    static const char svtag[] = {'S','A'};
    bam_aux_append(bamRead.get_data(),svtag,'Z',(svStr.size()+1),
                   (const uint8_t*)(svStr.c_str()));
}

/// get SV candidates from SA-tag split-read alignment
static
SVObservation
GetSplitSACandidate(
    const ReadScannerDerivOptions& dopt,
    const bam_record& localRead,
    const SimpleAlignment& localAlign,
    const SimpleAlignment& remoteAlign,
    const FRAGSOURCE::index_t fragSource)
{
    using namespace SVEvidenceType;
    static const index_t svSource(SPLIT_ALIGN);

    SVObservation sv;
    sv.evtype = svSource;
    sv.fragSource = fragSource;

    SVBreakend& localBreakend(sv.bp1);
    SVBreakend& remoteBreakend(sv.bp2);

    // use single-side evidence, have to read the supp read to get the
    // reverse edge. this protects against double-count:
    localBreakend.lowresEvidence.add(svSource);

    updateSABreakend(dopt, localAlign, localBreakend);
    updateSABreakend(dopt, remoteAlign, remoteBreakend);

    // If the local (bp1) alignment is split downstream (on the right side) then this read goes from bp1 -> bp2.
    // If it is a forward read (e.g. read1 on + strand), this means it's a forward read for this event.
    const bool isSplitDownstream(isSplitOpenDownstream(localAlign.path));
    const bool isReadFw = (localRead.is_first() == localRead.is_fwd_strand());
    if (dopt.isStranded)
    {
        if (isReadFw == isSplitDownstream)
        {
            sv.fwReads += 1;
        }
        else
        {
            sv.rvReads += 1;
        }
    }
    return sv;
}

/// extract poorly aligned read ends (semi-aligned and/or soft-clipped)
/// to internal candidate format
static
void
getSVCandidatesFromSemiAligned(
    const ReadScannerOptions& opt,
    const ReadScannerDerivOptions& dopt,
    const bam_record& bamRead,
    const SimpleAlignment& bamAlign,
    const FRAGSOURCE::index_t fragSource,
    const reference_contig_segment& refSeq,
    std::vector<SVObservation>& candidates)
{
    unsigned leadingMismatchLen(0);
    unsigned trailingMismatchLen(0);
    pos_t leadingRefPos(0), trailingRefPos(0);
    getSVBreakendCandidateSemiAligned(bamRead, bamAlign, refSeq,
                                      dopt.isUseOverlappingPairs,
                                      leadingMismatchLen, leadingRefPos,
                                      trailingMismatchLen, trailingRefPos);

    if ((leadingMismatchLen + trailingMismatchLen) >= bamRead.read_size()) return;

    using namespace SVEvidenceType;
    static const index_t svSource(SEMIALIGN);

    // semi-aligned reads don't define a full hypothesis, so they're always evidence for a 'complex' ie. undefined, event
    // in a fashion analogous to clipped reads
    static const bool isComplex(true);

    if (leadingMismatchLen >= opt.minSemiAlignedMismatchLen)
    {
        const pos_t pos(leadingRefPos);
        candidates.push_back(GetSplitSVCandidate(dopt,bamRead.target_id(),pos,pos,svSource, fragSource,isComplex));
    }

    if (trailingMismatchLen >= opt.minSemiAlignedMismatchLen)
    {
        const pos_t pos(trailingRefPos);
        candidates.push_back(GetSplitSVCandidate(dopt,bamRead.target_id(),pos,pos,svSource, fragSource,isComplex));
    }
}

bool
SVLocusScanner::
isProperPair(
    const bam_record& bamRead,
    const unsigned defaultReadGroupIndex) const
{
    if (! is_innie_pair(bamRead)) return false;

    const Range& ppr(_stats[defaultReadGroupIndex].properPair);
    const int32_t fragmentSize(std::abs(bamRead.template_size()));

    // we're seeing way to much large fragment garbage in cancers to use
    // vanilla proper pair criteria, push the max fragment size out a bit for now:
    static const float maxAnomFactor(1.5);
    if (fragmentSize > static_cast<int32_t>(maxAnomFactor*ppr.max)) return false;
    if (fragmentSize < ppr.min) return false;

    return true;
}

bool
SVLocusScanner::
isSVEvidence(
    const bam_record& bamRead,
    const unsigned defaultReadGroupIndex,
    const reference_contig_segment& refSeq,
    SVLocusEvidenceCount* incountsPtr) const
{
    // exclude innie read pairs which are anomalously short:
    const bool isAnom(isNonCompressedAnomalous(bamRead,defaultReadGroupIndex));
    const bool isSplit(bamRead.isSASplit());
    getAlignment(bamRead,_bamAlign);
    const bool isIndel(isLocalIndelEvidence(_bamAlign));
    const bool isAssm((_dopt.isSmallCandidates) && ((!isSplit) && isSemiAlignedEvidence(bamRead, _bamAlign, refSeq)));

    const bool isEvidence(isAnom || isSplit || isIndel || isAssm);

    if (nullptr != incountsPtr)
    {
        SVLocusEvidenceCount& incounts(*incountsPtr);
        incounts.total++;
        if (isAnom) incounts.anom++;
        if (isSplit) incounts.split++;
        if (isIndel) incounts.indel++;
        if (isAssm) incounts.assm++;

        if (! isEvidence) incounts.ignored++;

        if (isAnom)
        {
            if (isMateInsertionEvidenceCandidate(bamRead, getMinMapQ()))
            {
                // these counts are used to generate background noise rates in later candidate generation stages:
                incounts.remoteRecoveryCandidates++;
            }
        }
    }

    return isEvidence;
}

static
bool
isGoodShadow(
    const bam_record& bamRead,
    const std::string& lastQname)
{
#ifdef DEBUG_IS_SHADOW
    static const std::string logtag("isGoodShadow");
#endif

    if (! bamRead.is_paired()) return false;

    if (bamRead.isNonStrictSupplement()) return false;

    // sanity check that this is a shadow read:
    if (!bamRead.is_unmapped()) return false;
    if (bamRead.is_mate_unmapped()) return false;

    static const unsigned minAvgQualShadow = 25;
    if (get_avg_quality(bamRead) < minAvgQualShadow)
    {
        return false;
    }

    if (strcmp(bamRead.qname(),lastQname.c_str()) != 0)
    {
        // something went wrong here, shadows should have their singleton partner
        // preceding them in the BAM file.
#ifdef DEBUG_IS_SHADOW
        log_os << logtag << " ERROR: Shadow without matching singleton : " << bamRead.qname() << " vs " << lastQname << std::endl;
#endif
        return false;
    }

#ifdef DEBUG_IS_SHADOW
    log_os << logtag << " Found shadow!\n";
            << logtag << " this mapq  = " << ((unsigned int)bamRead.map_qual()) << std::endl;

/// get SV candidates from anomalous read pairs
static
void
getSVCandidatesFromPair(
    const ReadScannerOptions& opt,
    const ReadScannerDerivOptions& dopt,
    const SVLocusScanner::CachedReadGroupStats& rstats,
    const bam_record& localRead,
    const SimpleAlignment& localAlign,
    const bam_record* remoteReadPtr,
    std::vector<SVObservation>& candidates)
{
    if (! localRead.is_paired()) return;

    // don't count paired end evidence from SA-split reads twice:
    if (localRead.isNonStrictSupplement()) return;

    if (localRead.is_unmapped() || localRead.is_mate_unmapped()) return;

    // special case typically used for RNA-Seq analysis:
    if (opt.isIgnoreAnomProperPair && localRead.is_proper_pair()) return;

    // abstract remote alignment to SimpleAlignment object:
    const bool isRemote(nullptr != remoteReadPtr);
    const SimpleAlignment remoteAlign(isRemote ? getAlignment(*remoteReadPtr) : getFakeMateAlignment(localRead));

    AlignmentPairAnalyzer pairInspector(opt, dopt, rstats);
    pairInspector.reset(localAlign, remoteAlign, isRemote, localRead.is_first());

    if (! pairInspector.computeLargeEventRegionScale()) return;

    candidates.emplace_back();
    pairInspector.getSVObservation(candidates.back());

#ifdef DEBUG_SCANNER
    log_os << __FUNCTION__ << " evaluating pair sv for inclusion: " << candidates.back() << "\n";
#endif
}

void
buildTestBamRecord(
    bam_record& bamRead,
    int targetID,
    int pos,
    int mateTargetID,
    int matePos,
    int fragmentSize,
    int mapQ,
    std::string cigarString,
    std::string querySeq)
{
    bam1_t& bamData(*(bamRead.get_data()));

    // set qname
    {
        edit_bam_qname("buildTestBamRecord", bamData);
    }

    // set CIGAR
    {
        if (cigarString.empty())
        {
            cigarString = std::to_string(fragmentSize) + "M";
        }

        ALIGNPATH::path_t inputPath;
        cigar_to_apath(cigarString.c_str(), inputPath);
        edit_bam_cigar(inputPath, bamData);
    }

    // set read and qual
    {
        if ( querySeq.empty() )
        {
            querySeq = std::string(fragmentSize,'A');
        }
        const unsigned querySize(querySeq.length());
        // initialize test qual array to all Q30's:
        std::unique_ptr<uint8_t[]> qual(new uint8_t[querySize]);
        for (unsigned i(0); i<querySize; ++i)
        {
            qual[i] = 30;
        }
        edit_bam_read_and_quality(querySeq.c_str(), qual.get(), bamData);
    }

    // Set some defaults for the read
    bamRead.toggle_is_paired();
    bamRead.toggle_is_mate_fwd_strand();
    bamData.core.pos = pos;
    bamData.core.isize = fragmentSize;
    bamData.core.qual = mapQ;
    bamRead.set_target_id(targetID);

    // Set mate info
    bamData.core.mtid = mateTargetID;
    bamData.core.mpos = matePos;

    static const char nhTag[] = {'N','H'};
    static const char nmTag[] = {'N','M'};
    static const char rgTag[] = {'R','G'};
    bam_aux_append_unsigned(bamData, nhTag, 1);
    bam_aux_append_unsigned(bamData, nmTag, 1);
    bam_aux_append_unsigned(bamData, rgTag, 1);
}

/// get SV candidates from shadow/singleton pairs
/// look for singletons, create candidateSV around conf. interval of shadow position
/// cache singletons? might be needed to remove poor quality shadows.
/// should be able to re-use code, follow soft-clipping example.
static
void
getSVCandidatesFromShadow(
    const ReadScannerOptions& opt,
    const SVLocusScanner::CachedReadGroupStats& rstats,
    const bam_record& localRead,
    const SimpleAlignment& localAlign,
    const bam_record* remoteReadPtr,
    TrackedCandidates& candidates)
{
    using namespace SVEvidenceType;
    static const index_t svSource(SHADOW);

    static const bool isComplex(true);
    pos_t singletonGenomePos(0);
    int targetId(0);
    if (NULL == remoteReadPtr)
    {
        if (!localRead.is_unmapped()) return;
        // need to take care of this case
        // need to rely on cached mapq and qname
        return;
        if (!isGoodShadow(localRead,lastMapq,lastQname,opt.minSingletonMapqGraph))
        {
            return;
        }
        singletonGenomePos = localAlign.pos;
        targetId           = localRead.target_id();
    }
    else
    {
        // have both reads, straightforward from here
        const bam_record& remoteRead(*remoteReadPtr);
        const SimpleAlignment remoteAlign(remoteRead);

        if (localRead.is_mate_unmapped())
        {
            // remote read is shadow candidate
            if (!isGoodShadow(remoteRead,localRead.map_qual(),localRead.qname(),opt.minSingletonMapqGraph))
            {
                return;
            }
            singletonGenomePos = localAlign.pos;
            targetId = remoteRead.target_id();
        }
        else if (localRead.is_unmapped())
        {
            // local is shadow candidate
            if (!isGoodShadow(localRead,remoteRead.map_qual(),remoteRead.qname(),opt.minSingletonMapqGraph))
            {
                return;
            }
            singletonGenomePos = remoteAlign.pos;
            targetId = localRead.target_id();
        }
        else
        {
            // none unmapped, skip this one
            return;
        }
    }
    const pos_t properPairRangeOffset = static_cast<pos_t>(rstats.properPair.min + (rstats.properPair.max-rstats.properPair.min)/2);
    const pos_t shadowGenomePos = singletonGenomePos + properPairRangeOffset;
    candidates.push_back(GetSplitSVCandidate(opt,targetId,shadowGenomePos,shadowGenomePos, svSource, isComplex));
}