else {
//valid paramters for this command
string AlignArray[] = {"fasta", "flip", "oligos", "maxambig", "maxhomop", "minlength", "maxlength", "qfile",
- "qthreshold", "qaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
+ "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
it = parameters.find("fasta");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["fasta"] = inputDir + it->second; }
}
it = parameters.find("oligos");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["oligos"] = inputDir + it->second; }
}
it = parameters.find("qfile");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["qfile"] = inputDir + it->second; }
}
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
outputDir = "";
- outputDir += hasPath(fastaFile); //if user entered a file with a path then preserve it
+ outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
}
+
//check for optional parameter and set defaults
// ...at some point should added some additional type checking...
string temp;
temp = validParameter.validFile(parameters, "flip", false);
if (temp == "not found"){ flip = 0; }
- else if(isTrue(temp)) { flip = 1; }
+ else if(m->isTrue(temp)) { flip = 1; }
temp = validParameter.validFile(parameters, "oligos", true);
if (temp == "not found"){ oligoFile = ""; }
temp = validParameter.validFile(parameters, "maxlength", false); if (temp == "not found") { temp = "0"; }
convert(temp, maxLength);
-
temp = validParameter.validFile(parameters, "bdiffs", false); if (temp == "not found") { temp = "0"; }
convert(temp, bdiffs);
temp = validParameter.validFile(parameters, "qthreshold", false); if (temp == "not found") { temp = "0"; }
convert(temp, qThreshold);
- temp = validParameter.validFile(parameters, "qtrim", false); if (temp == "not found") { temp = "F"; }
- qtrim = isTrue(temp);
+ temp = validParameter.validFile(parameters, "qtrim", false); if (temp == "not found") { temp = "F"; }
+ qtrim = m->isTrue(temp);
+
+ temp = validParameter.validFile(parameters, "rollaverage", false); if (temp == "not found") { temp = "0"; }
+ convert(temp, qRollAverage);
+
+ temp = validParameter.validFile(parameters, "qwindowaverage", false);if (temp == "not found") { temp = "0"; }
+ convert(temp, qWindowAverage);
+
+ temp = validParameter.validFile(parameters, "qwindowsize", false); if (temp == "not found") { temp = "100"; }
+ convert(temp, qWindowSize);
+
+ temp = validParameter.validFile(parameters, "qstepsize", false); if (temp == "not found") { temp = "10"; }
+ convert(temp, qWindowStep);
temp = validParameter.validFile(parameters, "qaverage", false); if (temp == "not found") { temp = "0"; }
convert(temp, qAverage);
temp = validParameter.validFile(parameters, "allfiles", false); if (temp == "not found") { temp = "F"; }
- allFiles = isTrue(temp);
+ allFiles = m->isTrue(temp);
- temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found") { temp = "1"; }
convert(temp, processors);
if(allFiles && oligoFile == ""){
numFPrimers = 0; //this needs to be initialized
numRPrimers = 0;
- string trimSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "trim.fasta";
+ string trimSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.fasta";
outputNames.push_back(trimSeqFile);
- string scrapSeqFile = outputDir + getRootName(getSimpleName(fastaFile)) + "scrap.fasta";
+ string scrapSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.fasta";
outputNames.push_back(scrapSeqFile);
- string groupFile = outputDir + getRootName(getSimpleName(fastaFile)) + "groups";
+ string trimQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.qual";
+ outputNames.push_back(trimQualFile);
+ string scrapQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.qual";
+ outputNames.push_back(scrapQualFile);
+ string groupFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "groups";
vector<string> fastaFileNames;
+ vector<string> qualFileNames;
if(oligoFile != ""){
outputNames.push_back(groupFile);
- getOligos(fastaFileNames);
+ getOligos(fastaFileNames, qualFileNames);
}
-
- if(qFileName != "") { setLines(qFileName, qLines); }
+ vector<unsigned long int> fastaFilePos;
+ vector<unsigned long int> qFilePos;
+
+ setLines(fastaFile, qFileName, fastaFilePos, qFilePos);
+
+ for (int i = 0; i < (fastaFilePos.size()-1); i++) {
+ lines.push_back(new linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
+ if (qFileName != "") { qLines.push_back(new linePair(qFilePos[i], qFilePos[(i+1)])); }
+ }
+ if(qFileName == "") { qLines = lines; } //files with duds
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
- ifstream inFASTA;
- int numSeqs;
- openInputFile(fastaFile, inFASTA);
- getNumSeqs(inFASTA, numSeqs);
- inFASTA.close();
-
- lines.push_back(new linePair(0, numSeqs));
-
- driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
+
+ driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames, lines[0], qLines[0]);
for (int j = 0; j < fastaFileNames.size(); j++) {
rename((fastaFileNames[j] + toString(getpid()) + ".temp").c_str(), fastaFileNames[j].c_str());
}
+ if(qFileName != ""){
+ for (int j = 0; j < qualFileNames.size(); j++) {
+ rename((qualFileNames[j] + toString(getpid()) + ".temp").c_str(), qualFileNames[j].c_str());
+ }
+ }
}else{
- setLines(fastaFile, lines);
- if(qFileName == "") { qLines = lines; }
-
- createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames);
+ createProcessesCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames);
rename((trimSeqFile + toString(processIDS[0]) + ".temp").c_str(), trimSeqFile.c_str());
rename((scrapSeqFile + toString(processIDS[0]) + ".temp").c_str(), scrapSeqFile.c_str());
rename((groupFile + toString(processIDS[0]) + ".temp").c_str(), groupFile.c_str());
+
+ if(qFileName != ""){
+ rename((trimQualFile + toString(processIDS[0]) + ".temp").c_str(), trimQualFile.c_str());
+ rename((scrapQualFile + toString(processIDS[0]) + ".temp").c_str(), scrapQualFile.c_str());
+ }
+
+
for (int j = 0; j < fastaFileNames.size(); j++) {
rename((fastaFileNames[j] + toString(processIDS[0]) + ".temp").c_str(), fastaFileNames[j].c_str());
}
+ if(qFileName != ""){
+ for (int j = 0; j < qualFileNames.size(); j++) {
+ rename((qualFileNames[j] + toString(getpid()) + ".temp").c_str(), qualFileNames[j].c_str());
+ }
+ }
+
//append files
for(int i=1;i<processors;i++){
- appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
+ m->appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
remove((trimSeqFile + toString(processIDS[i]) + ".temp").c_str());
- appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
+ m->appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
remove((scrapSeqFile + toString(processIDS[i]) + ".temp").c_str());
- appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
+
+ m->appendFiles((trimQualFile + toString(processIDS[i]) + ".temp"), trimQualFile);
+ remove((trimQualFile + toString(processIDS[i]) + ".temp").c_str());
+ m->appendFiles((scrapQualFile + toString(processIDS[i]) + ".temp"), scrapQualFile);
+ remove((scrapQualFile + toString(processIDS[i]) + ".temp").c_str());
+
+ m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
for (int j = 0; j < fastaFileNames.size(); j++) {
- appendFiles((fastaFileNames[j] + toString(processIDS[i]) + ".temp"), fastaFileNames[j]);
+ m->appendFiles((fastaFileNames[j] + toString(processIDS[i]) + ".temp"), fastaFileNames[j]);
remove((fastaFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
}
+
+ if(qFileName != ""){
+ for (int j = 0; j < qualFileNames.size(); j++) {
+ m->appendFiles((qualFileNames[j] + toString(processIDS[i]) + ".temp"), qualFileNames[j]);
+ remove((qualFileNames[j] + toString(processIDS[i]) + ".temp").c_str());
+ }
+ }
+
+
}
}
if (m->control_pressed) { return 0; }
#else
- ifstream inFASTA;
- int numSeqs;
- openInputFile(fastaFile, inFASTA);
- getNumSeqs(inFASTA, numSeqs);
- inFASTA.close();
-
- lines.push_back(new linePair(0, numSeqs));
-
- driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, groupFile, fastaFileNames, lines[0], lines[0]);
+ driverCreateTrim(fastaFile, qFileName, trimSeqFile, scrapSeqFile, trimQualFile, scrapQualFile, groupFile, fastaFileNames, qualFileNames, lines[0], qLines[0]);
+ for (int j = 0; j < fastaFileNames.size(); j++) {
+ rename((fastaFileNames[j] + toString(j) + ".temp").c_str(), fastaFileNames[j].c_str());
+ }
+ if(qFileName != ""){
+ for (int j = 0; j < qualFileNames.size(); j++) {
+ rename((qualFileNames[j] + toString(j) + ".temp").c_str(), qualFileNames[j].c_str());
+ }
+ }
+
if (m->control_pressed) { return 0; }
#endif
for(int i=0;i<fastaFileNames.size();i++){
- if (isBlank(fastaFileNames[i])) { remove(fastaFileNames[i].c_str()); }
+ if (m->isBlank(fastaFileNames[i])) { remove(fastaFileNames[i].c_str()); }
+ else if (filesToRemove.count(fastaFileNames[i]) > 0) { remove(fastaFileNames[i].c_str()); }
else {
ifstream inFASTA;
string seqName;
- //openInputFile(getRootName(fastaFile) + groupVector[i] + ".fasta", inFASTA);
- openInputFile(fastaFileNames[i], inFASTA);
+ m->openInputFile(fastaFileNames[i], inFASTA);
ofstream outGroups;
- string outGroupFilename = outputDir + getRootName(getSimpleName(fastaFileNames[i])) + "groups";
- openOutputFile(outGroupFilename, outGroups);
- //openOutputFile(outputDir + getRootName(getSimpleName(fastaFile)) + groupVector[i] + ".groups", outGroups);
+ string outGroupFilename = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + "groups";
+ m->openOutputFile(outGroupFilename, outGroups);
outputNames.push_back(outGroupFilename);
string thisGroup = "";
}
}
+ if(qFileName != ""){
+ for(int i=0;i<qualFileNames.size();i++){
+ if (m->isBlank(qualFileNames[i])) { remove(qualFileNames[i].c_str()); }
+ else if (filesToRemove.count(qualFileNames[i]) > 0) { remove(qualFileNames[i].c_str()); }
+ else {
+ ifstream inQual;
+ string seqName;
+ m->openInputFile(qualFileNames[i], inQual);
+// ofstream outGroups;
+//
+// string thisGroup = "";
+// if (i > comboStarts) {
+// map<string, int>::iterator itCombo;
+// for(itCombo=combos.begin();itCombo!=combos.end(); itCombo++){
+// if(itCombo->second == i){ thisGroup = itCombo->first; combos.erase(itCombo); break; }
+// }
+// }
+// else{ thisGroup = groupVector[i]; }
+
+ inQual.close();
+ }
+ }
+ }
+
+
if (m->control_pressed) {
for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
return 0;
}
/**************************************************************************************/
-int TrimSeqsCommand::driverCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames, linePair* line, linePair* qline) {
+
+int TrimSeqsCommand::driverCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames, linePair* line, linePair* qline) {
+
try {
ofstream outFASTA;
- int able = openOutputFile(trimFile, outFASTA);
+ int able = m->openOutputFile(trimFile, outFASTA);
ofstream scrapFASTA;
- openOutputFile(scrapFile, scrapFASTA);
+ m->openOutputFile(scrapFile, scrapFASTA);
+
+ ofstream outQual;
+ ofstream scrapQual;
+ if(qFileName != ""){
+ m->openOutputFile(trimQFile, outQual);
+ m->openOutputFile(scrapQFile, scrapQual);
+ }
ofstream outGroups;
vector<ofstream*> fastaFileNames;
+ vector<ofstream*> qualFileNames;
+
+ cout << "here" << endl;
if (oligoFile != "") {
- openOutputFile(groupFile, outGroups);
+ m->openOutputFile(groupFile, outGroups);
for (int i = 0; i < fastaNames.size(); i++) {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
+ if(qFileName != ""){
+ qualFileNames.push_back(new ofstream((qualNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
+ }
#else
fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(i) + ".temp").c_str(), ios::ate));
+ if(qFileName != ""){
+ qualFileNames.push_back(new ofstream((qualNames[i] + toString(i) + ".temp").c_str(), ios::ate));
+ }
#endif
}
}
-
+cout << "here " << filename << endl;
ifstream inFASTA;
- openInputFile(filename, inFASTA);
-
- ifstream qFile;
- if(qFileName != "") { openInputFile(qFileName, qFile); }
-
- qFile.seekg(qline->start);
+ m->openInputFile(filename, inFASTA);
inFASTA.seekg(line->start);
+ cout << "here " << qFileName << endl;
+ ifstream qFile;
+ if(qFileName != "") { m->openInputFile(qFileName, qFile); qFile.seekg(qline->start); }
- for(int i=0;i<line->num;i++){
+ bool done = false;
+ int count = 0;
+ cout << "here" << endl;
+ while (!done) {
if (m->control_pressed) {
- inFASTA.close(); outFASTA.close(); scrapFASTA.close(); if (oligoFile != "") { outGroups.close(); } if(qFileName != "") { qFile.close(); }
+ inFASTA.close(); outFASTA.close(); scrapFASTA.close();
+ if (oligoFile != "") { outGroups.close(); }
+
for(int i=0;i<fastaFileNames.size();i++){ fastaFileNames[i]->close(); delete fastaFileNames[i]; }
+
+ if(qFileName != ""){
+ qFile.close();
+ for(int i=0;i<qualFileNames.size();i++){ qualFileNames[i]->close(); delete qualFileNames[i]; }
+ }
for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
+
return 0;
}
int success = 1;
- Sequence currSeq(inFASTA);
- cout << i << '\t' << currSeq.getName() << endl;
+ Sequence currSeq(inFASTA); m->gobble(inFASTA);
+
+ QualityScores currQual;
+ if(qFileName != ""){
+ currQual = QualityScores(qFile, currSeq.getNumBases()); m->gobble(qFile);
+ }
+
string origSeq = currSeq.getUnaligned();
if (origSeq != "") {
int groupBar, groupPrime;
string trashCode = "";
int currentSeqsDiffs = 0;
-
+
if(qFileName != ""){
- if(qThreshold != 0) { success = stripQualThreshold(currSeq, qFile); }
- else if(qAverage != 0) { success = cullQualAverage(currSeq, qFile); }
-
+ if(qThreshold != 0) { success = currQual.stripQualThreshold(currSeq, qThreshold); }
+ else if(qAverage != 0) { success = currQual.cullQualAverage(currSeq, qAverage); }
+ else if(qRollAverage != 0) { success = currQual.stripQualRollingAverage(currSeq, qRollAverage); }
+ else if(qWindowAverage != 0){ success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage); }
+
if (qtrim == 1 && (origSeq.length() != currSeq.getUnaligned().length())) {
success = 0; //if you don't want to trim and the sequence does not meet quality requirements, move to scrap
}
- if(!success) { trashCode += 'q'; }
+ if(!success) { trashCode += 'q'; }
}
if(barcodes.size() != 0){
- success = stripBarcode(currSeq, groupBar);
- if(success > bdiffs){ trashCode += 'b'; }
+ success = stripBarcode(currSeq, currQual, groupBar);
+ if(success > bdiffs) { trashCode += 'b'; }
else{ currentSeqsDiffs += success; }
}
if(numFPrimers != 0){
- success = stripForward(currSeq, groupPrime);
- if(success > pdiffs){ trashCode += 'f'; }
+ success = stripForward(currSeq, currQual, groupPrime);
+ if(success > pdiffs) { trashCode += 'f'; }
else{ currentSeqsDiffs += success; }
}
- if (currentSeqsDiffs > tdiffs) { trashCode += 't'; }
+ if (currentSeqsDiffs > tdiffs) { trashCode += 't'; }
if(numRPrimers != 0){
- success = stripReverse(currSeq);
- if(!success){ trashCode += 'r'; }
+ success = stripReverse(currSeq, currQual);
+ if(!success) { trashCode += 'r'; }
}
if(minLength > 0 || maxLength > 0){
success = cullLength(currSeq);
- if(!success){ trashCode += 'l'; }
+ if(!success) { trashCode += 'l'; }
}
if(maxHomoP > 0){
success = cullHomoP(currSeq);
- if(!success){ trashCode += 'h'; }
+ if(!success) { trashCode += 'h'; }
}
if(maxAmbig != -1){
success = cullAmbigs(currSeq);
- if(!success){ trashCode += 'n'; }
+ if(!success) { trashCode += 'n'; }
}
- if(flip){ currSeq.reverseComplement(); } // should go last
+ if(flip){ currSeq.reverseComplement(); currQual.flipQScores(); } // should go last
if(trashCode.length() == 0){
currSeq.setAligned(currSeq.getUnaligned());
currSeq.printSequence(outFASTA);
+ currQual.printQScores(outQual);
+
if(barcodes.size() != 0){
string thisGroup = groupVector[groupBar];
int indexToFastaFile = groupBar;
}
}
outGroups << currSeq.getName() << '\t' << thisGroup << endl;
-
if(allFiles){
- currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
+ currSeq.printSequence(*fastaFileNames[indexToFastaFile]);
+
+ if(qFileName != ""){
+ currQual.printQScores(*qualFileNames[indexToFastaFile]);
+ }
}
}
}
currSeq.setUnaligned(origSeq);
currSeq.setAligned(origSeq);
currSeq.printSequence(scrapFASTA);
+ currQual.printQScores(scrapQual);
}
+ count++;
}
- gobble(inFASTA);
+
+ unsigned long int pos = inFASTA.tellg();
+ if ((pos == -1) || (pos >= line->end)) { break; }
+
+ //report progress
+ if((count) % 1000 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
+
}
+ //report progress
+ if((count) % 1000 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
+
inFASTA.close();
outFASTA.close();
scrapFASTA.close();
if (oligoFile != "") { outGroups.close(); }
- if(qFileName != "") { qFile.close(); }
+ if(qFileName != "") { qFile.close(); scrapQual.close(); outQual.close(); }
for(int i=0;i<fastaFileNames.size();i++){
fastaFileNames[i]->close();
delete fastaFileNames[i];
}
- return 0;
+ if(qFileName != ""){
+ for(int i=0;i<qualFileNames.size();i++){
+ qualFileNames[i]->close();
+ delete qualFileNames[i];
+ }
+ }
+
+ return count;
}
catch(exception& e) {
m->errorOut(e, "TrimSeqsCommand", "driverCreateTrim");
exit(1);
}
}
+
/**************************************************************************************************/
-int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string groupFile, vector<string> fastaNames) {
+
+int TrimSeqsCommand::createProcessesCreateTrim(string filename, string qFileName, string trimFile, string scrapFile, string trimQFile, string scrapQFile, string groupFile, vector<string> fastaNames, vector<string> qualNames) {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
process++;
}else if (pid == 0){
- driverCreateTrim(filename, qFileName, (trimFile + toString(getpid()) + ".temp"), (scrapFile + toString(getpid()) + ".temp"), (groupFile + toString(getpid()) + ".temp"), fastaNames, lines[process], qLines[process]);
+ driverCreateTrim(filename, qFileName, (trimFile + toString(getpid()) + ".temp"), (scrapFile + toString(getpid()) + ".temp"), (trimQFile + toString(getpid()) + ".temp"), (scrapQFile + toString(getpid()) + ".temp"), (groupFile + toString(getpid()) + ".temp"), fastaNames, qualNames, lines[process], qLines[process]);
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
exit(1);
}
}
+
/**************************************************************************************************/
-int TrimSeqsCommand::setLines(string filename, vector<linePair*>& lines) {
+int TrimSeqsCommand::setLines(string filename, string qfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos) {
try {
- lines.clear();
+ //set file positions for fasta file
+ fastaFilePos = m->divideFile(filename, processors);
- vector<long int> positions;
+ if (qfilename == "") { return processors; }
- ifstream inFASTA;
- openInputFile(filename, inFASTA);
+ //get name of first sequence in each chunk
+ map<string, int> firstSeqNames;
+ for (int i = 0; i < (fastaFilePos.size()-1); i++) {
+ ifstream in;
+ m->openInputFile(filename, in);
+ in.seekg(fastaFilePos[i]);
+
+ Sequence temp(in);
+ firstSeqNames[temp.getName()] = i;
+
+ in.close();
+ }
+
+ //seach for filePos of each first name in the qfile and save in qfileFilePos
+ ifstream inQual;
+ m->openInputFile(qfilename, inQual);
string input;
- while(!inFASTA.eof()){
- input = getline(inFASTA);
+ while(!inQual.eof()){
+ input = m->getline(inQual);
if (input.length() != 0) {
- if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
+ if(input[0] == '>'){ //this is a sequence name line
+ istringstream nameStream(input);
+
+ string sname = ""; nameStream >> sname;
+ sname = sname.substr(1);
+
+ map<string, int>::iterator it = firstSeqNames.find(sname);
+
+ if(it != firstSeqNames.end()) { //this is the start of a new chunk
+ unsigned long int pos = inQual.tellg();
+ qfileFilePos.push_back(pos - input.length() - 1);
+ firstSeqNames.erase(it);
+ }
+ }
}
+
+ if (firstSeqNames.size() == 0) { break; }
}
- inFASTA.close();
+ inQual.close();
- int numFastaSeqs = positions.size();
-
+ if (firstSeqNames.size() != 0) {
+ for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
+ m->mothurOut(it->first + " is in your fasta file and not in your quality file, not using quality file."); m->mothurOutEndLine();
+ }
+ qFileName = "";
+ return processors;
+ }
+
+ //get last file position of qfile
FILE * pFile;
- long size;
+ unsigned long int size;
//get num bytes in file
- pFile = fopen (filename.c_str(),"rb");
+ pFile = fopen (qfilename.c_str(),"rb");
if (pFile==NULL) perror ("Error opening file");
else{
fseek (pFile, 0, SEEK_END);
fclose (pFile);
}
- int numSeqsPerProcessor = numFastaSeqs / processors;
-
- for (int i = 0; i < processors; i++) {
-
- long int startPos = positions[ i * numSeqsPerProcessor ];
- if(i == processors - 1){
- numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
- }else{
- long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
- }
- lines.push_back(new linePair(startPos, numSeqsPerProcessor));
- }
+ qfileFilePos.push_back(size);
- return numFastaSeqs;
+ return processors;
}
catch(exception& e) {
m->errorOut(e, "TrimSeqsCommand", "setLines");
}
//***************************************************************************************************************
-void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec){ //vector<ofstream*>& outFASTAVec
+void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
try {
ifstream inOligos;
- openInputFile(oligoFile, inOligos);
+ m->openInputFile(oligoFile, inOligos);
ofstream test;
groupVector.push_back(group);
if(allFiles){
- if (group != "") { //there is a group for this primer
- outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
- outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
+ outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
+ if(qFileName != ""){
+ outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ }
+ if (group == "") { //if there is not a group for this primer, then this file will not get written to, but we add it to keep the indexes correct
+ filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
+ if(qFileName != ""){
+ filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ }
+ }else {
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
+ if(qFileName != ""){
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ }
}
}
groupVector.push_back(group);
if(allFiles){
- outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
- outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(index) + "." + group + ".fasta"));
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
+ outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
+ if(qFileName != ""){
+ outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ }
}
}else{ m->mothurOut(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine(); }
}
- gobble(inOligos);
+ m->gobble(inOligos);
}
inOligos.close();
for (map<string, int>::iterator itBar = barcodes.begin(); itBar != barcodes.end(); itBar++) {
for (map<string, int>::iterator itPrime = primers.begin(); itPrime != primers.end(); itPrime++) {
if (groupVector[itPrime->second] != "") { //there is a group for this primer
- outputNames.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(itBar->second) + "." + groupVector[itBar->second] + "." + toString(itPrime->second) + "." + groupVector[itPrime->second] + ".fasta"));
- outFASTAVec.push_back((outputDir + getRootName(getSimpleName(fastaFile)) + toString(itBar->second) + "." + groupVector[itBar->second] + "." + toString(itPrime->second) + "." + groupVector[itPrime->second] + ".fasta"));
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
+ outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".fasta"));
combos[(groupVector[itBar->second] + "." + groupVector[itPrime->second])] = outFASTAVec.size()-1;
+
+ if(qFileName != ""){
+ outQualVec.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupVector[itBar->second] + "." + groupVector[itPrime->second] + ".qual"));
+ }
}
}
}
}
//***************************************************************************************************************
-int TrimSeqsCommand::stripBarcode(Sequence& seq, int& group){
+int TrimSeqsCommand::stripBarcode(Sequence& seq, QualityScores& qual, int& group){
try {
string rawSequence = seq.getUnaligned();
if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
group = it->second;
seq.setUnaligned(rawSequence.substr(oligo.length()));
+
+ if(qual.getName() != ""){
+ qual.trimQScores(oligo.length(), -1);
+ }
+
success = 0;
break;
}
else{ //use the best match
group = minGroup;
seq.setUnaligned(rawSequence.substr(minPos));
+
+ if(qual.getName() != ""){
+ qual.trimQScores(minPos, -1);
+ }
success = minDiff;
}
//***************************************************************************************************************
-int TrimSeqsCommand::stripForward(Sequence& seq, int& group){
+int TrimSeqsCommand::stripForward(Sequence& seq, QualityScores& qual, int& group){
try {
string rawSequence = seq.getUnaligned();
int success = pdiffs + 1; //guilty until proven innocent
if(compareDNASeq(oligo, rawSequence.substr(0,oligo.length()))){
group = it->second;
seq.setUnaligned(rawSequence.substr(oligo.length()));
+ if(qual.getName() != ""){
+ qual.trimQScores(oligo.length(), -1);
+
+ }
success = 0;
break;
}
else{ //use the best match
group = minGroup;
seq.setUnaligned(rawSequence.substr(minPos));
+ if(qual.getName() != ""){
+ qual.trimQScores(minPos, -1);
+ }
success = minDiff;
}
//***************************************************************************************************************
-bool TrimSeqsCommand::stripReverse(Sequence& seq){
+bool TrimSeqsCommand::stripReverse(Sequence& seq, QualityScores& qual){
try {
string rawSequence = seq.getUnaligned();
bool success = 0; //guilty until proven innocent
if(compareDNASeq(oligo, rawSequence.substr(rawSequence.length()-oligo.length(),oligo.length()))){
seq.setUnaligned(rawSequence.substr(0,rawSequence.length()-oligo.length()));
+ if(qual.getName() != ""){
+ qual.trimQScores(-1, rawSequence.length()-oligo.length());
+ }
success = 1;
break;
}
else if(oligo[i] == 'B' && (seq[i] != 'C' && seq[i] != 'T' && seq[i] != 'G')) { countDiffs++; }
else if(oligo[i] == 'D' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'G')) { countDiffs++; }
else if(oligo[i] == 'H' && (seq[i] != 'A' && seq[i] != 'T' && seq[i] != 'C')) { countDiffs++; }
- else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G')) { countDiffs++; }
-
+ else if(oligo[i] == 'V' && (seq[i] != 'A' && seq[i] != 'C' && seq[i] != 'G')) { countDiffs++; }
}
+
}
return countDiffs;
}
//***************************************************************************************************************
-bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
- try {
+//bool TrimSeqsCommand::stripQualThreshold(Sequence& seq, ifstream& qFile){
+// try {
+//
// string rawSequence = seq.getUnaligned();
-// int seqLength; // = rawSequence.length();
-// string name, temp, temp2;
+// int seqLength = seq.getNumBases();
+// bool success = 0; //guilty until proven innocent
+// string name;
//
// qFile >> name;
+// if (name[0] == '>') { if(name.substr(1) != seq.getName()) { m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); } }
+//
+// while (!qFile.eof()) { char c = qFile.get(); if (c == 10 || c == 13){ break; } }
//
-// //get rest of line
-// temp = "";
-// while (!qFile.eof()) {
-// char c = qFile.get();
-// if (c == 10 || c == 13){ break; }
-// else { temp += c; }
-// }
-//
-// int pos = temp.find("length");
-// if (pos == temp.npos) { m->mothurOut("Cannot find length in qfile for " + seq.getName()); m->mothurOutEndLine(); seqLength = 0; }
-// else {
-// string tempLength = temp.substr(pos);
-// istringstream iss (tempLength,istringstream::in);
-// iss >> temp;
+// int score;
+// int end = seqLength;
+//
+// for(int i=0;i<seqLength;i++){
+// qFile >> score;
+//
+// if(score < qThreshold){
+// end = i;
+// break;
+// }
+// }
+// for(int i=end+1;i<seqLength;i++){
+// qFile >> score;
// }
//
-// splitAtEquals(temp2, temp); //separates length=242, temp=length, temp2=242
-// convert(temp, seqLength); //converts string to int
-//
-// if (name.length() != 0) { if(name.substr(1) != seq.getName()) { m->mothurOut("sequence name mismatch btwn fasta and qual file"); m->mothurOutEndLine(); } }
-
- string rawSequence = seq.getUnaligned();
- int seqLength = seq.getNumBases();
- bool success = 0; //guilty until proven innocent
- string name;
-
- qFile >> name;
- if (name[0] == '>') { if(name.substr(1) != seq.getName()) { m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); } }
-
- while (!qFile.eof()) { char c = qFile.get(); if (c == 10 || c == 13){ break; } }
-
- int score;
- int end = seqLength;
-
- for(int i=0;i<seqLength;i++){
- qFile >> score;
-
- if(score < qThreshold){
- end = i;
- break;
- }
- }
- for(int i=end+1;i<seqLength;i++){
- qFile >> score;
- }
-
- seq.setUnaligned(rawSequence.substr(0,end));
-
- return 1;
- }
- catch(exception& e) {
- m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
- exit(1);
- }
-}
+// seq.setUnaligned(rawSequence.substr(0,end));
+//
+// return 1;
+// }
+// catch(exception& e) {
+// m->errorOut(e, "TrimSeqsCommand", "stripQualThreshold");
+// exit(1);
+// }
+//}
//***************************************************************************************************************
-bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
- try {
- string rawSequence = seq.getUnaligned();
- int seqLength = seq.getNumBases();
- bool success = 0; //guilty until proven innocent
- string name;
-
- qFile >> name;
- if (name[0] == '>') { if(name.substr(1) != seq.getName()) { m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); } }
-
- while (!qFile.eof()) { char c = qFile.get(); if (c == 10 || c == 13){ break; } }
-
- float score;
- float average = 0;
-
- for(int i=0;i<seqLength;i++){
- qFile >> score;
- average += score;
- }
- average /= seqLength;
-
- if(average >= qAverage) { success = 1; }
- else { success = 0; }
-
- return success;
- }
- catch(exception& e) {
- m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
- exit(1);
- }
-}
+//bool TrimSeqsCommand::cullQualAverage(Sequence& seq, ifstream& qFile){
+// try {
+// string rawSequence = seq.getUnaligned();
+// int seqLength = seq.getNumBases();
+// bool success = 0; //guilty until proven innocent
+// string name;
+//
+// qFile >> name;
+// if (name[0] == '>') { if(name.substr(1) != seq.getName()) { m->mothurOut("sequence name mismatch btwn fasta: " + seq.getName() + " and qual file: " + name); m->mothurOutEndLine(); } }
+//
+// while (!qFile.eof()) { char c = qFile.get(); if (c == 10 || c == 13){ break; } }
+//
+// float score;
+// float average = 0;
+//
+// for(int i=0;i<seqLength;i++){
+// qFile >> score;
+// average += score;
+// }
+// average /= seqLength;
+//
+// if(average >= qAverage) { success = 1; }
+// else { success = 0; }
+//
+// return success;
+// }
+// catch(exception& e) {
+// m->errorOut(e, "TrimSeqsCommand", "cullQualAverage");
+// exit(1);
+// }
+//}
//***************************************************************************************************************
projects / mothur.git / blobdiff