int main(int argc, char* argv[]) {
string cmpFileName, movieFileName;
int argi = 3;
int numMetrics = 8;
map<string,bool> metricOptions;
int maxElements = 0;
//
// Default is all options are true
//
CreateMetricOptions(metricOptions);
string metricList = "";
bool useCcs = false;
bool byRead = false;
bool failOnMissingData = false;
CommandLineParser clp;
bool printVersion = false;
clp.RegisterStringOption("basFileName", &movieFileName, "The input {bas,pls}.h5 or input.fofn.", true);
clp.RegisterStringOption("cmpFileName", &cmpFileName, "The cmp.h5 file to load pulse information into.", true);
clp.RegisterPreviousFlagsAsHidden();
clp.RegisterStringOption("metrics", &metricList, "The a string delimited list of metrics (with no spaces).The "
"valid options are: QualityValue, ClassifierQV, MergeQV, StartFrame,"
"PulseWidth, pkmid, IPD, and Light.");
clp.RegisterFlagOption("useccs", &useCcs, "Load pulse information for CCS sequences and not raw bases.");
clp.RegisterFlagOption("byread", &byRead, "Load pulse information by read rather than buffering an entire pls.h5 file. "
"This option will soon be deprecated and on by default.");
clp.RegisterIntOption("maxElements", &maxElements, "Set a limit on the size of pls/bas file to buffer in.", CommandLineParser::PositiveInteger);
clp.RegisterFlagOption("failOnMissingData", &failOnMissingData, "Exit if any data fields are missing from the bas.h5 or pls.h5 input that are required to load a metric. Defualt is a warning.");
clp.SetProgramSummary("Load pulse information such as inter pulse distance, or quality information into the cmp.h5 file."
"This allows one to analyze kinetic and quality information by alignment column.");
clp.ParseCommandLine(argc, argv);
if (printVersion) {
cout << VERSION << endl;
exit(1);
}
if (metricList == "") {
SetDefaultMetricOptions(metricOptions);
}
else {
ParseMetricsList(metricList, metricOptions);
}
//
// Always read in basecalls since they are used to check the sanity
// of the alignment indices.
//
metricOptions["Basecall"] = true;
//
// Translate from the metrics to be loaded to the ones that are
// required to compute them.
//
vector<string> datasetFields;
RequirementMap fieldRequirements;
BuildRequirementMap(fieldRequirements);
StoreDatasetFieldsFromPulseFields(metricOptions, fieldRequirements, datasetFields);
vector<string> movieFileNames;
vector<string> fofnMovieNames;
FileOfFileNames::StoreFileOrFileList(movieFileName, movieFileNames);
HDFBasReader hdfBasReader;
HDFPlsReader hdfPlsReader;
HDFCCSReader<SMRTSequence> hdfCcsReader;
vector<string> baseFileFields, pulseFileFields;
int fieldIndex;
bool useBaseFile = false, usePulseFile = false;
for (fieldIndex = 0; fieldIndex < datasetFields.size(); fieldIndex++) {
if (hdfBasReader.ContainsField(datasetFields[fieldIndex])) {
useBaseFile = true;
baseFileFields.push_back(datasetFields[fieldIndex]);
}
}
if (maxElements != 0) {
hdfBasReader.maxAllocNElements = maxElements;
hdfPlsReader.maxAllocNElements = maxElements;
}
//
// For now, all runs will attempt to use information from a .bas
// file, since it's assumed that if one has alignments, one has a
// .bas file.
//
useBaseFile = true;
//
// Add some default fields.
//
hdfBasReader.IncludeField("Basecall");
hdfBasReader.IncludeField("PulseIndex");
hdfBasReader.InitializeFields(baseFileFields);
for (fieldIndex = 0; fieldIndex < datasetFields.size(); fieldIndex++) {
if (hdfPlsReader.ContainsField(datasetFields[fieldIndex])) {
usePulseFile = true;
pulseFileFields.push_back(datasetFields[fieldIndex]);
}
}
if (usePulseFile) {
hdfPlsReader.InitializeFields(pulseFileFields);
}
hdfPlsReader.IncludeField("NumEvent");
int nMovies = movieFileNames.size();
int movieIndex;
MovieNameToArrayIndex movieNameMap;
//
// Initialize movies. This accomplishes two tasks. First, all movie
// files are opened and initialized, so that if there are data
// fields missing the program will exit now rather than in the
// middle of loading pulses.
// Next, a list of movie names is created in fofnMovieNames. The
// cmp file does not necessarily index movies in the order of the
// fofn, and so when loading pulses from a movie indexed by a cmp
// file, one needs to look up the file name of the movie. This is
// done by scanning the fofnMovieNames list in order until the movie
// is found.
for (movieIndex = 0; movieIndex < nMovies; movieIndex++) {
if (!hdfBasReader.Initialize(movieFileNames[movieIndex])) {
cout << "ERROR, could not initialize HDF file "
<< movieFileNames[movieIndex] << " for reading bases." << endl;
exit(1);
}
else {
fofnMovieNames.push_back(hdfBasReader.GetMovieName());
movieNameMap[hdfBasReader.GetMovieName()] = movieIndex;
hdfBasReader.Close();
}
//
// The pulse file is optional.
//
if (usePulseFile) {
if (hdfPlsReader.Initialize(movieFileNames[movieIndex]) == 0) {
usePulseFile = false;
}
}
}
CmpFile cmpFile;
/*
* These readers pull information from the same pls file.
*/
HDFCmpFile<CmpAlignment> cmpReader;
if (cmpReader.Initialize(cmpFileName, H5F_ACC_RDWR) == 0) {
cout << "ERROR, could not open the cmp file." << endl;
exit(0);
}
cmpReader.Read(cmpFile);
string commandLine;
clp.CommandLineToString(argc, argv, commandLine);
string versionStr(VERSION);
AppendPerforceChangelist(PERFORCE_VERSION_STRING, versionStr);
cmpReader.fileLogGroup.AddEntry(commandLine, "Loading pulse metrics", "loadPulses", GetTimestamp(), versionStr);
//
// Group alignment indices by movie so that they may be processed one movie at a time
// later on. The movie indices set keeps track of all indices
// listed in alignment files. This keeps a reference to all
// alignments in memory at once. At the time of writing this, most
// projects will have at most a few million alignments, and so the
// size of this structure is modest.
//
UInt alignmentIndex;
map<int, vector<int> > movieIndexSets;
for (alignmentIndex = 0; alignmentIndex < cmpFile.alnInfo.alignments.size(); alignmentIndex++) {
movieIndexSets[cmpFile.alnInfo.alignments[alignmentIndex].GetMovieId()].push_back(alignmentIndex);
}
vector<float> computedPulseField;
string alignedSequence;
string readSequence;
vector<unsigned char> byteAlignment;
int m;
vector<int> baseToAlignmentMap;
//
// Load pulses from movies in order they appear in the input fofn.
//
int fofnMovieIndex;
for (fofnMovieIndex = 0; fofnMovieIndex < fofnMovieNames.size(); fofnMovieIndex++) {
if (cmpFile.readType == ReadType::CCS or useCcs) {
hdfBasReader.SetReadBasesFromCCS();
hdfCcsReader.Initialize(movieFileNames[fofnMovieIndex]);
}
hdfBasReader.Initialize(movieFileNames[fofnMovieIndex]);
BaseFile baseFile;
PulseFile pulseFile;
if (byRead == false) {
//
// Read the entire bas file at once, and then extract values
// from memory. This can be faster depending on the chunk
// size and size of the movie.
//
hdfBasReader.ReadBaseFile(baseFile);
hdfBasReader.Close();
}
else {
//
// Reads are scanned one by instead of caching all. It is
// still necessary to read in some of the datasets entirely,
// in particular the start positions and hole numbers.
//
// This is repeated below for a pulse file. Since the pulse
// and base files are separate objects, the scan data is
// read into each separately. Somehow later the information
// should be merged into just one.
if (hdfBasReader.scanDataReader.fileHasScanData) {
hdfBasReader.scanDataReader.Read(baseFile.scanData);
}
baseFile.readStartPositions.resize(hdfBasReader.nReads+1);
baseFile.readStartPositions[0] = 0;
hdfBasReader.GetAllReadLengths(baseFile.readLengths);
int i;
assert(baseFile.readLengths.size() + 1 == baseFile.readStartPositions.size());
for (i = 1; i < hdfBasReader.nReads + 1; i++ ) {
baseFile.readStartPositions[i] = baseFile.readLengths[i-1] + baseFile.readStartPositions[i-1];
}
//
// Although the whole bas file isn't being read in, it is
// necessary to read in which hole numbers are contained in this
// bas file since it is possible that the alignment for a
// particular hole number may be in a different input bas.h5
// file even if it is the same movie.
//
hdfBasReader.GetAllHoleNumbers(baseFile.holeNumbers);
}
set<uint32_t> moviePartHoleNumbers;
copy(baseFile.holeNumbers.begin(), baseFile.holeNumbers.end(), inserter(moviePartHoleNumbers, moviePartHoleNumbers.begin()));
if (usePulseFile) {
hdfPlsReader.Initialize(movieFileNames[fofnMovieIndex]);
hdfPlsReader.IncludeField("NumEvent");
hdfPlsReader.IncludeField("StartFrame");
if (byRead == false) {
hdfPlsReader.ReadPulseFile(pulseFile);
hdfPlsReader.Close();
}
else {
if (usePulseFile) {
pulseFile.pulseStartPositions.resize(hdfBasReader.nReads+1);
pulseFile.pulseStartPositions[0] = 0;
hdfPlsReader.GetAllNumEvent(pulseFile.numEvent);
int i;
for (i = 1; i < hdfBasReader.nReads + 1; i++ ) {
pulseFile.pulseStartPositions[i] = pulseFile.numEvent[i-1] + pulseFile.pulseStartPositions[i-1];
}
if (hdfPlsReader.scanDataReader.fileHasScanData) {
hdfPlsReader.scanDataReader.Read(pulseFile.scanData);
}
}
}
}
string cmpFileMovieName;
for (m = 0; m < cmpFile.movieInfo.name.size(); m++) {
//
// First find the file name for the movie 'm'
//
cmpFileMovieName = cmpFile.movieInfo.name[m];
int fofnMovieIndex;
if (baseFile.GetMovieName() == cmpFileMovieName) {
break;
}
}
//
// If the movie specified in the input.fofn is not found in the
// cmp file, that indicates something bad is happeing. Either the
// input.fofn was not used to generate the cmp.h5 file, or no
// alignments were found between the input bas.h5 and the
// reference. That shouldn't happen.
//
if (m == cmpFile.movieInfo.name.size()) {
cout << "WARNING: The movie indexed in the compare file " << cmpFileMovieName << " is not listed in the file " << movieFileName << endl;
continue;
}
//
// Open the movie and load its pulses into memory.
//
movieIndex = cmpFile.movieInfo.id[m];
int movieAlignmentIndex;
float NaN = 0.0/0.0;
UChar missingQualityValue = 255;
HalfWord missingFrameRateValue = USHRT_MAX;
unsigned int missingPulseIndex = UINT_MAX;
//
// Since usePulseFile is set when the input file is a pulseFile,
// and ReadType::CCS becomes the read type when the alignments are
// ccs, when pulse files are specified for de novo ccs alignments,
// they will be opened as pulse files. Since the de novo ccs
// sequences do not have pulse file information, the auto-reading
// of pulse files needs to be disabled. Do that here.
//
if (cmpFile.readType == ReadType::CCS or useCcs) {
usePulseFile = false;
}
//
// Now check the sanity of metric options.
//
map<string,bool>::iterator metricIt;
for (metricIt = metricOptions.begin(); metricIt != metricOptions.end(); ++metricIt) {
if (metricIt->second == false) {
continue;
}
bool metricMayBeComputed = true;
if (cmpFile.readType == ReadType::CCS and
metricIt->first != "QualityValue" and
metricIt->first != "DeletionQV" and
metricIt->first != "SubstitutionQV" and
metricIt->first != "InsertionQV" and
metricIt->first != "DeletionTag" and
metricIt->first != "SubstitutionTag" and
metricIt->first != "Basecall") {
cout << "ERROR! The metric " << metricIt->first << " cannot be loaded into de novo ccs alignemnts." << endl;
// exit(0);
metricMayBeComputed = false;
}
if (metricIt->first == "IPD") {
//
// The field requirements for IPD are special.
//
if ((useBaseFile and !hdfBasReader.FieldIsIncluded("PreBaseFrames")) or
(usePulseFile and (!hdfPlsReader.FieldIsIncluded("StartFrame") and
!hdfPlsReader.FieldIsIncluded("WidthInFrames")))) {
metricMayBeComputed = false;
}
}
else {
if (fieldRequirements.find(metricIt->first) != fieldRequirements.end()) {
//
// There are requirements for this field. Make sure all are
// present before trying to compute this field.
//
int requirementIndex;
for (requirementIndex = 0; requirementIndex < fieldRequirements[metricIt->first].size(); ++requirementIndex) {
string requirement;
requirement = fieldRequirements[metricIt->first][requirementIndex];
if (((useBaseFile == false or ((hdfBasReader.includedFields.find(requirement) == hdfBasReader.includedFields.end() or
hdfBasReader.includedFields[requirement] == false))) and
((usePulseFile == false or (hdfPlsReader.includedFields.find(requirement) == hdfPlsReader.includedFields.end() or
hdfPlsReader.includedFields[requirement] == false))))) {
metricMayBeComputed = false;
}
}
}
else {
//
// There are no requirements for this field, so it must exist as
// a datset in either the bas or pls file.
//
if ((useBaseFile == false or ((hdfBasReader.includedFields.find(metricIt->first) == hdfBasReader.includedFields.end() or
hdfBasReader.includedFields[metricIt->first] == false))) and
(usePulseFile == false or (((hdfPlsReader.includedFields.find(metricIt->first) == hdfPlsReader.includedFields.end() or
hdfPlsReader.includedFields[metricIt->first] == false))))) {
metricMayBeComputed = false;
}
}
}
if (metricMayBeComputed == false) {
if (failOnMissingData) {
cout << "ERROR";
}
else {
cout << "WARNING";
}
cout << ": There is insufficient data to compute metric: " << metricIt->first << " in the file " << movieFileNames[fofnMovieIndex] << " ";
cout << " It will be ignored." << endl;
if (failOnMissingData) {
exit(1);
}
metricOptions[metricIt->first] = false;
}
}
UInt i;
//
// This is currently used as a sentinal for showing that an array
// element does not have a value stored for it, as in deleted
// bases.
//
vector<int> pulseIndexArray;
vector<unsigned int> statTime;
if (metricOptions["WhenStarted"]) {
string whenStarted;
if (hdfPlsReader.scanDataReader.useWhenStarted == false) {
cout << "ERROR! Attempting to read WhenStarted from "
<< movieFileNames[fofnMovieIndex]
<< " but the attriubte does not exist." << endl;
exit(1);
}
hdfPlsReader.scanDataReader.ReadWhenStarted(whenStarted);
if (!cmpReader.movieInfoGroup.whenStartedArray.IsInitialized()) {
cmpReader.movieInfoGroup.whenStartedArray.Initialize(cmpReader.movieInfoGroup.movieInfoGroup, "WhenStarted");
}
cmpReader.movieInfoGroup.whenStartedArray.Write(&whenStarted, 1);
}
if (AnyFieldRequiresFrameRate(datasetFields)) {
if (useBaseFile) {
cmpReader.movieInfoGroup.StoreFrameRate(m, baseFile.GetFrameRate());
}
else if (usePulseFile) {
cmpReader.movieInfoGroup.StoreFrameRate(m, pulseFile.GetFrameRate());
}
}
//
// An index set is a set of indices into the alignment array that
// are of reads generated by this movie. Load pulses for all
// alignments generated for this movie.
//
//
// Movie index sets should be sorted by alignment index. Build a lookup table for this.
//
std::vector<std::pair<int,int> > toFrom;
for (movieAlignmentIndex = 0; movieAlignmentIndex < movieIndexSets[movieIndex].size(); movieAlignmentIndex++) {
alignmentIndex = movieIndexSets[movieIndex][movieAlignmentIndex];
toFrom.push_back(std::pair<int,int>(cmpFile.alnInfo.alignments[alignmentIndex].GetAlignmentId(), movieAlignmentIndex));
}
// orders by first by default.
std::sort(toFrom.begin(), toFrom.end());
//
// Load metrics for alignments from movie 'movieIndex'.
//
cout << "loading " << movieIndexSets[movieIndex].size() << " alignments for movie " << movieIndex << endl;
for (movieAlignmentIndex = 0; movieAlignmentIndex < movieIndexSets[movieIndex].size(); movieAlignmentIndex++) {
alignmentIndex = movieIndexSets[movieIndex][toFrom[movieAlignmentIndex].second];
//
// Alignments are groupsd by ref group id then movie id.
//
int refGroupId = cmpFile.alnInfo.alignments[alignmentIndex].GetRefGroupId();
int movieId = cmpFile.alnInfo.alignments[alignmentIndex].GetMovieId();
UInt holeNumber = cmpFile.alnInfo.alignments[alignmentIndex].GetHoleNumber();
//
// Since the movie may be split into multiple parts, look to see
// if this hole number is one of the ones covered by this
// set. If it is not, just continue. It will be loaded on
// another pass through a different movie part.
//
if (moviePartHoleNumbers.find(holeNumber) == moviePartHoleNumbers.end()) {
continue;
}
//
// Now locate where this movie is stored.
//
if (cmpReader.refGroupIdToArrayIndex.find(refGroupId) == cmpReader.refGroupIdToArrayIndex.end()) {
cout << "ERROR! An alignment " << alignmentIndex << " is specified with reference group " << endl
<< refGroupId << " that is not found as an alignment group." << endl;
exit(1);
}
int refGroupIndex = cmpReader.refGroupIdToArrayIndex[refGroupId];
//
// Now find the group containing the alignment for this movie.
//
if (cmpReader.refAlignGroups[refGroupIndex]->movieIdToIndex.find(movieId) ==
cmpReader.refAlignGroups[refGroupIndex]->movieIdToIndex.end()) {
cout << "ERROR! An alignment " << alignmentIndex << " is specified with movie index " << endl
<< movieId << " that is not found in the alignment group " << refGroupIndex << endl;
exit(1);
}
int readGroupIndex = cmpReader.refAlignGroups[refGroupIndex]->movieIdToIndex[movieId];
//
// First do sanity check on the read to make sure the pules and the bases match.
//
//
// Look to see if the output HDF arrays need to be created.
//
UInt offsetBegin, offsetEnd;
offsetBegin = cmpFile.alnInfo.alignments[alignmentIndex].GetOffsetBegin();
offsetEnd = cmpFile.alnInfo.alignments[alignmentIndex].GetOffsetEnd();
int alignedSequenceLength = offsetEnd - offsetBegin;
if (alignedSequenceLength >= 0) {
alignedSequence.resize(alignedSequenceLength);
byteAlignment.resize(alignedSequenceLength);
}
//
// Read the alignment string. All alignments
//
cmpReader.refAlignGroups[refGroupIndex]->readGroups[readGroupIndex]->alignmentArray.Read(offsetBegin,
offsetEnd,
&byteAlignment[0]);
//
// Convert to something we can compare easily.
//
ByteAlignmentToQueryString(&byteAlignment[0], byteAlignment.size(), &alignedSequence[0]);
//
// Do a sanity check to make sure the pulses and the alignment
// make sense. The main check is to see if the query sequence
// in the alignment is the same as the query sequence in the
// read.
//
//
// First pull out the bases corresponding to this read.
//
int queryStart = cmpFile.alnInfo.alignments[alignmentIndex].GetQueryStart();
int queryEnd = cmpFile.alnInfo.alignments[alignmentIndex].GetQueryEnd();
// Build a map of where
CreateSequenceToAlignmentMap(byteAlignment,
baseToAlignmentMap);
//
// Condense gaps in the alignment for easy comparison.
//
//
RemoveGaps(alignedSequence, alignedSequence);
//
// Query the cmp file for a way to look up a read based on
// coordinate information. For Astro reads, the coords are
// based on x and y. For Springfield, it is read index. The
// base files should be able to look up reads by x,y or by
// index.
//
int readIndex;
if (cmpFile.platformId == Astro) {
cout << "ASTRO pulse loading is deprecated." << endl;
exit(0);
}
if (baseFile.LookupReadIndexByHoleNumber(holeNumber, readIndex) == false) {
cout << "ERROR! Alignment has hole number " << holeNumber << " that is not in the movie. " << endl;
assert(0);
}
int readStart, readLength, alignBaseStart, alignBaseEnd, alignBaseLength;
readStart = baseFile.readStartPositions[readIndex];
readLength = baseFile.readStartPositions[readIndex+1] - baseFile.readStartPositions[readIndex];
alignBaseStart = readStart + queryStart;
alignBaseEnd = readStart + queryEnd;
alignBaseLength = alignBaseEnd - alignBaseStart;
int pulseStart;
if (usePulseFile) {
pulseStart = pulseFile.pulseStartPositions[readIndex];
}
//
// This maps from pulse to a base, since there are more pulses
// called than bases, and the is one pulse for every base.
//
pulseIndexArray.resize(readLength);
SMRTSequence sourceRead;
unsigned int numPasses;
//
// These are not allocated in the regular allocate function
// since they are only used in loadPulses. (maybe I should
// subclass SMRTSequence here).
//
if (byRead) {
// Read in the data from the bas file if it exsts.
if (useBaseFile) {
hdfBasReader.GetReadAt(readIndex, sourceRead);
if (cmpFile.readType == ReadType::CCS or useCcs) {
numPasses = hdfCcsReader.GetNumPasses(readIndex);
}
}
// Read in the data from the pls file if it exists.
if (usePulseFile) {
hdfPlsReader.GetReadAt(readIndex, sourceRead.pulseIndex, sourceRead);
}
}
else {
//
// The entire base/pulse file was read in, so copy data from that into a read
// For the data used in the read, it is possible to simply
// reference the data, but for the pls file it is necessary
// to copy since there is a packing of data.
//
if (useBaseFile) {
baseFile.CopyReadAt(readIndex, sourceRead);
if (cmpFile.readType == ReadType::CCS or useCcs) {
numPasses = hdfCcsReader.GetNumPasses(readIndex);
}
}
if (usePulseFile) {
//
// Copy the subset of pulses that correspond to the ones called as bases.
//
int i;
for (i = 0; i < readLength; i++) {
pulseIndexArray[i] = pulseStart + baseFile.pulseIndex[readStart + i];
}
pulseFile.CopyReadAt(readIndex, &pulseIndexArray[0], sourceRead);
}
}
readSequence.resize(queryEnd - queryStart);
CapQualityValues(sourceRead);
copy((char*) (sourceRead.seq + queryStart),
(char*) (sourceRead.seq + queryEnd),
readSequence.begin());
bool stringsMatch = true;
if (alignedSequence.size() != readSequence.size() or alignedSequence != readSequence) {
cout << "ERROR, the query sequence does not match the aligned query sequence." << endl;
cout << "HoleNumber: "<< holeNumber << ", MovieName: " << cmpFileMovieName;
cout << " ,ReadIndex: " << (int) readIndex <<
cout << ", qStart: "<< queryStart << ", qEnd: " << queryEnd << endl;
cout << "Aligned sequence: "<< endl;
cout << alignedSequence << endl;
cout << "Original sequence: " << endl;
cout << readSequence << endl;
assert(0);
}
/*
* Compute any necessary data fields. These usually involve
* using differences of pulse indices, pulse widths, etc..
* Missing fields are stored as 0's.
*/
vector<float> readPulseMetric;
vector<float> floatMetric;
vector<UChar> qvMetric;
vector<HalfWord> frameRateMetric;
vector<uint32_t> timeMetric;
int ungappedAlignedSequenceLength = alignedSequence.size();
floatMetric.resize(alignedSequenceLength+1);
readPulseMetric.resize(alignedSequenceLength+1);
qvMetric.resize(alignedSequenceLength+1);
frameRateMetric.resize(alignedSequenceLength+1);
timeMetric.resize(alignedSequenceLength+1);
UInt i;
UInt pi;
HDFCmpExperimentGroup* expGroup = cmpReader.refAlignGroups[refGroupIndex]->readGroups[readGroupIndex];
if (cmpFile.readType == ReadType::CCS or useCcs) {
if (!cmpReader.alnInfoGroup.numPasses.IsInitialized()) {
cmpReader.alnInfoGroup.InitializeNumPasses();
}
cmpReader.alnInfoGroup.numPasses.WriteToPos(&numPasses, 1, alignmentIndex);
}
if (metricOptions["StartTimeOffset"] == true) {
if (!expGroup->startTimeOffset.IsInitialized()) {
expGroup->startTimeOffset.Initialize(expGroup->experimentGroup, "StartTimeOffset");
}
unsigned int readStartTimeOffset = sourceRead.startFrame[queryStart];
expGroup->startTimeOffset.WriteToPos(&readStartTimeOffset, 1, alignmentIndex);
}
if (metricOptions["QualityValue"] == true) {
if (!expGroup->qualityValue.IsInitialized()) {
expGroup->qualityValue.Initialize(expGroup->experimentGroup, "QualityValue");
}
// Store start time normalized to frame rate.
fill(qvMetric.begin(), qvMetric.end(), missingQualityValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
qvMetric[baseToAlignmentMap[i]] = sourceRead.qual[queryStart + i];
}
qvMetric[qvMetric.size()-1] = 0;
expGroup->qualityValue.WriteToPos(&qvMetric[0], qvMetric.size(), offsetBegin);
}
if (metricOptions["InsertionQV"] == true) {
if (!expGroup->insertionQV.IsInitialized()) {
expGroup->insertionQV.Initialize(expGroup->experimentGroup, "InsertionQV");
}
// Store start time normalized to frame rate.
fill(qvMetric.begin(), qvMetric.end(), missingQualityValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
qvMetric[baseToAlignmentMap[i]] = sourceRead.insertionQV[queryStart+ i];
}
qvMetric[qvMetric.size()-1] = 0;
expGroup->insertionQV.WriteToPos(&qvMetric[0], qvMetric.size(), offsetBegin);
}
if (metricOptions["MergeQV"] == true) {
if (!expGroup->mergeQV.IsInitialized()) {
expGroup->mergeQV.Initialize(expGroup->experimentGroup, "MergeQV");
}
// Store start time normalized to frame rate.
fill(qvMetric.begin(), qvMetric.end(), missingQualityValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
qvMetric[baseToAlignmentMap[i]] = sourceRead.mergeQV[queryStart+ i];
}
qvMetric[qvMetric.size()-1] = 0;
expGroup->mergeQV.WriteToPos(&qvMetric[0], qvMetric.size(), offsetBegin);
}
if (metricOptions["DeletionQV"] == true) {
if (!expGroup->deletionQV.IsInitialized()) {
expGroup->deletionQV.Initialize(expGroup->experimentGroup, "DeletionQV");
}
// Store start time normalized to frame rate.
fill(qvMetric.begin(), qvMetric.end(), missingQualityValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
qvMetric[baseToAlignmentMap[i]] = sourceRead.deletionQV[queryStart+i];
}
qvMetric[qvMetric.size()-1] = 0;
expGroup->deletionQV.WriteToPos(&qvMetric[0], qvMetric.size(), offsetBegin);
}
if (metricOptions["DeletionTag"] == true) {
if (!expGroup->deletionTag.IsInitialized()) {
expGroup->deletionTag.Initialize(expGroup->experimentGroup, "DeletionTag");
}
vector<char> readDeletionTagMetric;
readDeletionTagMetric.resize(readPulseMetric.size());
// Store start time normalized to frame rate.
for (i = 0; i < readDeletionTagMetric.size()-1; i++ ) {
readDeletionTagMetric[i] = '-';
}
readDeletionTagMetric[i] = '\0';
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
assert(baseToAlignmentMap[i] < readDeletionTagMetric.size());
readDeletionTagMetric[baseToAlignmentMap[i]] = sourceRead.deletionTag[queryStart+i];
}
readDeletionTagMetric[readDeletionTagMetric.size()-1] = 0;
expGroup->deletionTag.WriteToPos(&readDeletionTagMetric[0], readDeletionTagMetric.size(), offsetBegin);
}
if (metricOptions["PulseIndex"] == true) {
if (!expGroup->pulseIndex.IsInitialized()) {
expGroup->pulseIndex.Initialize(expGroup->experimentGroup, "PulseIndex");
}
vector<uint32_t> readPulseIndexMetric;
fill(readPulseIndexMetric.begin(), readPulseIndexMetric.end(), missingPulseIndex);
readPulseIndexMetric.resize(readPulseMetric.size());
// Store start time normalized to frame rate.
assert(readPulseIndexMetric.size() > 0);
for (i = 0; i < readPulseIndexMetric.size(); i++ ) {
readPulseIndexMetric[i] = 0;
}
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
readPulseIndexMetric[baseToAlignmentMap[i]] = sourceRead.pulseIndex[queryStart+i];
}
readPulseIndexMetric[readPulseIndexMetric.size()-1] = 0;
expGroup->pulseIndex.WriteToPos(&readPulseIndexMetric[0], readPulseIndexMetric.size(), offsetBegin);
}
if (metricOptions["SubstitutionTag"] == true) {
if (!expGroup->substitutionTag.IsInitialized()) {
expGroup->substitutionTag.Initialize(expGroup->experimentGroup, "SubstitutionTag");
}
vector<char> readSubstitutionTagMetric;
readSubstitutionTagMetric.resize(readPulseMetric.size());
// Store start time normalized to frame rate.
for (i = 0; i < readSubstitutionTagMetric.size()-1; i++ ) {
readSubstitutionTagMetric[i] = '-';
}
readSubstitutionTagMetric[i] = '\0';
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
readSubstitutionTagMetric[baseToAlignmentMap[i]] = sourceRead.substitutionTag[queryStart+i];
}
readSubstitutionTagMetric[readSubstitutionTagMetric.size()-1] = 0;
expGroup->substitutionTag.WriteToPos(&readSubstitutionTagMetric[0], readSubstitutionTagMetric.size(), offsetBegin);
}
if (metricOptions["SubstitutionQV"] == true) {
if (!expGroup->substitutionQV.IsInitialized()) {
expGroup->substitutionQV.Initialize(expGroup->experimentGroup, "SubstitutionQV");
}
// Store start time normalized to frame rate.
fill(qvMetric.begin(), qvMetric.end(), missingQualityValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
qvMetric[baseToAlignmentMap[i]] = sourceRead.substitutionQV[queryStart+i];
}
qvMetric[qvMetric.size()-1] = 0;
expGroup->substitutionQV.WriteToPos(&qvMetric[0], qvMetric.size(), offsetBegin);
}
if (metricOptions["ClassifierQV"] == true) {
if (!expGroup->classifierQV.IsInitialized()) {
expGroup->classifierQV.Initialize(expGroup->experimentGroup, "ClassifierQV");
}
// Store start time normalized to frame rate.
fill(floatMetric.begin(), floatMetric.end(), missingQualityValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
floatMetric[baseToAlignmentMap[i]] = sourceRead.classifierQV[i+queryStart];
}
floatMetric[floatMetric.size()-1] = 0;
expGroup->classifierQV.WriteToPos(&floatMetric[0], floatMetric.size(), offsetBegin);
}
if (metricOptions["StartFrame"] == true) {
if (!expGroup->startTime.IsInitialized()) {
expGroup->startTime.Initialize(expGroup->experimentGroup, "StartFrame");
}
if (useBaseFile) {
sourceRead.startFrame = new unsigned int[sourceRead.length];
copy(sourceRead.preBaseFrames, &sourceRead.preBaseFrames[sourceRead.length], sourceRead.startFrame);
for (i = 0; i < sourceRead.length-1; i++) {
sourceRead.startFrame[i+1] += sourceRead.widthInFrames[i];
}
partial_sum(sourceRead.startFrame, &sourceRead.startFrame[sourceRead.length], sourceRead.startFrame);
}
// Store start time normalized to frame rate.
fill(timeMetric.begin(), timeMetric.end(), missingPulseIndex);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
timeMetric[baseToAlignmentMap[i]] = sourceRead.startFrame[i+queryStart];
}
timeMetric[timeMetric.size()-1] = 0;
expGroup->startTime.WriteToPos(&timeMetric[0], timeMetric.size(), offsetBegin);
}
if (metricOptions["PulseWidth"] == true) {
if (!expGroup->pulseWidth.IsInitialized()) {
expGroup->pulseWidth.Initialize(expGroup->experimentGroup, "PulseWidth");
}
// Store start time normalized to frame rate.
fill(frameRateMetric.begin(), frameRateMetric.end(), missingFrameRateValue);
//
// For legacy reasons, it's possible the width in frames is
// stored in the bas file. If this is the case, use the width
// in frames there. Otherwise, use the width in frames stored
// in the pls file.
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
frameRateMetric[baseToAlignmentMap[i]] = sourceRead.widthInFrames[queryStart + i];
}
frameRateMetric[frameRateMetric.size()-1] = 0;
expGroup->pulseWidth.WriteToPos(&frameRateMetric[0], frameRateMetric.size(), offsetBegin);
}
if (metricOptions["PreBaseFrames"] == true) {
if (!expGroup->preBaseFrames.IsInitialized()) {
expGroup->preBaseFrames.Initialize(expGroup->experimentGroup, "PreBaseFrames");
}
// Compute width in frames normalized to frame rate.
fill(frameRateMetric.begin(), frameRateMetric.end(), missingFrameRateValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
frameRateMetric[baseToAlignmentMap[i]] = sourceRead.preBaseFrames[i+queryStart];
}
frameRateMetric[frameRateMetric.size()-1] = 0;
expGroup->preBaseFrames.WriteToPos(&frameRateMetric[0], frameRateMetric.size(), offsetBegin);
}
if (metricOptions["WidthInFrames"] == true) {
if (!expGroup->widthInFrames.IsInitialized()) {
expGroup->widthInFrames.Initialize(expGroup->experimentGroup, "WidthInFrames");
}
// Compute width in frames normalized to frame rate.
fill(frameRateMetric.begin(), frameRateMetric.end(), missingFrameRateValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
if (usePulseFile) {
frameRateMetric[baseToAlignmentMap[i]] = sourceRead.widthInFrames[i+queryStart];
}
else {
frameRateMetric[baseToAlignmentMap[i]] = sourceRead.widthInFrames[i+queryStart];
}
}
frameRateMetric[frameRateMetric.size()-1] = 0;
expGroup->widthInFrames.WriteToPos(&frameRateMetric[0], frameRateMetric.size(), offsetBegin);
}
if (metricOptions["pkmid"] == true) {
if (!expGroup->pkmid.IsInitialized()) {
expGroup->pkmid.Initialize(expGroup->experimentGroup, "pkmid");
}
for (i = 0; i < readPulseMetric.size(); i++ ) {
readPulseMetric[i] = NaN;
}
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
readPulseMetric[baseToAlignmentMap[i]] = sourceRead.midSignal[i+queryStart];
}
readPulseMetric[readPulseMetric.size()-1] = 0;
expGroup->pkmid.WriteToPos(&readPulseMetric[0], readPulseMetric.size(), offsetBegin);
}
if (metricOptions["IPD"] == true) {
if (!expGroup->ipd.IsInitialized()) {
expGroup->ipd.Initialize(expGroup->experimentGroup, "IPD");
}
fill(frameRateMetric.begin(), frameRateMetric.end(), missingFrameRateValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
//
// The IPD is undefined for the first base in a read.
//
if (usePulseFile ) {
if (queryStart == 0 and i == 0) {
frameRateMetric[baseToAlignmentMap[i]] = 0;
}
else {
frameRateMetric[baseToAlignmentMap[i]] = (sourceRead.startFrame[i+queryStart]
- sourceRead.startFrame[i+queryStart-1]
- sourceRead.widthInFrames[i+queryStart-1]);
}
}
else if (useBaseFile) {
frameRateMetric[baseToAlignmentMap[i]] = sourceRead.preBaseFrames[i + queryStart];
}
}
frameRateMetric[frameRateMetric.size()-1] = 0;
expGroup->ipd.WriteToPos(&frameRateMetric[0], frameRateMetric.size(), offsetBegin);
}
if (metricOptions["Light"] == true) {
if (!expGroup->light.IsInitialized()) {
expGroup->light.Initialize(expGroup->experimentGroup, "Light");
}
fill(frameRateMetric.begin(), frameRateMetric.end(), missingFrameRateValue);
for (i = 0; i < ungappedAlignedSequenceLength; i++ ) {
frameRateMetric[baseToAlignmentMap[i]] = sourceRead.meanSignal[i+queryStart];
frameRateMetric[baseToAlignmentMap[i]] = (frameRateMetric[baseToAlignmentMap[i]] *
sourceRead.widthInFrames[i+queryStart]);
}
frameRateMetric[frameRateMetric.size()-1] = 0;
expGroup->light.WriteToPos(&frameRateMetric[0], frameRateMetric.size(), offsetBegin);
}
sourceRead.Free();
Free(sourceRead.meanSignal);
Free(sourceRead.maxSignal);
Free(sourceRead.midSignal);
Free(sourceRead.startFrame);
Free(sourceRead.classifierQV);
Free(sourceRead.widthInFrames);
}
if (byRead == true) {
if (useBaseFile) {
hdfBasReader.Close();
}
if (cmpFile.readType == ReadType::CCS or useCcs) {
hdfCcsReader.Close();
}
if (usePulseFile) {
hdfPlsReader.Close();
}
}
} // done loading movies
cmpReader.Close();
}
TEST(SubreadsTest, EndToEnd_Multiple)
{
// setup
const string movieName = "m140905_042212_sidney_c100564852550000001823085912221377_s1_X0";
vector<string> baxFilenames;
baxFilenames.push_back(tests::Data_Dir + "/" + movieName + ".1.bax.h5");
const string generatedBam = movieName + ".subreads.bam";
const string scrapBam = movieName + ".scraps.bam";
// run conversion
const int result = RunBax2Bam(baxFilenames, "--subread");
EXPECT_EQ(0, result);
// open BAX reader on original data
HDFBasReader baxReader;
baxReader.IncludeField("Basecall");
baxReader.IncludeField("DeletionQV");
baxReader.IncludeField("DeletionTag");
baxReader.IncludeField("InsertionQV");
baxReader.IncludeField("PreBaseFrames");
baxReader.IncludeField("MergeQV");
baxReader.IncludeField("SubstitutionQV");
baxReader.IncludeField("HQRegionSNR");
// not using SubTag or PulseWidth here
string baxBasecallerVersion;
string baxBindingKit;
string baxSequencingKit;
const int initOk = baxReader.Initialize(baxFilenames.front());
EXPECT_EQ(1, initOk);
if (initOk == 1) {
if (baxReader.scanDataReader.fileHasScanData && baxReader.scanDataReader.initializedRunInfoGroup) {
if (baxReader.scanDataReader.runInfoGroup.ContainsAttribute("BindingKit")) {
HDFAtom<std::string> bkAtom;
if (bkAtom.Initialize(baxReader.scanDataReader.runInfoGroup, "BindingKit")) {
bkAtom.Read(baxBindingKit);
bkAtom.dataspace.close();
}
}
if (baxReader.scanDataReader.runInfoGroup.ContainsAttribute("SequencingKit")) {
HDFAtom<std::string> skAtom;
if (skAtom.Initialize(baxReader.scanDataReader.runInfoGroup, "SequencingKit")) {
skAtom.Read(baxSequencingKit);
skAtom.dataspace.close();
}
}
}
baxReader.GetChangeListID(baxBasecallerVersion);
}
// read region table info
boost::scoped_ptr<HDFRegionTableReader> regionTableReader(new HDFRegionTableReader);
RegionTable regionTable;
std::string fn = baxFilenames.front();
EXPECT_TRUE(regionTableReader->Initialize(fn) != 0);
regionTable.Reset();
regionTableReader->ReadTable(regionTable);
regionTableReader->Close();
EXPECT_NO_THROW(
{
// open BAM file
BamFile bamFile(generatedBam);
// check BAM header information
const BamHeader& header = bamFile.Header();
EXPECT_EQ(string("1.5"), header.Version());
EXPECT_EQ(string("unknown"), header.SortOrder());
EXPECT_EQ(string("3.0.1"), header.PacBioBamVersion());
EXPECT_TRUE(header.Sequences().empty());
EXPECT_TRUE(header.Comments().empty());
ASSERT_FALSE(header.Programs().empty());
const vector<string> readGroupIds = header.ReadGroupIds();
ASSERT_FALSE(readGroupIds.empty());
const ReadGroupInfo& rg = header.ReadGroup(readGroupIds.front());
string rawId = movieName + "//SUBREAD";
string md5Id;
MakeMD5(rawId, md5Id, 8);
EXPECT_EQ(md5Id, rg.Id());
EXPECT_EQ(string("PACBIO"), rg.Platform());
EXPECT_EQ(movieName, rg.MovieName());
EXPECT_TRUE(rg.SequencingCenter().empty());
EXPECT_TRUE(rg.Date().empty());
EXPECT_TRUE(rg.FlowOrder().empty());
EXPECT_TRUE(rg.KeySequence().empty());
EXPECT_TRUE(rg.Library().empty());
EXPECT_TRUE(rg.Programs().empty());
EXPECT_TRUE(rg.PredictedInsertSize().empty());
EXPECT_TRUE(rg.Sample().empty());
EXPECT_EQ("SUBREAD", rg.ReadType());
EXPECT_EQ(baxBasecallerVersion, rg.BasecallerVersion());
EXPECT_EQ(baxBindingKit, rg.BindingKit());
EXPECT_EQ(baxSequencingKit, rg.SequencingKit());
EXPECT_EQ(75, std::stod(rg.FrameRateHz()));
EXPECT_EQ("dq", rg.BaseFeatureTag(BaseFeature::DELETION_QV));
EXPECT_EQ("dt", rg.BaseFeatureTag(BaseFeature::DELETION_TAG));
EXPECT_EQ("iq", rg.BaseFeatureTag(BaseFeature::INSERTION_QV));
EXPECT_EQ("ip", rg.BaseFeatureTag(BaseFeature::IPD));
EXPECT_EQ("mq", rg.BaseFeatureTag(BaseFeature::MERGE_QV));
EXPECT_EQ("sq", rg.BaseFeatureTag(BaseFeature::SUBSTITUTION_QV));
EXPECT_FALSE(rg.HasBaseFeature(BaseFeature::SUBSTITUTION_TAG));
EXPECT_EQ(FrameCodec::V1, rg.IpdCodec());
// compare 1st record from each file
SMRTSequence baxRecord;
UInt holeNumber = 0;
vector<float> hqSnr;
size_t intervalIdx = 0;
vector<SubreadInterval> subreadIntervals;
size_t numTested = 0;
EntireFileQuery entireFile(bamFile);
for (BamRecord& bamRecord : entireFile) {
if (intervalIdx >= subreadIntervals.size())
{
while (baxReader.GetNext(baxRecord))
{
holeNumber = baxRecord.zmwData.holeNumber;
ComputeSubreadIntervals(&subreadIntervals, regionTable, holeNumber);
/* this is for debugging subread interval problems
int hqStart = 0;
int hqEnd = 0;
int hqScore = 0;
LookupHQRegion(holeNumber,
regionTable,
hqStart,
hqEnd,
hqScore);
vector<ReadInterval> subreadIntervals_;
CollectSubreadIntervals(baxRecord, ®ionTable, subreadIntervals_);
for (int i = subreadIntervals_.size() - 1; i >= 0; --i)
{
auto& in = subreadIntervals_[i];
int inStart = max(hqStart, in.start);
int inEnd = min(hqEnd, in.end);
if (inEnd <= inStart)
subreadIntervals_.erase(subreadIntervals_.begin() + i);
}
cerr << "hqRegion: " << hqStart << ", " << hqEnd << endl;
cerr << "subreadRegions:" << endl;
for (const auto& in : subreadIntervals_)
cerr << " l, r: " << in.start << ", " << in.end << endl;
cerr << "adapterDerived:" << endl;
for (const auto& in : subreadIntervals)
cerr << " l, r: " << in.Start << ", " << in.End << endl;
cerr << endl;
// */
if (subreadIntervals.empty())
continue;
intervalIdx = 0;
hqSnr.clear();
hqSnr.push_back(baxRecord.HQRegionSnr('A'));
hqSnr.push_back(baxRecord.HQRegionSnr('C'));
hqSnr.push_back(baxRecord.HQRegionSnr('G'));
hqSnr.push_back(baxRecord.HQRegionSnr('T'));
EXPECT_GT(hqSnr[0], 0);
EXPECT_GT(hqSnr[1], 0);
EXPECT_GT(hqSnr[2], 0);
EXPECT_GT(hqSnr[3], 0);
goto compare;
}
goto cleanup;
}
compare:
const BamRecordImpl& bamRecordImpl = bamRecord.Impl();
EXPECT_EQ(4680,bamRecordImpl.Bin());
EXPECT_EQ(0, bamRecordImpl.InsertSize());
EXPECT_EQ(255, bamRecordImpl.MapQuality());
EXPECT_EQ(-1, bamRecordImpl.MatePosition());
EXPECT_EQ(-1, bamRecordImpl.MateReferenceId());
EXPECT_EQ(-1, bamRecordImpl.Position());
EXPECT_EQ(-1, bamRecordImpl.ReferenceId());
EXPECT_FALSE(bamRecordImpl.IsMapped());
const int subreadStart = subreadIntervals[intervalIdx].Start;
const int subreadEnd = subreadIntervals[intervalIdx].End;
const string expectedName = movieName + "/" +
to_string(holeNumber) + "/" +
to_string(subreadStart) + "_" +
to_string(subreadEnd);
EXPECT_EQ(expectedName, bamRecordImpl.Name());
using PacBio::BAM::QualityValue;
using PacBio::BAM::QualityValues;
const DNALength length = subreadEnd - subreadStart;
string expectedSequence;
expectedSequence.assign((const char*)baxRecord.seq + subreadStart, length);
const string bamSequence = bamRecord.Sequence();
const QualityValues bamQualities = bamRecord.Qualities();
EXPECT_EQ(expectedSequence, bamSequence);
EXPECT_TRUE(bamQualities.empty());
const QualityValues bamDeletionQVs = bamRecord.DeletionQV();
const QualityValues bamInsertionQVs = bamRecord.InsertionQV();
const QualityValues bamMergeQVs = bamRecord.MergeQV();
const QualityValues bamSubstitutionQVs = bamRecord.SubstitutionQV();
for (size_t i = 0; i < length; ++i) {
const size_t pos = subreadStart + i;
EXPECT_EQ((QualityValue)baxRecord.GetDeletionQV(pos), bamDeletionQVs.at(i));
EXPECT_EQ((QualityValue)baxRecord.GetInsertionQV(pos), bamInsertionQVs.at(i));
EXPECT_EQ((QualityValue)baxRecord.GetMergeQV(pos), bamMergeQVs.at(i));
EXPECT_EQ((QualityValue)baxRecord.GetSubstitutionQV(pos), bamSubstitutionQVs.at(i));
}
if (baxRecord.deletionTag)
{
string expectedDeletionTags;
expectedDeletionTags.assign((char*)baxRecord.deletionTag + subreadStart,
(char*)baxRecord.deletionTag + subreadStart + length);
const string& bamDeletionTags = bamRecord.DeletionTag();
EXPECT_EQ(expectedDeletionTags, bamDeletionTags);
}
if (baxRecord.substitutionTag)
{
string expectedSubstitutionTags;
expectedSubstitutionTags.assign((char*)baxRecord.substitutionTag + subreadStart,
(char*)baxRecord.substitutionTag + subreadStart + length);
const string& bamSubstitutionTags = bamRecord.SubstitutionTag();
EXPECT_EQ(expectedSubstitutionTags, bamSubstitutionTags);
}
// TODO: IPDs
const LocalContextFlags ctxFlags = subreadIntervals[intervalIdx].LocalContextFlags;
EXPECT_EQ(md5Id, bamRecord.ReadGroupId());
EXPECT_EQ(movieName, bamRecord.MovieName());
EXPECT_EQ(1, bamRecord.NumPasses());
EXPECT_EQ(holeNumber, bamRecord.HoleNumber());
EXPECT_EQ(subreadStart, bamRecord.QueryStart());
EXPECT_EQ(subreadEnd, bamRecord.QueryEnd());
EXPECT_EQ(hqSnr, bamRecord.SignalToNoise());
EXPECT_EQ(ctxFlags, bamRecord.LocalContextFlags());
numTested++;
intervalIdx++;
}
cleanup:
EXPECT_GT(numTested, 1);
// cleanup
baxReader.Close();
RemoveFile(generatedBam);
RemoveFile(scrapBam);
}); // EXPECT_NO_THROW
int main(int argc, char* argv[]) {
string plsFileName, fastaOutName;
vector<string> plsFileNames;
bool trimByRegion, maskByRegion;
trimByRegion = false;
maskByRegion = false;
int argi = 3;
RegionTable regionTable;
string regionsFOFNName = "";
vector<string> regionFileNames;
bool splitSubreads = true;
int minSubreadLength = 0;
bool addSimulatedData = false;
bool printSimulatedCoordinate = false;
bool printSimulatedSequenceIndex = false;
bool printFastq = false;
bool printCcs = false;
int lineLength = 50;
int minReadScore = 0;
bool encodeCCSPassesInTitle = false;
vector<int> holeNumbers;
CommandLineParser clp;
bool printOnlyBest = false;
clp.SetProgramName("pls2fasta");
clp.RegisterStringOption("file.pls.h5", &plsFileName, "Input pls/bas.h5 file.", true);
clp.RegisterStringOption("out.fasta", &fastaOutName, "Output fasta/fastq file.", true);
clp.RegisterPreviousFlagsAsHidden();
clp.RegisterFlagOption("trimByRegion", &trimByRegion, "Trim away low quality regions.");
clp.RegisterFlagOption("maskByRegion", &maskByRegion, "Mask low quality regions with 'N'.");
clp.RegisterStringOption("regionTable", ®ionsFOFNName, "Optional HDF file with a /PulseData/Regions dataset.");
clp.RegisterIntOption("minSubreadLength", &minSubreadLength, "Do not write subreads less than the specified length.", CommandLineParser::PositiveInteger);
clp.RegisterFlagOption("noSplitSubreads", &splitSubreads, "Do not split reads on adapter sequences.");
clp.RegisterIntListOption("holeNumber", &holeNumbers, "Only print this hole number (or list of numbers).");
clp.RegisterFlagOption("fastq", &printFastq, "Print in FASTQ format with quality.");
clp.RegisterFlagOption("ccs", &printCcs, "Print de novo CCS sequences");
clp.RegisterFlagOption("passesInTitle", &encodeCCSPassesInTitle, "Append /N_passes/ to ccs sequence title, where"
" N is the number of passes.");
clp.RegisterIntOption("lineLength", &lineLength, "Specify fasta/fastq line length", CommandLineParser::PositiveInteger);
clp.RegisterIntOption("minReadScore", &minReadScore, "Minimum read score to print a read. The score is "
"a number between 0 and 1000 and represents the expected accuracy percentage * 10. "
"A typical value would be between 750 and 800. This does not apply to ccs reads.", CommandLineParser::NonNegativeInteger);
clp.RegisterFlagOption("best", &printOnlyBest, "If a CCS sequence exists, print this. Otherwise, print the longest"
"subread. This does not support fastq.");
clp.SetProgramSummary("Converts bas.h5 files to fasta or fastq files. Although fasta files are provided"
" with every run, they are not trimmed nor split into subreads. This program takes "
"additional annotation information, such as the subread coordinates and high quality regions "
"and uses them to create fasta sequences that are substrings of all bases called. Most of the time "
"you will want to trim low quality reads, so you should specify -trimByRegion.");
clp.ParseCommandLine(argc, argv);
if (trimByRegion and maskByRegion) {
cout << "ERROR! You cannot both trim and mask regions. Use one or the other." << endl;
exit(1);
}
if (printFastq) {
// Setting lineLength to 0 flags to print on one line.
lineLength = 0;
}
if (FileOfFileNames::IsFOFN(plsFileName)) {
FileOfFileNames::FOFNToList(plsFileName, plsFileNames);
}
else {
plsFileNames.push_back(plsFileName);
}
if (regionsFOFNName == "") {
regionFileNames = plsFileNames;
}
else {
if (FileOfFileNames::IsFOFN(regionsFOFNName)) {
FileOfFileNames::FOFNToList(regionsFOFNName, regionFileNames);
}
else {
regionFileNames.push_back(regionsFOFNName);
}
}
ofstream fastaOut;
CrucialOpen(fastaOutName, fastaOut);
int plsFileIndex;
HDFRegionTableReader hdfRegionReader;
sort(holeNumbers.begin(), holeNumbers.end());
for (plsFileIndex = 0; plsFileIndex < plsFileNames.size(); plsFileIndex++) {
if (trimByRegion or maskByRegion or splitSubreads) {
hdfRegionReader.Initialize(regionFileNames[plsFileIndex]);
hdfRegionReader.ReadTable(regionTable);
regionTable.SortTableByHoleNumber();
}
HDFCCSReader<CCSSequence> ccsReader;
HDFBasReader smrtReader;
if (printCcs) {
ccsReader.Initialize(plsFileNames[plsFileIndex]);
}
else {
smrtReader.Initialize(plsFileNames[plsFileIndex]);
}
vector<ReadInterval> subreadIntervals;;
SMRTSequence seq;
CCSSequence ccsSeq;
while (true) {
if (printCcs == true or printOnlyBest == true) {
if (ccsReader.GetNext(ccsSeq) == false) {
break;
}
else {
seq = ccsSeq.unrolledRead;
}
}
else {
if (smrtReader.GetNext(seq) == false) {
break;
}
}
if (holeNumbers.size() != 0 and
binary_search(holeNumbers.begin(), holeNumbers.end(), seq.zmwData.holeNumber) == false) {
continue;
}
if (encodeCCSPassesInTitle) {
assert(printCcs);
string title = ccsSeq.title;
stringstream titleStrm;
titleStrm << title << "/"<<ccsSeq.numPasses << "_passes/";
ccsSeq.CopyTitle(titleStrm.str());
}
if (printCcs == true or (printOnlyBest and ccsSeq.length > 0)) {
//
// The reason for checking to see if the length is greater
// than 0 again is in case -ccs flag is specified, but the ccs
// sequence is empty, nothing should be printed. When
// printing only the best this block should only be entered if
// a ccs sequence exists, because the rest of the loop is
// skipped after this.
if (ccsSeq.length > 0) {
if (printFastq == false) {
ccsSeq.PrintSeq(fastaOut);
}
else {
ccsSeq.PrintFastq(fastaOut, lineLength);
}
}
ccsSeq.Free();
seq.Free();
continue;
}
if (seq.length == 0) {
continue;
}
//
// Determine the high quality boundaries of the read. This is
// the full read is no hq regions exist, or it is stated to
// ignore regions.
//
DNALength hqReadStart, hqReadEnd;
int hqRegionScore;
if (GetReadTrimCoordinates(seq, seq.zmwData, regionTable, hqReadStart, hqReadEnd, hqRegionScore) == false or
(trimByRegion == false and maskByRegion == false)) {
hqReadStart = 0;
hqReadEnd = seq.length;
}
//
// Mask off the low quality portions of the reads.
//
if (maskByRegion) {
if (hqReadStart > 0) {
fill(&seq.seq[0], &seq.seq[hqReadStart], 'N');
}
if (hqReadEnd != seq.length) {
fill(&seq.seq[hqReadEnd], &seq.seq[seq.length], 'N');
}
}
//
// Now possibly print the full read with masking. This could be handled by making a
//
if (splitSubreads == false) {
ReadInterval wholeRead(0, seq.length);
// The set of subread intervals is just the entire read.
subreadIntervals.clear();
subreadIntervals.push_back(wholeRead);
}
else {
//
// Print subread coordinates no matter whether or not reads have subreads.
//
subreadIntervals.clear(); // clear old, new intervals are appended.
CollectSubreadIntervals(seq, ®ionTable, subreadIntervals);
}
//
// Output all subreads as separate sequences.
//
int intvIndex;
SMRTSequence bestSubreadSequence;
int bestSubreadScore = -1;
int bestSubreadIndex = 0;
int bestSubreadStart = 0, bestSubreadEnd = 0;
SMRTSequence bestSubread;
for (intvIndex = 0; intvIndex < subreadIntervals.size(); intvIndex++) {
SMRTSequence subreadSequence, subreadSequenceRC;
subreadSequence.subreadStart = subreadIntervals[intvIndex].start;
subreadSequence.subreadEnd = subreadIntervals[intvIndex].end;
//
// When trimming by region, only output the parts of the
// subread that overlap the hq region.
//
if (trimByRegion == true) {
subreadSequence.subreadStart = max((DNALength) subreadIntervals[intvIndex].start, hqReadStart);
subreadSequence.subreadEnd = min((DNALength) subreadIntervals[intvIndex].end, hqReadEnd);
}
if (subreadSequence.subreadStart >= subreadSequence.subreadEnd or
subreadSequence.subreadEnd - subreadSequence.subreadStart <= minSubreadLength) {
//
// There is no high qualty portion of this subread. Skip it.
//
continue;
}
if (hqRegionScore < minReadScore) {
continue;
}
//
// Print the subread, adding the coordinates as part of the title.
//
subreadSequence.ReferenceSubstring(seq, subreadSequence.subreadStart,
subreadSequence.subreadEnd - subreadSequence.subreadStart);
stringstream titleStream;
titleStream << seq.title;
if (splitSubreads) {
//
// Add the subread coordinates if splitting on subread.
//
titleStream << "/"
<< subreadSequence.subreadStart
<< "_" << subreadSequence.subreadEnd;
}
subreadSequence.CopyTitle(titleStream.str());
//
// Eventually replace with WriterAgglomerate.
//
if (printOnlyBest == false) {
if (subreadSequence.length > 0) {
if (printFastq == false) {
((FASTASequence*)&subreadSequence)->PrintSeq(fastaOut);
}
else {
subreadSequence.PrintFastq(fastaOut, lineLength);
}
}
delete[] subreadSequence.title;
}
else {
int subreadWeightedScore = subreadSequence.length * hqRegionScore;
if (subreadWeightedScore > bestSubreadScore) {
bestSubreadIndex = intvIndex;
bestSubread = subreadSequence;
bestSubreadScore = subreadWeightedScore;
}
}
}
if (printOnlyBest) {
if (bestSubreadScore >= 0) {
if (printFastq == false) {
bestSubread.PrintSeq(fastaOut);
}
else {
bestSubread.PrintFastq(fastaOut, bestSubread.length);
}
bestSubread.Free();
}
}
ccsSeq.Free();
seq.Free();
}
if (printCcs or printOnlyBest) {
ccsReader.Close();
}
else {
smrtReader.Close();
}
hdfRegionReader.Close();
}
}