#include "trimoligos.h"
+
//**********************************************************************************************************************
vector<string> TrimSeqsCommand::setParameters(){
try {
CommandParameter preorient("checkorient", "Boolean", "", "F", "", "", "","",false,false,true); parameters.push_back(preorient);
CommandParameter pmaxambig("maxambig", "Number", "", "-1", "", "", "","",false,false); parameters.push_back(pmaxambig);
CommandParameter pmaxhomop("maxhomop", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pmaxhomop);
- CommandParameter pminlength("minlength", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pminlength);
+ CommandParameter pminlength("minlength", "Number", "", "1", "", "", "","",false,false); parameters.push_back(pminlength);
CommandParameter pmaxlength("maxlength", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pmaxlength);
CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(ppdiffs);
CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(pbdiffs);
CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
CommandParameter pallfiles("allfiles", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pallfiles);
CommandParameter pkeepforward("keepforward", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pkeepforward);
+ CommandParameter plogtransform("logtransform", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(plogtransform);
CommandParameter pqtrim("qtrim", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pqtrim);
CommandParameter pqthreshold("qthreshold", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pqthreshold);
CommandParameter pqaverage("qaverage", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pqaverage);
string helpString = "";
helpString += "The trim.seqs command reads a fastaFile and creates 2 new fasta files, .trim.fasta and scrap.fasta, as well as group files if you provide and oligos file.\n";
helpString += "The .trim.fasta contains sequences that meet your requirements, and the .scrap.fasta contains those which don't.\n";
- helpString += "The trim.seqs command parameters are fasta, name, count, flip, checkorient, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim, keepfirst, removelast and allfiles.\n";
+ helpString += "The trim.seqs command parameters are fasta, name, count, flip, checkorient, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim, keepfirst, removelast, logtransform and allfiles.\n";
helpString += "The fasta parameter is required.\n";
helpString += "The flip parameter will output the reverse compliment of your trimmed sequence. The default is false.\n";
- helpString += "The checkorient parameter will check the reverse compliment of the sequence if the barcodes and primers cannot be found in the forward. The default is false.\n";
+ helpString += "The checkorient parameter will check look for the reverse compliment of the barcode or primer in the sequence. If found the sequence is flipped. The default is false.\n";
helpString += "The oligos parameter allows you to provide an oligos file.\n";
helpString += "The name parameter allows you to provide a names file with your fasta file.\n";
helpString += "The count parameter allows you to provide a count file with your fasta file.\n";
helpString += "The qwindowaverage parameter allows you to set a minimum average quality score allowed over a window. \n";
helpString += "The rollaverage parameter allows you to set a minimum rolling average quality score allowed over a window. \n";
helpString += "The qstepsize parameter allows you to set a number of bases to move the window over. Default=1.\n";
+ helpString += "The logtransform parameter allows you to indicate you want the averages for the qwindowaverage, rollaverage and qaverage to be calculated using a logtransform. Default=F.\n";
helpString += "The allfiles parameter will create separate group and fasta file for each grouping. The default is F.\n";
helpString += "The keepforward parameter allows you to indicate whether you want the forward primer removed or not. The default is F, meaning remove the forward primer.\n";
helpString += "The qtrim parameter will trim sequence from the point that they fall below the qthreshold and put it in the .trim file if set to true. The default is T.\n";
temp = validParameter.validFile(parameters, "maxhomop", false); if (temp == "not found") { temp = "0"; }
m->mothurConvert(temp, maxHomoP);
- temp = validParameter.validFile(parameters, "minlength", false); if (temp == "not found") { temp = "0"; }
+ temp = validParameter.validFile(parameters, "minlength", false); if (temp == "not found") { temp = "1"; }
m->mothurConvert(temp, minLength);
temp = validParameter.validFile(parameters, "maxlength", false); if (temp == "not found") { temp = "0"; }
temp = validParameter.validFile(parameters, "keepforward", false); if (temp == "not found") { temp = "F"; }
keepforward = m->isTrue(temp);
+ temp = validParameter.validFile(parameters, "logtransform", false); if (temp == "not found") { temp = "F"; }
+ logtransform = m->isTrue(temp);
+
temp = validParameter.validFile(parameters, "checkorient", false); if (temp == "not found") { temp = "F"; }
reorient = m->isTrue(temp);
vector<vector<string> > fastaFileNames;
vector<vector<string> > qualFileNames;
vector<vector<string> > nameFileNames;
+ map<string, string> uniqueFastaNames;// so we don't add the same groupfile multiple times
map<string, string> variables;
variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFile));
if (countfile != "") {
CountTable ct;
- ct.readTable(countfile);
+ ct.readTable(countfile, true, false);
nameCount = ct.getNameMap();
outputNames.push_back(trimCountFile);
outputNames.push_back(scrapCountFile);
string outputGroupFileName;
if(oligoFile != ""){
- createGroup = getOligos(fastaFileNames, qualFileNames, nameFileNames);
+ createGroup = getOligos(fastaFileNames, qualFileNames, nameFileNames, uniqueFastaNames);
if ((createGroup) && (countfile == "")){
map<string, string> myvariables;
myvariables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFile));
outputNames.push_back(outputGroupFileName); outputTypes["group"].push_back(outputGroupFileName);
}
}
-
+
if (m->control_pressed) { return 0; }
//fills lines and qlines
if (m->control_pressed) { return 0; }
if(allFiles){
- map<string, string> uniqueFastaNames;// so we don't add the same groupfile multiple times
map<string, string>::iterator it;
set<string> namesToRemove;
for(int i=0;i<fastaFileNames.size();i++){
m->mothurRemove(nameFileNames[i][j]);
namesToRemove.insert(nameFileNames[i][j]);
}
- }else{
- it = uniqueFastaNames.find(fastaFileNames[i][j]);
- if (it == uniqueFastaNames.end()) {
- uniqueFastaNames[fastaFileNames[i][j]] = barcodeNameVector[i];
- }
+ uniqueFastaNames.erase(fastaFileNames[i][j]); //remove from list for group file print
}
}
}
if (countfile != "") { //create countfile with group info included
CountTable* ct = new CountTable();
- ct->readTable(trimCountFile);
+ ct->readTable(trimCountFile, true, false);
map<string, int> justTrimmedNames = ct->getNameMap();
delete ct;
int count = 0;
bool moreSeqs = 1;
- int numBarcodes = barcodes.size();
TrimOligos* trimOligos = NULL;
- if (pairedOligos) { trimOligos = new TrimOligos(pdiffs, bdiffs, 0, 0, pairedPrimers, pairedBarcodes); numBarcodes = pairedBarcodes.size(); }
- else { trimOligos = new TrimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, revPrimer, linker, spacer); }
+ if (pairedOligos) { trimOligos = new TrimOligos(pdiffs, bdiffs, 0, 0, oligos.getPairedPrimers(), oligos.getPairedBarcodes()); }
+ else { trimOligos = new TrimOligos(pdiffs, bdiffs, ldiffs, sdiffs, oligos.getPrimers(), oligos.getBarcodes(), oligos.getReversePrimers(), oligos.getLinkers(), oligos.getSpacers()); }
TrimOligos* rtrimOligos = NULL;
if (reorient) {
- //create reoriented primer and barcode pairs
- map<int, oligosPair> rpairedPrimers, rpairedBarcodes;
- for (map<int, oligosPair>::iterator it = pairedPrimers.begin(); it != pairedPrimers.end(); it++) {
- oligosPair tempPair(reverseOligo((it->second).reverse), (reverseOligo((it->second).forward))); //reversePrimer, rc ForwardPrimer
- rpairedPrimers[it->first] = tempPair;
- //cout << reverseOligo((it->second).reverse) << '\t' << (reverseOligo((it->second).forward)) << '\t' << primerNameVector[it->first] << endl;
- }
- for (map<int, oligosPair>::iterator it = pairedBarcodes.begin(); it != pairedBarcodes.end(); it++) {
- oligosPair tempPair(reverseOligo((it->second).reverse), (reverseOligo((it->second).forward))); //reverseBarcode, rc ForwardBarcode
- rpairedBarcodes[it->first] = tempPair;
- //cout << reverseOligo((it->second).reverse) << '\t' << (reverseOligo((it->second).forward)) << '\t' << barcodeNameVector[it->first] << endl;
- }
- rtrimOligos = new TrimOligos(pdiffs, bdiffs, 0, 0, rpairedPrimers, rpairedBarcodes); numBarcodes = rpairedBarcodes.size();
+ rtrimOligos = new TrimOligos(pdiffs, bdiffs, 0, 0, oligos.getReorientedPairedPrimers(), oligos.getReorientedPairedBarcodes()); numBarcodes = oligos.getReorientedPairedBarcodes().size();
}
while (moreSeqs) {
int currentSeqsDiffs = 0;
Sequence currSeq(inFASTA); m->gobble(inFASTA);
- //cout << currSeq.getName() << '\t' << currSeq.getUnaligned().length() << endl;
+ //cout << currSeq.getName() << '\t' << currSeq.getUnaligned() << endl;
Sequence savedSeq(currSeq.getName(), currSeq.getAligned());
QualityScores currQual; QualityScores savedQual;
if(numBarcodes != 0){
success = trimOligos->stripBarcode(currSeq, currQual, barcodeIndex);
- if(success > bdiffs) { trashCode += 'b'; }
+ if(success > bdiffs) {
+ trashCode += 'b';
+ }
else{ currentSeqsDiffs += success; }
}
-
+ //cout << currSeq.getName() << '\t' << currSeq.getUnaligned() << endl;
if(numSpacers != 0){
success = trimOligos->stripSpacer(currSeq, currQual);
if(success > sdiffs) { trashCode += 's'; }
if(numFPrimers != 0){
success = trimOligos->stripForward(currSeq, currQual, primerIndex, keepforward);
if(success > pdiffs) {
- //if (pairedOligos) { trashCode += trimOligos->getTrashCode(); }
- //else { trashCode += 'f'; }
- trashCode += 'f';
+ trashCode += 'f';
}
else{ currentSeqsDiffs += success; }
}
int thisBarcodeIndex = 0;
int thisPrimerIndex = 0;
-
+ //cout << currSeq.getName() << '\t' << savedSeq.getUnaligned() << endl;
if(numBarcodes != 0){
thisSuccess = rtrimOligos->stripBarcode(savedSeq, savedQual, thisBarcodeIndex);
- if(thisSuccess > bdiffs) { thisTrashCode += 'b'; }
+ if(thisSuccess > bdiffs) { thisTrashCode += "b"; }
else{ thisCurrentSeqsDiffs += thisSuccess; }
}
-
+ //cout << currSeq.getName() << '\t' << savedSeq.getUnaligned() << endl;
if(numFPrimers != 0){
thisSuccess = rtrimOligos->stripForward(savedSeq, savedQual, thisPrimerIndex, keepforward);
- if(thisSuccess > pdiffs) {
- //if (pairedOligos) { thisTrashCode += rtrimOligos->getTrashCode(); }
- //else { thisTrashCode += 'f'; }
- thisTrashCode += 'f';
- }
+ if(thisSuccess > pdiffs) { thisTrashCode += "f"; }
else{ thisCurrentSeqsDiffs += thisSuccess; }
}
savedQual.flipQScores();
currQual.setScores(savedQual.getScores());
}
- }
+ }else { trashCode += "(" + thisTrashCode + ")"; }
}
if(keepFirst != 0){
int origLength = currSeq.getNumBases();
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); }
+ else if(qAverage != 0) { success = currQual.cullQualAverage(currSeq, qAverage, logtransform); }
+ else if(qRollAverage != 0) { success = currQual.stripQualRollingAverage(currSeq, qRollAverage, logtransform); }
+ else if(qWindowAverage != 0){ success = currQual.stripQualWindowAverage(currSeq, qWindowStep, qWindowSize, qWindowAverage, logtransform); }
else { success = 1; }
//you don't want to trim, if it fails above then scrap it
if(trashCode.length() == 0){
string thisGroup = "";
- if (createGroup) {
- if(numBarcodes != 0){
- thisGroup = barcodeNameVector[barcodeIndex];
- if (numFPrimers != 0) {
- if (primerNameVector[primerIndex] != "") {
- if(thisGroup != "") {
- thisGroup += "." + primerNameVector[primerIndex];
- }else {
- thisGroup = primerNameVector[primerIndex];
- }
- }
- }
- }
- }
+ if (createGroup) { thisGroup = oligos.getGroupName(barcodeIndex, primerIndex); }
int pos = thisGroup.find("ignore");
if (pos == string::npos) {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//loop through and create all the processes you want
while (process != processors) {
- int pid = fork();
+ pid_t pid = fork();
if (pid > 0) {
processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
for(int i=0;i<tempFASTAFileNames.size();i++){
for(int j=0;j<tempFASTAFileNames[i].size();j++){
if (tempFASTAFileNames[i][j] != "") {
- tempFASTAFileNames[i][j] += toString(getpid()) + ".temp";
+ tempFASTAFileNames[i][j] += m->mothurGetpid(process) + ".temp";
m->openOutputFile(tempFASTAFileNames[i][j], temp); temp.close();
if(qFileName != ""){
- tempPrimerQualFileNames[i][j] += toString(getpid()) + ".temp";
+ tempPrimerQualFileNames[i][j] += m->mothurGetpid(process) + ".temp";
m->openOutputFile(tempPrimerQualFileNames[i][j], temp); temp.close();
}
if(nameFile != ""){
- tempNameFileNames[i][j] += toString(getpid()) + ".temp";
+ tempNameFileNames[i][j] += m->mothurGetpid(process) + ".temp";
m->openOutputFile(tempNameFileNames[i][j], temp); temp.close();
}
}
driverCreateTrim(filename,
qFileName,
- (trimFASTAFileName + toString(getpid()) + ".temp"),
- (scrapFASTAFileName + toString(getpid()) + ".temp"),
- (trimQualFileName + toString(getpid()) + ".temp"),
- (scrapQualFileName + toString(getpid()) + ".temp"),
- (trimNameFileName + toString(getpid()) + ".temp"),
- (scrapNameFileName + toString(getpid()) + ".temp"),
- (trimCountFileName + toString(getpid()) + ".temp"),
- (scrapCountFileName + toString(getpid()) + ".temp"),
- (groupFile + toString(getpid()) + ".temp"),
+ (trimFASTAFileName + m->mothurGetpid(process) + ".temp"),
+ (scrapFASTAFileName + m->mothurGetpid(process) + ".temp"),
+ (trimQualFileName + m->mothurGetpid(process) + ".temp"),
+ (scrapQualFileName + m->mothurGetpid(process) + ".temp"),
+ (trimNameFileName + m->mothurGetpid(process) + ".temp"),
+ (scrapNameFileName + m->mothurGetpid(process) + ".temp"),
+ (trimCountFileName + m->mothurGetpid(process) + ".temp"),
+ (scrapCountFileName + m->mothurGetpid(process) + ".temp"),
+ (groupFile + m->mothurGetpid(process) + ".temp"),
tempFASTAFileNames,
tempPrimerQualFileNames,
tempNameFileNames,
lines[process],
qLines[process]);
- if (m->debug) { m->mothurOut("[DEBUG]: " + toString(lines[process].start) + '\t' + toString(qLines[process].start) + '\t' + toString(getpid()) + '\n'); }
+ if (m->debug) { m->mothurOut("[DEBUG]: " + toString(lines[process].start) + '\t' + toString(qLines[process].start) + '\t' + m->mothurGetpid(process) + '\n'); }
//pass groupCounts to parent
if(createGroup){
ofstream out;
- string tempFile = filename + toString(getpid()) + ".num.temp";
+ string tempFile = filename + m->mothurGetpid(process) + ".num.temp";
m->openOutputFile(tempFile, out);
out << groupCounts.size() << endl;
tempPrimerQualFileNames,
tempNameFileNames,
lines[h].start, lines[h].end, qLines[h].start, qLines[h].end, m,
- pdiffs, bdiffs, ldiffs, sdiffs, tdiffs, primers, barcodes, revPrimer, linker, spacer, pairedBarcodes, pairedPrimers, pairedOligos,
- primerNameVector, barcodeNameVector, createGroup, allFiles, keepforward, keepFirst, removeLast,
- qWindowStep, qWindowSize, qWindowAverage, qtrim, qThreshold, qAverage, qRollAverage,
+ pdiffs, bdiffs, ldiffs, sdiffs, tdiffs, oligoFile,
+ createGroup, allFiles, keepforward, keepFirst, removeLast,
+ qWindowStep, qWindowSize, qWindowAverage, qtrim, qThreshold, qAverage, qRollAverage, logtransform,
minLength, maxAmbig, maxHomoP, maxLength, flip, reorient, nameMap, nameCount);
pDataArray.push_back(tempTrim);
string sname = ""; nameStream >> sname;
sname = sname.substr(1);
- for (int i = 0; i < sname.length(); i++) {
- if (sname[i] == ':') { sname[i] = '_'; m->changedSeqNames = true; }
- }
+ m->checkName(sname);
map<string, int>::iterator it = firstSeqNames.find(sname);
//***************************************************************************************************************
-bool TrimSeqsCommand::getOligos(vector<vector<string> >& fastaFileNames, vector<vector<string> >& qualFileNames, vector<vector<string> >& nameFileNames){
+bool TrimSeqsCommand::getOligos(vector<vector<string> >& fastaFileNames, vector<vector<string> >& qualFileNames, vector<vector<string> >& nameFileNames, map<string, string>& fastaFile2Group){
try {
- ifstream inOligos;
+
+ bool allBlank = false;
+ oligos.read(oligoFile);
+
+ if (m->control_pressed) { return false; } //error in reading oligos
+
+ if (oligos.hasPairedBarcodes()) {
+ pairedOligos = true;
+ numFPrimers = oligos.getPairedPrimers().size();
+ numBarcodes = oligos.getPairedBarcodes().size();
+ }else {
+ pairedOligos = false;
+ numFPrimers = oligos.getPrimers().size();
+ numBarcodes = oligos.getBarcodes().size();
+ }
+
+ numLinkers = oligos.getLinkers().size();
+ numSpacers = oligos.getSpacers().size();
+ numRPrimers = oligos.getReversePrimers().size();
+
+ vector<string> groupNames = oligos.getGroupNames();
+ if (groupNames.size() == 0) { allFiles = 0; allBlank = true; }
+
+
+ fastaFileNames.resize(oligos.getBarcodeNames().size());
+ for(int i=0;i<fastaFileNames.size();i++){
+ for(int j=0;j<oligos.getPrimerNames().size();j++){ fastaFileNames[i].push_back(""); }
+ }
+
+ if(qFileName != "") { qualFileNames = fastaFileNames; }
+ if(nameFile != "") { nameFileNames = fastaFileNames; }
+
+
+ if (allFiles) {
+ set<string> uniqueNames; //used to cleanup outputFileNames
+ if (pairedOligos) {
+ map<int, oligosPair> barcodes = oligos.getPairedBarcodes();
+ map<int, oligosPair> primers = oligos.getPairedPrimers();
+ for(map<int, oligosPair>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
+ for(map<int, oligosPair>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
+
+ string primerName = oligos.getPrimerName(itPrimer->first);
+ string barcodeName = oligos.getBarcodeName(itBar->first);
+
+ if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
+ else if ((primerName == "") && (barcodeName == "")) { } //do nothing
+ else {
+ string comboGroupName = "";
+ string fastaFileName = "";
+ string qualFileName = "";
+ string nameFileName = "";
+ string countFileName = "";
+
+ if(primerName == ""){
+ comboGroupName = barcodeName;
+ }else{
+ if(barcodeName == ""){
+ comboGroupName = primerName;
+ }
+ else{
+ comboGroupName = barcodeName + "." + primerName;
+ }
+ }
+
+
+ ofstream temp;
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFile));
+ variables["[tag]"] = comboGroupName;
+ fastaFileName = getOutputFileName("fasta", variables);
+ if (uniqueNames.count(fastaFileName) == 0) {
+ outputNames.push_back(fastaFileName);
+ outputTypes["fasta"].push_back(fastaFileName);
+ uniqueNames.insert(fastaFileName);
+ fastaFile2Group[fastaFileName] = comboGroupName;
+ }
+
+ fastaFileNames[itBar->first][itPrimer->first] = fastaFileName;
+ m->openOutputFile(fastaFileName, temp); temp.close();
+
+ if(qFileName != ""){
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(qFileName));
+ qualFileName = getOutputFileName("qfile", variables);
+ if (uniqueNames.count(qualFileName) == 0) {
+ outputNames.push_back(qualFileName);
+ outputTypes["qfile"].push_back(qualFileName);
+ }
+
+ qualFileNames[itBar->first][itPrimer->first] = qualFileName;
+ m->openOutputFile(qualFileName, temp); temp.close();
+ }
+
+ if(nameFile != ""){
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(nameFile));
+ nameFileName = getOutputFileName("name", variables);
+ if (uniqueNames.count(nameFileName) == 0) {
+ outputNames.push_back(nameFileName);
+ outputTypes["name"].push_back(nameFileName);
+ }
+
+ nameFileNames[itBar->first][itPrimer->first] = nameFileName;
+ m->openOutputFile(nameFileName, temp); temp.close();
+ }
+ }
+ }
+ }
+ }else {
+ map<string, int> barcodes = oligos.getBarcodes() ;
+ map<string, int> primers = oligos.getPrimers();
+ for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
+ for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
+
+ string primerName = oligos.getPrimerName(itPrimer->second);
+ string barcodeName = oligos.getBarcodeName(itBar->second);
+
+ if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
+ else if ((primerName == "") && (barcodeName == "")) { } //do nothing
+ else {
+ string comboGroupName = "";
+ string fastaFileName = "";
+ string qualFileName = "";
+ string nameFileName = "";
+ string countFileName = "";
+
+ if(primerName == ""){
+ comboGroupName = barcodeName;
+ }else{
+ if(barcodeName == ""){
+ comboGroupName = primerName;
+ }
+ else{
+ comboGroupName = barcodeName + "." + primerName;
+ }
+ }
+
+
+
+ ofstream temp;
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFile));
+ variables["[tag]"] = comboGroupName;
+ fastaFileName = getOutputFileName("fasta", variables);
+ if (uniqueNames.count(fastaFileName) == 0) {
+ outputNames.push_back(fastaFileName);
+ outputTypes["fasta"].push_back(fastaFileName);
+ uniqueNames.insert(fastaFileName);
+ fastaFile2Group[fastaFileName] = comboGroupName;
+ }
+
+ fastaFileNames[itBar->second][itPrimer->second] = fastaFileName;
+ m->openOutputFile(fastaFileName, temp); temp.close();
+
+ if(qFileName != ""){
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(qFileName));
+ qualFileName = getOutputFileName("qfile", variables);
+ if (uniqueNames.count(qualFileName) == 0) {
+ outputNames.push_back(qualFileName);
+ outputTypes["qfile"].push_back(qualFileName);
+ }
+
+ qualFileNames[itBar->second][itPrimer->second] = qualFileName;
+ m->openOutputFile(qualFileName, temp); temp.close();
+ }
+
+ if(nameFile != ""){
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(nameFile));
+ nameFileName = getOutputFileName("name", variables);
+ if (uniqueNames.count(nameFileName) == 0) {
+ outputNames.push_back(nameFileName);
+ outputTypes["name"].push_back(nameFileName);
+ }
+
+ nameFileNames[itBar->second][itPrimer->second] = nameFileName;
+ m->openOutputFile(nameFileName, temp); temp.close();
+ }
+ }
+ }
+ }
+ }
+
+ }
+
+
+
+ if (allBlank) {
+ m->mothurOut("[WARNING]: your oligos file does not contain any group names. mothur will not create a groupfile."); m->mothurOutEndLine();
+ allFiles = false;
+ return false;
+ }
+
+
+
+ /*ifstream inOligos;
m->openInputFile(oligoFile, inOligos);
ofstream test;
while(!inOligos.eof()){
- inOligos >> type;
+ inOligos >> type;
if (m->debug) { m->mothurOut("[DEBUG]: reading type - " + type + ".\n"); }
// get rest of line in case there is a primer name
while (!inOligos.eof()) {
char c = inOligos.get();
- if (c == 10 || c == 13){ break; }
+ if (c == 10 || c == 13 || c == -1){ break; }
else if (c == 32 || c == 9){;} //space or tab
else { group += c; }
}
}
if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0; }
-
+
//add in potential combos
if(barcodeNameVector.size() == 0){
barcodes[""] = 0;
}
}
}
- numFPrimers = primers.size();
- if (pairedOligos) { numFPrimers = pairedPrimers.size(); }
- numRPrimers = revPrimer.size();
- numLinkers = linker.size();
- numSpacers = spacer.size();
-
- bool allBlank = true;
- for (int i = 0; i < barcodeNameVector.size(); i++) {
- if (barcodeNameVector[i] != "") {
- allBlank = false;
- break;
- }
- }
- for (int i = 0; i < primerNameVector.size(); i++) {
- if (primerNameVector[i] != "") {
- allBlank = false;
- break;
- }
- }
-
- if (allBlank) {
- m->mothurOut("[WARNING]: your oligos file does not contain any group names. mothur will not create a groupfile."); m->mothurOutEndLine();
- allFiles = false;
- return false;
- }
-
- return true;
+ */
+ return true;
}
catch(exception& e) {
if(qscores.getName() != ""){
qscores.trimQScores(-1, keepFirst);
}
+
+// sequence.printSequence(cout);cout << endl;
+
sequence.trim(keepFirst);
+
+// sequence.printSequence(cout);cout << endl << endl;;
+
return success;
}
catch(exception& e) {
}
}
-//********************************************************************/
-string TrimSeqsCommand::reverseOligo(string oligo){
- try {
- string reverse = "";
-
- for(int i=oligo.length()-1;i>=0;i--){
-
- if(oligo[i] == 'A') { reverse += 'T'; }
- else if(oligo[i] == 'T'){ reverse += 'A'; }
- else if(oligo[i] == 'U'){ reverse += 'A'; }
-
- else if(oligo[i] == 'G'){ reverse += 'C'; }
- else if(oligo[i] == 'C'){ reverse += 'G'; }
-
- else if(oligo[i] == 'R'){ reverse += 'Y'; }
- else if(oligo[i] == 'Y'){ reverse += 'R'; }
-
- else if(oligo[i] == 'M'){ reverse += 'K'; }
- else if(oligo[i] == 'K'){ reverse += 'M'; }
-
- else if(oligo[i] == 'W'){ reverse += 'W'; }
- else if(oligo[i] == 'S'){ reverse += 'S'; }
-
- else if(oligo[i] == 'B'){ reverse += 'V'; }
- else if(oligo[i] == 'V'){ reverse += 'B'; }
-
- else if(oligo[i] == 'D'){ reverse += 'H'; }
- else if(oligo[i] == 'H'){ reverse += 'D'; }
-
- else { reverse += 'N'; }
- }
-
-
- return reverse;
- }
- catch(exception& e) {
- m->errorOut(e, "TrimSeqsCommand", "reverseOligo");
- exit(1);
- }
-}
//***************************************************************************************************************
projects / mothur.git / blobdiff