void ReadAlignmentArray(int alignmentIndex, ByteAlignment &alignmentArray) {
CmpAlignment cmpAlignment;
alnInfoGroup.ReadCmpAlignment(alignmentIndex, cmpAlignment);
//
// Cache some stats about the read, and where it was aligned to.
//
int queryStart = cmpAlignment.GetQueryStart();
int queryEnd = cmpAlignment.GetQueryEnd();
int refGroupId = cmpAlignment.GetRefGroupId();
int alnGroupId = cmpAlignment.GetAlnGroupId();
int refGroupIndex = refGroupIdToArrayIndex[refGroupId];
if (alnGroupIdToReadGroupName.find(alnGroupId) ==
alnGroupIdToReadGroupName.end()) {
cout << "INTERNAL ERROR! Could not find read group name for alignment "
<< "group with Id " << alnGroupId << "." << endl;
assert(0);
}
string readGroupName = alnGroupIdToReadGroupName[alnGroupId];
if (refAlignGroups[refGroupIndex]->experimentNameToIndex.find(readGroupName) ==
refAlignGroups[refGroupIndex]->experimentNameToIndex.end()) {
cout << "Internal ERROR! The read group name " << readGroupName << " is specified as part of "
<< " the path in alignment " << alignmentIndex
<< " though it does not exist in the ref align group specified for this alignment." << endl;
assert(0);
}
int readGroupIndex = refAlignGroups[refGroupIndex]->experimentNameToIndex[readGroupName];
HDFCmpExperimentGroup* expGroup = refAlignGroups[refGroupIndex]->readGroups[readGroupIndex];
int offsetBegin = cmpAlignment.GetOffsetBegin();
int offsetEnd = cmpAlignment.GetOffsetEnd();
int alignedSequenceLength = offsetEnd - offsetBegin;
if (alignedSequenceLength >= 0) {
alignmentArray.resize(alignedSequenceLength);
}
else {
return;
}
//
// Read the alignment string. All alignments
//
refAlignGroups[refGroupIndex]->readGroups[readGroupIndex]->alignmentArray.Read(offsetBegin,
offsetEnd,
&alignmentArray[0]);
}
int main(int argc, char* argv[])
{
std::string outFileName;
unsigned contextLength = 5;
int minSamples = 500;
int maxSamples = 1000;
if (argc < 3) {
PrintUsage();
std::exit(EXIT_FAILURE);
}
int argi = 1;
std::string cmpH5FileName;
cmpH5FileName = argv[argi++];
outFileName = argv[argi++];
int minAverageQual = 0;
bool onlyMaxLength = false;
while (argi < argc) {
if (strcmp(argv[argi], "-contextLength") == 0) {
contextLength = atoi(argv[++argi]);
} else if (strcmp(argv[argi], "-minSamples") == 0) {
minSamples = atoi(argv[++argi]);
} else if (strcmp(argv[argi], "-maxSamples") == 0) {
maxSamples = atoi(argv[++argi]);
} else if (strcmp(argv[argi], "-onlyMaxLength") == 0) {
onlyMaxLength = true;
} else {
PrintUsage();
std::cout << "ERROR, bad option: " << argv[argi] << std::endl;
std::exit(EXIT_FAILURE);
}
++argi;
}
std::map<std::string, ScoredLength> maxLengthMap;
OutputSampleListSet samples(contextLength);
SMRTSequence read;
std::ofstream sampleOut;
CrucialOpen(outFileName, sampleOut, std::ios::out | std::ios::binary);
int fileNameIndex;
int numContextsReached = 0;
int numContexts = 1 << (contextLength * 2);
ReaderAgglomerate reader;
samples.keyLength = contextLength;
HDFCmpFile<CmpAlignment> cmpReader;
cmpReader.IncludeField("QualityValue");
cmpReader.IncludeField("DeletionQV");
cmpReader.IncludeField("InsertionQV");
cmpReader.IncludeField("SubstitutionQV");
cmpReader.IncludeField("SubstitutionTag");
cmpReader.IncludeField("DeletionTag");
cmpReader.IncludeField("PulseIndex");
cmpReader.IncludeField("WidthInFrames");
cmpReader.IncludeField("PreBaseFrames");
if (cmpReader.Initialize(cmpH5FileName, H5F_ACC_RDWR) == 0) {
std::cout << "ERROR, could not open the cmp file." << std::endl;
std::exit(EXIT_FAILURE);
}
std::cout << "Reading cmp file." << std::endl;
CmpFile cmpFile;
cmpReader.ReadAlignmentDescriptions(cmpFile);
cmpReader.ReadStructure(cmpFile);
std::cout << "done reading structure." << std::endl;
int alignmentIndex;
int nAlignments = cmpReader.alnInfoGroup.GetNAlignments();
std::vector<int> alignmentToBaseMap;
for (alignmentIndex = 0; alignmentIndex < nAlignments and !samples.Sufficient();
alignmentIndex++) {
//
// For ease of use, store the length of the alignment to make another model.
//
ByteAlignment alignmentArray;
cmpReader.ReadAlignmentArray(alignmentIndex, alignmentArray);
Alignment alignment;
ByteAlignmentToAlignment(alignmentArray, alignment);
std::string readSequence, refSequence;
readSequence.resize(alignmentArray.size());
refSequence.resize(alignmentArray.size());
DNASequence readDNA, refDNA;
ByteAlignmentToQueryString(&alignmentArray[0], alignmentArray.size(), &readSequence[0]);
ByteAlignmentToRefString(&alignmentArray[0], alignmentArray.size(), &refSequence[0]);
RemoveGaps(readSequence, readSequence);
RemoveGaps(refSequence, refSequence);
readDNA.seq = (Nucleotide*)readSequence.c_str();
readDNA.length = readSequence.size();
refDNA.seq = (Nucleotide*)refSequence.c_str();
refDNA.length = refSequence.size();
CmpAlignment cmpAlignment;
cmpReader.ImportReadFromCmpH5(alignmentIndex, cmpAlignment, read);
CreateAlignmentToSequenceMap(alignmentArray, alignmentToBaseMap);
if (read.length < contextLength) {
continue;
}
int subreadLength = (cmpFile.alnInfo.alignments[alignmentIndex].GetQueryEnd() -
cmpFile.alnInfo.alignments[alignmentIndex].GetQueryStart());
if (onlyMaxLength == false) {
samples.lengths.push_back(subreadLength);
} else {
int score = (cmpAlignment.GetNMatch() - cmpAlignment.GetNMismatch() -
cmpAlignment.GetNInsertions() - cmpAlignment.GetNDeletions());
std::stringstream nameStrm;
nameStrm << cmpAlignment.GetMovieId() << "_" << cmpAlignment.GetHoleNumber();
std::string nameStr = nameStrm.str();
if (maxLengthMap.find(nameStr) == maxLengthMap.end()) {
maxLengthMap[nameStr] = ScoredLength(score, subreadLength);
}
}
int sampleEnd = alignmentArray.size() - contextLength / 2;
int a;
for (a = contextLength / 2; a < sampleEnd; a++) {
// Make sure the context begins on a real nucleotide.
while (a < sampleEnd and ((RefChar[alignmentArray[a]] == ' '))) {
a++;
}
//
// Move ab back to an index where there are contextLength/2 non-gap
// characters, counted by nb
//
int ab; //num bases
int ae; //alignment end
ab = a - 1;
int nb = 0, ne = 0;
while (true) {
if (RefChar[alignmentArray[ab]] != ' ') {
nb++;
}
if (ab == 0 or nb == static_cast<int>(contextLength) / 2) break;
ab--;
}
//
// Advance ae to an index where there are contextLength/2 non-gap
// characters, counted by ne.
//
ae = a + 1;
while (ae < static_cast<int>(alignmentArray.size()) and
ne < static_cast<int>(contextLength) / 2) {
if (RefChar[alignmentArray[ae]] != ' ') {
ne++;
}
ae++;
}
//
// Make sure there are no edge effects that prevent a context of the correct length from being assigned.
//
if (nb + ne + 1 != static_cast<int>(contextLength)) {
continue;
}
int ai;
std::string context;
for (ai = ab; ai < ae; ai++) {
if (RefChar[alignmentArray[ai]] != ' ') {
context.push_back(RefChar[alignmentArray[ai]]);
}
}
assert(context.size() == contextLength);
//
// Now create the context.
//
OutputSample sample;
//
// This context is a deletion, create that.
//
sample.type = OutputSample::Deletion;
//
// This context is either an insertion or substitution
//
// Look to see if the previous aligned position was an
// insertion, and move back as far as the insertion extends.
int aq = a - 1;
int sampleLength;
if (QueryChar[alignmentArray[a]] == ' ') {
sample.type = OutputSample::Deletion;
sampleLength = 0;
} else if (RefChar[alignmentArray[aq]] == ' ') {
while (aq > 0 and RefChar[alignmentArray[aq]] == ' ' and
QueryChar[alignmentArray[aq]] != ' ') {
aq--;
}
sample.type = OutputSample::Insertion;
sampleLength = a - aq;
} else if (QueryChar[alignmentArray[a]] == RefChar[alignmentArray[aq]]) {
sample.type = OutputSample::Match;
sampleLength = 1;
} else {
sample.type = OutputSample::Substitution;
sampleLength = 1;
}
sample.Resize(sampleLength);
if (sampleLength > 0) {
int seqPos = alignmentToBaseMap[aq];
if (seqPos < static_cast<int>(read.length)) {
sample.CopyFromSeq(read, seqPos, sampleLength);
std::string nucs;
for (size_t n = 0; n < sample.nucleotides.size(); n++) {
char c = sample.nucleotides[n];
assert(c == 'A' or c == 'T' or c == 'G' or c == 'C');
nucs.push_back(sample.nucleotides[n]);
}
}
}
samples.AppendOutputSample(context, sample);
}
read.Free();
}
if (onlyMaxLength) {
std::map<std::string, ScoredLength>::iterator maxScoreIt;
for (maxScoreIt = maxLengthMap.begin(); maxScoreIt != maxLengthMap.end(); ++maxScoreIt) {
std::cout << maxScoreIt->second.length << std::endl;
samples.lengths.push_back(maxScoreIt->second.length);
}
}
samples.Write(sampleOut);
return 0;
}
void ImportReadFromCmpH5(int alignmentIndex, SMRTSequence &read) {
CmpAlignment cmpAlignment;
alnInfoGroup.ReadCmpAlignment(alignmentIndex, cmpAlignment);
//
// Cache some stats about the read, and where it was aligned to.
//
int queryStart = cmpAlignment.GetQueryStart();
int queryEnd = cmpAlignment.GetQueryEnd();
read.holeNumber = cmpAlignment.GetHoleNumber();
int refGroupId = cmpAlignment.GetRefGroupId();
int alnGroupId = cmpAlignment.GetAlnGroupId();
int refGroupIndex = refGroupIdToArrayIndex[refGroupId];
if (alnGroupIdToReadGroupName.find(alnGroupId) == alnGroupIdToReadGroupName.end()) {
cout << "INTERNAL ERROR! Could not find read group name for alignment "
<< "group with Id " << alnGroupId << "." << endl;
assert(0);
}
string readGroupName = alnGroupIdToReadGroupName[alnGroupId];
if (refAlignGroups[refGroupIndex]->experimentNameToIndex.find(readGroupName) ==
refAlignGroups[refGroupIndex]->experimentNameToIndex.end()) {
cout << "Internal ERROR! The read group name " << readGroupName << " is specified as part of "
<< " the path in alignment " << alignmentIndex
<< " though it does not exist in the ref align group specified for this alignment." << endl;
assert(0);
}
int readGroupIndex = refAlignGroups[refGroupIndex]->experimentNameToIndex[readGroupName];
HDFCmpExperimentGroup* expGroup = refAlignGroups[refGroupIndex]->readGroups[readGroupIndex];
int offsetBegin = cmpAlignment.GetOffsetBegin();
int offsetEnd = cmpAlignment.GetOffsetEnd();
int alignedSequenceLength = offsetEnd - offsetBegin;
string alignedSequence;
string readSequence;
vector<unsigned char> byteAlignment;
if (alignedSequenceLength >= 0) {
alignedSequence.resize(alignedSequenceLength);
byteAlignment.resize(alignedSequenceLength);
}
//
// Read the alignment string. All alignments
//
refAlignGroups[refGroupIndex]->readGroups[readGroupIndex]->alignmentArray.Read(offsetBegin,
offsetEnd,
&byteAlignment[0]);
//
// Convert to something we can compare easily.
//
ByteAlignmentToQueryString(&byteAlignment[0], byteAlignment.size(), &alignedSequence[0]);
//
// Initialize the sequence of the read.
//
RemoveGaps(alignedSequence, readSequence);
//
// Make space for the sequence and all fields.
//
read.length = readSequence.size();
read.Allocate(read.length);
memcpy(read.seq, readSequence.c_str(), readSequence.size() * sizeof(char));
vector<int> baseToAlignmentMap;
CreateSequenceToAlignmentMap(byteAlignment, baseToAlignmentMap);
//
// Read in the quality values
//
vector<unsigned char> storedQVArray;
vector<UChar> qvValues;
vector<HalfWord> frameValues;
int length = offsetEnd - offsetBegin;
qvValues.resize(length);
frameValues.resize(length);
int i;
if (expGroup->experimentGroup.ContainsObject("QualityValue")) {
expGroup->qualityValue.Read(offsetBegin, offsetEnd, &qvValues[0]);
StoreQualityValueFromAlignment(qvValues, baseToAlignmentMap, &read.qual.data[0]);
int i;
for (i= 0; i < read.length; i++) {
assert(read.qual[i] < 100);
}
}
if (expGroup->experimentGroup.ContainsObject("InsertionQV")) {
expGroup->insertionQV.Read(offsetBegin, offsetEnd, &qvValues[0]);
StoreQualityValueFromAlignment(qvValues, baseToAlignmentMap, &read.insertionQV.data[0]);
}
if (expGroup->experimentGroup.ContainsObject("SubstitutionQV")) {
expGroup->substitutionQV.Read(offsetBegin, offsetEnd, &qvValues[0]);
StoreQualityValueFromAlignment(qvValues, baseToAlignmentMap, &read.substitutionQV.data[0]);
}
if (expGroup->experimentGroup.ContainsObject("DeletionQV")) {
expGroup->deletionQV.Read(offsetBegin, offsetEnd, &qvValues[0]);
StoreQualityValueFromAlignment(qvValues, baseToAlignmentMap, &read.deletionQV.data[0]);
}
if (expGroup->experimentGroup.ContainsObject("DeletionTag")) {
vector<char> deletionTagValues;
deletionTagValues.resize(offsetEnd-offsetBegin);
expGroup->deletionTag.Read(offsetBegin, offsetEnd, &deletionTagValues[0]);
StoreQualityValueFromAlignment(deletionTagValues, baseToAlignmentMap, read.deletionTag);
}
if (expGroup->experimentGroup.ContainsObject("SubstitutionTag")) {
vector<char> substitutionTagValues;
substitutionTagValues.resize(offsetEnd-offsetBegin);
expGroup->substitutionTag.Read(offsetBegin, offsetEnd, &substitutionTagValues[0]);
StoreQualityValueFromAlignment(substitutionTagValues, baseToAlignmentMap, read.substitutionTag);
}
if (expGroup->experimentGroup.ContainsObject("PulseIndex")) {
vector<uint32_t> pulseIndexValues;
pulseIndexValues.resize(offsetEnd-offsetBegin);
expGroup->pulseIndex.Read(offsetBegin, offsetEnd, &pulseIndexValues[0]);
StoreQualityValueFromAlignment(pulseIndexValues, baseToAlignmentMap, read.pulseIndex);
}
if (expGroup->experimentGroup.ContainsObject("PreBaseFrames")) {
expGroup->preBaseFrames.Read(offsetBegin, offsetEnd, &frameValues[0]);
StoreQualityValueFromAlignment(frameValues, baseToAlignmentMap, read.preBaseFrames);
}
if (expGroup->experimentGroup.ContainsObject("WidthInFrames")) {
expGroup->widthInFrames.Read(offsetBegin, offsetEnd, &frameValues[0]);
StoreQualityValueFromAlignment(frameValues, baseToAlignmentMap, read.widthInFrames);
}
}
void ReadAlignment(int alignmentIndex, AlignmentCandidate<FASTASequence, FASTASequence> &alignment) {
CmpAlignment cmpAln;
ReadAlignment(alignmentIndex, cmpAln);
string refSequence;
string readSequence;
readSequence.resize(cmpAln.alignmentArray.size());
refSequence.resize(cmpAln.alignmentArray.size());
ByteAlignmentToQueryString(&cmpAln.alignmentArray[0], cmpAln.alignmentArray.size(), &readSequence[0]);
ByteAlignmentToRefString(&cmpAln.alignmentArray[0], cmpAln.alignmentArray.size(), &refSequence[0]);
string ungappedRead, ungappedRef;
RemoveGaps(readSequence, ungappedRead);
RemoveGaps(refSequence, ungappedRef);
GappedStringsToAlignment(readSequence, refSequence, alignment);
FASTASequence qAlignedSeq, rAlignedSeq;
qAlignedSeq.seq = (Nucleotide*) &ungappedRead[0];
qAlignedSeq.length = ungappedRead.size();
rAlignedSeq.seq = (Nucleotide*) &ungappedRef[0];
rAlignedSeq.length = ungappedRef.size();
alignment.tAlignedSeq.Copy(rAlignedSeq);
alignment.qAlignedSeq.Copy(qAlignedSeq);
unsigned int qStart = cmpAln.GetQueryStart();
unsigned int tStart = cmpAln.GetRefStart();
alignment.tPos = cmpAln.GetRefStart();
alignment.qPos = cmpAln.GetQueryStart();
alignment.nIns = cmpAln.GetNInsertions();
alignment.nDel = cmpAln.GetNDeletions();
alignment.nMatch = cmpAln.GetNMatch();
alignment.nMismatch=cmpAln.GetNMismatch();
alignment.qStrand= 0;
alignment.tStrand = cmpAln.GetTStrand();
alignment.pctSimilarity = ((float)alignment.nMatch) / (alignment.nMatch + alignment.nMismatch + alignment.nIns + alignment.nDel);
alignment.mapQV = cmpAln.GetMapQV();
}