// NOTE: We have 2 different sets of offsets because the copied, new file differs in size than the existing one.
//
// Unsure which combination of write parameters were used on the original. Things like thread count,
// compression level, etc. can effect compression ratio, BGZF block sizes, etc. even though the BAM record
// content itself is equal. So we'll just track these index values separately, for now at least.
//
static
PbiRawData Test2Bam_ExistingIndex(void)
{
PbiRawData index = Test2Bam_CoreIndexData();
index.BasicData().fileOffset_ = { 33816576, 33825163, 33831333, 33834264, 33836542, 33838065, 33849818, 33863499, 33874621, 1392836608 };
return index;
}
static
PbiRawData Test2Bam_NewIndex(void)
{
PbiRawData index = Test2Bam_CoreIndexData();
index.BasicData().fileOffset_ = { 33816576, 236126208, 391315456, 469106688, 537067520, 587792384, 867303424, 1182793728, 1449787392, 1582628864 };
return index;
}
bool Accepts(const PbiRawData& idx, const size_t row) const
{
const auto& basicData = idx.BasicData();
// see if row's RGID known
const auto& rgId = basicData.rgId_.at(row);
const auto rgFound = lookup_.find(rgId);
if (rgFound == lookup_.end())
return false;
// see if row's ZMW known
const auto& zmwPtr = rgFound->second;
const auto zmw = basicData.holeNumber_.at(row);
const auto zmwFound = zmwPtr->find(zmw);
if (zmwFound == zmwPtr->end())
return false;
// see if row's QueryStart/QueryEnd known
// CCS names already covered in lookup construction phase
const auto& queryIntervals = zmwFound->second;
const auto qStart = basicData.qStart_.at(row);
const auto qEnd = basicData.qEnd_.at(row);
const auto queryInterval = std::make_pair(qStart, qEnd);
return queryIntervals.find(queryInterval) != queryIntervals.end();
}
bool PbiQueryLengthFilter::Accepts(const PbiRawData& idx, const size_t row) const
{
const auto& basicData = idx.BasicData();
const auto& qStart = basicData.qStart_.at(row);
const auto& qEnd = basicData.qEnd_.at(row);
const auto readLength = qEnd - qStart;
return CompareHelper(readLength);
}
bool PbiAlignedLengthFilter::Accepts(const PbiRawData& idx, const size_t row) const
{
const auto& mappedData = idx.MappedData();
const auto& aEnd = mappedData.aEnd_.at(row) ;
const auto& aStart = mappedData.aStart_.at(row);
const auto aLength = aEnd - aStart;
return CompareHelper(aLength);
}
PbiIndexPrivate::PbiIndexPrivate(const PbiRawData& rawIndex)
: filename_(rawIndex.Filename())
, version_(rawIndex.Version())
, sections_(rawIndex.FileSections())
, numReads_(rawIndex.NumReads())
, basicData_(rawIndex.BasicData())
, mappedData_(rawIndex.MappedData())
, referenceData_(rawIndex.ReferenceData())
, barcodeData_(rawIndex.BarcodeData())
{ }
bool PbiIdentityFilter::Accepts(const PbiRawData& idx, const size_t row) const
{
const auto& mappedData = idx.MappedData();
const auto& nMM = mappedData.nMM_.at(row);
const auto& nIndels = mappedData.NumDeletedAndInsertedBasesAt(row);
const auto& nDel = nIndels.first;
const auto& nIns = nIndels.second;
const auto& basicData = idx.BasicData();
const auto& qStart = basicData.qStart_.at(row);
const auto& qEnd = basicData.qEnd_.at(row);
const auto readLength = qEnd - qStart;
const auto nonMatches = nMM + nDel + nIns;
const float identity = 1.0 - (static_cast<float>(nonMatches)/static_cast<float>(readLength));
return CompareHelper(identity);
}
void PbiReferenceNameFilter::Initialize(const PbiRawData& idx) const
{
const auto pbiFilename = idx.Filename();
const auto bamFilename = pbiFilename.substr(0, pbiFilename.length() - 4);
const auto bamFile = BamFile{ bamFilename };
// single-value
if (rnameWhitelist_ == boost::none) {
const auto tId = bamFile.ReferenceId(rname_);
subFilter_ = PbiReferenceIdFilter{ tId, cmp_ };
}
// multi-value whitelist
else {
subFilter_ = PbiFilter(PbiFilter::UNION);
for (const auto& rname : rnameWhitelist_.get())
subFilter_.Add(PbiReferenceIdFilter{ bamFile.ReferenceId(rname) });
}
initialized_ = true;
}
static
PbiRawData Test2Bam_CoreIndexData(void)
{
PbiRawData rawData;
rawData.Version(PbiFile::Version_3_0_1);
rawData.FileSections(PbiFile::BASIC | PbiFile::MAPPED | PbiFile::REFERENCE);
rawData.NumReads(10);
PbiRawBasicData& basicData = rawData.BasicData();
basicData.rgId_ = { -1197849594, -1197849594, -1197849594, -1197849594, -1197849594, -1197849594, -1197849594, -1197849594, -1197849594, -1197849594 };
basicData.qStart_ = {48,387,0,9936,10232,7468,5557,7285,426,7064};
basicData.qEnd_ = {1132,1134,344,10187,10394,8906,7235,8657,1045,7421};
basicData.holeNumber_ = {49050,32328,32328,6469,6469,30983,13473,13473,19915,30983};
basicData.readQual_ = {0.6,0.6,0.6,0.6,0.6,0.6,0.6,0.6,0.6,0.6};
basicData.ctxtFlag_ = {0,0,0,0,0,0,0,0,0,0};
basicData.fileOffset_ = { 33816576, 33825163, 33831333, 33834264, 33836542, 33838065, 33849818, 33863499, 33874621, 1392836608 };
PbiRawMappedData& mappedData = rawData.MappedData();
mappedData.tId_ = {0,0,0,0,0,0,0,0,0,0};
mappedData.tStart_ = {0,302,675,2170,2203,3572,4506,4507,4592,4669};
mappedData.tEnd_ = {471,1019,1026,2397,2326,5015,6125,5850,5203,5011};
mappedData.aStart_ = {653,395,1,9960,10271,7468,5574,7285,441,7075};
mappedData.aEnd_ = {1129,1134,344,10185,10394,8906,7235,8647,1040,7418};
mappedData.revStrand_ = {0,1,0,1,0,1,1,0,1,0};
mappedData.nM_ = {460,704,339,216,118,1394,1581,1313,583,333};
mappedData.nMM_ = {0,0,0,0,0,0,0,0,0,0};
mappedData.mapQV_ = {254,254,254,254,254,254,254,254,254,254};
PbiRawReferenceData& referenceData = rawData.ReferenceData();
referenceData.entries_ = {
PbiReferenceEntry{0,0,10},
PbiReferenceEntry{4294967295,4294967295,4294967295}
};
return rawData;
}
static
void ExpectRawIndicesEqual(const PbiRawData& expected, const PbiRawData& actual)
{
// header data
EXPECT_EQ(expected.Version(), actual.Version());
EXPECT_EQ(expected.FileSections(), actual.FileSections());
EXPECT_EQ(expected.NumReads(), actual.NumReads());
// subread data
const PbiRawBasicData& e = expected.BasicData();
const PbiRawBasicData& a = actual.BasicData();
EXPECT_EQ(e.rgId_, a.rgId_);
EXPECT_EQ(e.qStart_, a.qStart_);
EXPECT_EQ(e.qEnd_, a.qEnd_);
EXPECT_EQ(e.holeNumber_, a.holeNumber_);
EXPECT_EQ(e.readQual_, a.readQual_);
EXPECT_EQ(e.ctxtFlag_, a.ctxtFlag_);
EXPECT_EQ(e.fileOffset_, a.fileOffset_);
// mapped data
EXPECT_EQ(expected.HasMappedData(), actual.HasMappedData());
if (expected.HasMappedData() && actual.HasMappedData()) {
const PbiRawMappedData& e = expected.MappedData();
const PbiRawMappedData& a = actual.MappedData();
EXPECT_EQ(e.tId_, a.tId_);
EXPECT_EQ(e.tStart_, a.tStart_);
EXPECT_EQ(e.tEnd_, a.tEnd_);
EXPECT_EQ(e.aStart_, a.aStart_);
EXPECT_EQ(e.aEnd_, a.aEnd_);
EXPECT_EQ(e.revStrand_, a.revStrand_);
EXPECT_EQ(e.nM_, a.nM_);
EXPECT_EQ(e.nMM_, a.nMM_);
EXPECT_EQ(e.mapQV_, a.mapQV_);
}
// reference data
EXPECT_EQ(expected.HasReferenceData(), actual.HasReferenceData());
if (expected.HasReferenceData() && actual.HasReferenceData()) {
const PbiRawReferenceData& e = expected.ReferenceData();
const PbiRawReferenceData& a = actual.ReferenceData();
EXPECT_EQ(e.entries_, a.entries_);
}
// barcode data
EXPECT_EQ(expected.HasBarcodeData(), actual.HasBarcodeData());
if (expected.HasBarcodeData() && actual.HasBarcodeData()) {
const PbiRawBarcodeData& e = expected.BarcodeData();
const PbiRawBarcodeData& a = actual.BarcodeData();
EXPECT_EQ(e.bcForward_, a.bcForward_);
EXPECT_EQ(e.bcReverse_, a.bcReverse_);
EXPECT_EQ(e.bcQual_, a.bcQual_);
}
}