bool QualityScores::stripQualThreshold(Sequence& sequence, double qThreshold){
try {
string rawSequence = sequence.getUnaligned();
int seqLength = sequence.getNumBases();
if(seqName != sequence.getName()){
m->mothurOut("sequence name mismatch btwn fasta: " + sequence.getName() + " and qual file: " + seqName);
m->mothurOutEndLine();
}
int end;
for(int i=0;i<seqLength;i++){
end = i;
if(qScores[i] < qThreshold){
break;
}
}
//every score passed
if (end == (seqLength-1)) { end = seqLength; }
sequence.setUnaligned(rawSequence.substr(0,end));
trimQScores(-1, end);
return 1;
}
catch(exception& e) {
m->errorOut(e, "QualityScores", "flipQScores");
exit(1);
}
}
bool QualityScores::stripQualWindowAverage(Sequence& sequence, int stepSize, int windowSize, double qThreshold){
try {
string rawSequence = sequence.getUnaligned();
int seqLength = sequence.getNumBases();
if(seqName != sequence.getName()){
m->mothurOut("sequence name mismatch between fasta: " + sequence.getName() + " and qual file: " + seqName);
m->mothurOutEndLine();
}
int end = windowSize;
int start = 0;
if(seqLength < windowSize) { return 0; }
while((start+windowSize) < seqLength){
double windowSum = 0.0000;
for(int i=start;i<end;i++){
windowSum += qScores[i];
}
double windowAverage = windowSum / (double)(end-start);
if(windowAverage < qThreshold){
end = end - stepSize;
break;
}
start += stepSize;
end = start + windowSize;
if(end >= seqLength){ end = seqLength; }
}
if(end == -1){ end = seqLength; }
//failed first window
if (end < windowSize) { return 0; }
sequence.setUnaligned(rawSequence.substr(0,end));
trimQScores(-1, end);
return 1;
}
catch(exception& e) {
m->errorOut(e, "QualityScores", "stripQualWindowAverage");
exit(1);
}
}
bool QualityScores::stripQualRollingAverage(Sequence& sequence, double qThreshold){
try {
string rawSequence = sequence.getUnaligned();
int seqLength = sequence.getNumBases();
if(seqName != sequence.getName()){
m->mothurOut("sequence name mismatch btwn fasta: " + sequence.getName() + " and qual file: " + seqName);
m->mothurOutEndLine();
}
int end = -1;
double rollingSum = 0.0000;
for(int i=0;i<seqLength;i++){
rollingSum += (double)qScores[i];
if(rollingSum / (double)(i+1) < qThreshold){
end = i;
break;
}
}
if(end == -1){ end = seqLength; }
sequence.setUnaligned(rawSequence.substr(0,end));
trimQScores(-1, end);
return 1;
}
catch(exception& e) {
m->errorOut(e, "QualityScores", "flipQScores");
exit(1);
}
}
//**********************************************************************************************************************
int AlignCommand::driver(linePair* filePos, string alignFName, string reportFName, string accnosFName, string filename){
try {
ofstream alignmentFile;
m->openOutputFile(alignFName, alignmentFile);
ofstream accnosFile;
m->openOutputFile(accnosFName, accnosFile);
NastReport report(reportFName);
ifstream inFASTA;
m->openInputFile(filename, inFASTA);
inFASTA.seekg(filePos->start);
bool done = false;
int count = 0;
//moved this into driver to avoid deep copies in windows paralellized version
Alignment* alignment;
int longestBase = templateDB->getLongestBase();
if (m->debug) { m->mothurOut("[DEBUG]: template longest base = " + toString(templateDB->getLongestBase()) + " \n"); }
if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase); }
else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); }
else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }
else if(align == "noalign") { alignment = new NoAlign(); }
else {
m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
m->mothurOutEndLine();
alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
}
while (!done) {
if (m->control_pressed) { break; }
Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
report.setCandidate(candidateSeq);
int origNumBases = candidateSeq->getNumBases();
string originalUnaligned = candidateSeq->getUnaligned();
int numBasesNeeded = origNumBases * threshold;
if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
if (candidateSeq->getUnaligned().length()+1 > alignment->getnRows()) {
if (m->debug) { m->mothurOut("[DEBUG]: " + candidateSeq->getName() + " " + toString(candidateSeq->getUnaligned().length()) + " " + toString(alignment->getnRows()) + " \n"); }
alignment->resize(candidateSeq->getUnaligned().length()+2);
}
Sequence temp = templateDB->findClosestSequence(candidateSeq);
Sequence* templateSeq = new Sequence(temp.getName(), temp.getAligned());
float searchScore = templateDB->getSearchScore();
Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
Sequence* copy;
Nast* nast2;
bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below
//since nast does not make a copy of hte sequence passed, and it is used by the reporter below
//you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
//so this bool tells you if you need to delete it
//if there is a possibility that this sequence should be reversed
if (candidateSeq->getNumBases() < numBasesNeeded) {
string wasBetter = "";
//if the user wants you to try the reverse
if (flip) {
//get reverse compliment
copy = new Sequence(candidateSeq->getName(), originalUnaligned);
copy->reverseComplement();
if (m->debug) { m->mothurOut("[DEBUG]: flipping " + candidateSeq->getName() + " \n"); }
//rerun alignment
Sequence temp2 = templateDB->findClosestSequence(copy);
Sequence* templateSeq2 = new Sequence(temp2.getName(), temp2.getAligned());
if (m->debug) { m->mothurOut("[DEBUG]: closest template " + temp2.getName() + " \n"); }
searchScore = templateDB->getSearchScore();
nast2 = new Nast(alignment, copy, templateSeq2);
if (m->debug) { m->mothurOut("[DEBUG]: completed Nast2 " + candidateSeq->getName() + " flipped numBases = " + toString(copy->getNumBases()) + " old numbases = " + toString(candidateSeq->getNumBases()) +" \n"); }
//check if any better
if (copy->getNumBases() > candidateSeq->getNumBases()) {
candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better
delete templateSeq;
templateSeq = templateSeq2;
delete nast;
nast = nast2;
needToDeleteCopy = true;
wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
}else{
wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
delete nast2;
delete templateSeq2;
delete copy;
}
if (m->debug) { m->mothurOut("[DEBUG]: done.\n"); }
}
//create accnos file with names
accnosFile << candidateSeq->getName() << wasBetter << endl;
}
report.setTemplate(templateSeq);
report.setSearchParameters(search, searchScore);
report.setAlignmentParameters(align, alignment);
report.setNastParameters(*nast);
alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
report.print();
delete nast;
delete templateSeq;
if (needToDeleteCopy) { delete copy; }
count++;
}
delete candidateSeq;
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
unsigned long long pos = inFASTA.tellg();
if ((pos == -1) || (pos >= filePos->end)) { break; }
#else
if (inFASTA.eof()) { break; }
#endif
//report progress
if((count) % 100 == 0){ m->mothurOutJustToScreen(toString(count) + "\n"); }
}
//report progress
if((count) % 100 != 0){ m->mothurOutJustToScreen(toString(count) + "\n"); }
delete alignment;
alignmentFile.close();
inFASTA.close();
accnosFile.close();
return count;
}
catch(exception& e) {
m->errorOut(e, "AlignCommand", "driver");
exit(1);
}
}
int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector<unsigned long long>& MPIPos){
try {
string outputString = "";
MPI_Status statusReport;
MPI_Status statusAlign;
MPI_Status statusAccnos;
MPI_Status status;
int pid;
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
NastReport report;
if (pid == 0) {
outputString = report.getHeaders();
int length = outputString.length();
char* buf = new char[length];
memcpy(buf, outputString.c_str(), length);
MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport);
delete buf;
}
Alignment* alignment;
int longestBase = templateDB->getLongestBase();
if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase); }
else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); }
else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }
else if(align == "noalign") { alignment = new NoAlign(); }
else {
m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
m->mothurOutEndLine();
alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
}
for(int i=0;i<num;i++){
if (m->control_pressed) { delete alignment; return 0; }
//read next sequence
int length = MPIPos[start+i+1] - MPIPos[start+i];
char* buf4 = new char[length];
//memcpy(buf4, outputString.c_str(), length);
MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
string tempBuf = buf4;
delete buf4;
if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
istringstream iss (tempBuf,istringstream::in);
Sequence* candidateSeq = new Sequence(iss);
report.setCandidate(candidateSeq);
int origNumBases = candidateSeq->getNumBases();
string originalUnaligned = candidateSeq->getUnaligned();
int numBasesNeeded = origNumBases * threshold;
if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
alignment->resize(candidateSeq->getUnaligned().length()+1);
}
Sequence temp = templateDB->findClosestSequence(candidateSeq);
Sequence* templateSeq = &temp;
float searchScore = templateDB->getSearchScore();
Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
Sequence* copy;
Nast* nast2;
bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below
//since nast does not make a copy of hte sequence passed, and it is used by the reporter below
//you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
//so this bool tells you if you need to delete it
//if there is a possibility that this sequence should be reversed
if (candidateSeq->getNumBases() < numBasesNeeded) {
string wasBetter = "";
//if the user wants you to try the reverse
if (flip) {
//get reverse compliment
copy = new Sequence(candidateSeq->getName(), originalUnaligned);
copy->reverseComplement();
//rerun alignment
Sequence temp2 = templateDB->findClosestSequence(copy);
Sequence* templateSeq2 = &temp2;
searchScore = templateDB->getSearchScore();
nast2 = new Nast(alignment, copy, templateSeq2);
//check if any better
if (copy->getNumBases() > candidateSeq->getNumBases()) {
candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better
templateSeq = templateSeq2;
delete nast;
nast = nast2;
needToDeleteCopy = true;
wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
}else{
wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";
delete nast2;
delete copy;
}
}
//create accnos file with names
outputString = candidateSeq->getName() + wasBetter + "\n";
//send results to parent
int length = outputString.length();
char* buf = new char[length];
memcpy(buf, outputString.c_str(), length);
MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos);
delete buf;
MPIWroteAccnos = true;
}
report.setTemplate(templateSeq);
report.setSearchParameters(search, searchScore);
report.setAlignmentParameters(align, alignment);
report.setNastParameters(*nast);
outputString = ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n";
//send results to parent
int length = outputString.length();
char* buf2 = new char[length];
memcpy(buf2, outputString.c_str(), length);
MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign);
delete buf2;
outputString = report.getReport();
//send results to parent
length = outputString.length();
char* buf3 = new char[length];
memcpy(buf3, outputString.c_str(), length);
MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport);
delete buf3;
delete nast;
if (needToDeleteCopy) { delete copy; }
}
delete candidateSeq;
//report progress
if((i+1) % 100 == 0){ cout << (toString(i+1)) << endl; }
}
//report progress
if((num) % 100 != 0){ cout << (toString(num)) << endl; }
return 1;
}
catch(exception& e) {
m->errorOut(e, "AlignCommand", "driverMPI");
exit(1);
}
}
//**********************************************************************************************************************
void alignDriver(alignStruct* params) {
try {
NastReport report;
ifstream inFASTA;
params->util.openInputFile(params->inputFilename, inFASTA);
inFASTA.seekg(params->filePos.start);
bool done = false;
long long count = 0;
long long numFlipped_0 = 0;
long long numFlipped_1 = 0;
//moved this into driver to avoid deep copies in windows paralellized version
Alignment* alignment;
int longestBase = params->templateDB->getLongestBase();
if (params->m->getDebug()) { params->m->mothurOut("[DEBUG]: template longest base = " + toString(longestBase) + " \n"); }
if(params->alignMethod == "gotoh") { alignment = new GotohOverlap(params->gapOpen, params->gapExtend, params->match, params->misMatch, longestBase); }
else if(params->alignMethod == "needleman") { alignment = new NeedlemanOverlap(params->gapOpen, params->match, params->misMatch, longestBase); }
else if(params->alignMethod == "blast") { alignment = new BlastAlignment(params->gapOpen, params->gapExtend, params->match, params->misMatch); }
else if(params->alignMethod == "noalign") { alignment = new NoAlign(); }
else {
params->m->mothurOut(params->alignMethod + " is not a valid alignment option. I will run the command using needleman.");
params->m->mothurOutEndLine();
alignment = new NeedlemanOverlap(params->gapOpen, params->match, params->misMatch, longestBase);
}
while (!done) {
if (params->m->getControl_pressed()) { break; }
Sequence* candidateSeq = new Sequence(inFASTA); params->util.gobble(inFASTA);
report.setCandidate(candidateSeq);
int origNumBases = candidateSeq->getNumBases();
string originalUnaligned = candidateSeq->getUnaligned();
int numBasesNeeded = origNumBases * params->threshold;
if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
if (candidateSeq->getUnaligned().length()+1 > alignment->getnRows()) {
if (params->m->getDebug()) { params->m->mothurOut("[DEBUG]: " + candidateSeq->getName() + " " + toString(candidateSeq->getUnaligned().length()) + " " + toString(alignment->getnRows()) + " \n"); }
alignment->resize(candidateSeq->getUnaligned().length()+2);
}
float searchScore;
Sequence temp = params->templateDB->findClosestSequence(candidateSeq, searchScore);
Sequence* templateSeq = new Sequence(temp.getName(), temp.getAligned());
Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
Sequence* copy;
Nast* nast2;
bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below
//since nast does not make a copy of hte sequence passed, and it is used by the reporter below
//you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
//so this bool tells you if you need to delete it
//if there is a possibility that this sequence should be reversed
if (candidateSeq->getNumBases() < numBasesNeeded) {
numFlipped_1++;
string wasBetter = "";
//if the user wants you to try the reverse
if (params->flip) {
//get reverse compliment
copy = new Sequence(candidateSeq->getName(), originalUnaligned);
copy->reverseComplement();
if (params->m->getDebug()) { params->m->mothurOut("[DEBUG]: flipping " + candidateSeq->getName() + " \n"); }
//rerun alignment
Sequence temp2 = params->templateDB->findClosestSequence(copy, searchScore);
Sequence* templateSeq2 = new Sequence(temp2.getName(), temp2.getAligned());
if (params->m->getDebug()) { params->m->mothurOut("[DEBUG]: closest template " + temp2.getName() + " \n"); }
nast2 = new Nast(alignment, copy, templateSeq2);
if (params->m->getDebug()) { params->m->mothurOut("[DEBUG]: completed Nast2 " + candidateSeq->getName() + " flipped numBases = " + toString(copy->getNumBases()) + " old numbases = " + toString(candidateSeq->getNumBases()) +" \n"); }
//check if any better
if (copy->getNumBases() > candidateSeq->getNumBases()) {
candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better
delete templateSeq;
templateSeq = templateSeq2;
delete nast;
nast = nast2;
needToDeleteCopy = true;
wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
numFlipped_0++;
}else{
wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
delete nast2;
delete templateSeq2;
delete copy;
}
if (params->m->getDebug()) { params->m->mothurOut("[DEBUG]: done.\n"); }
}
//create accnos file with names
params->accnosWriter->write(candidateSeq->getName() + wasBetter + "\n");
}
report.setTemplate(templateSeq);
report.setSearchParameters(params->search, searchScore);
report.setAlignmentParameters(params->alignMethod, alignment);
report.setNastParameters(*nast);
params->alignWriter->write('>' + candidateSeq->getName() + '\n' + candidateSeq->getAligned() + '\n');
params->reportWriter->write(report.getReport());
delete nast;
delete templateSeq;
if (needToDeleteCopy) { delete copy; }
count++;
}
delete candidateSeq;
#if defined NON_WINDOWS
unsigned long long pos = inFASTA.tellg();
if ((pos == -1) || (pos >= params->filePos.end)) { break; }
#else
if (count == params->filePos.end) { break; }
#endif
//report progress
if((count) % 1000 == 0){ params->m->mothurOutJustToScreen(toString(count) + "\n"); }
}
//report progress
if((count) % 1000 != 0){ params->m->mothurOutJustToScreen(toString(count) + "\n"); }
params->numSeqs += count;
params->flippedResults[0] += numFlipped_0;
params->flippedResults[1] += numFlipped_1;
delete alignment;
inFASTA.close();
}
catch(exception& e) {
params->m->errorOut(e, "AlignCommand", "driver");
exit(1);
}
}
