+int ParseFastaQCommand::findGroup(fastqRead2 thisRead, int& barcode, int& primer, TrimOligos*& trimOligos, int numBarcodes, int numPrimers) {
+ try {
+ int success = 1;
+ string trashCode = "";
+ int currentSeqsDiffs = 0;
+
+ Sequence currSeq(thisRead.seq.getName(), thisRead.seq.getAligned());
+ QualityScores currQual; currQual.setScores(convertQual(thisRead.quality));
+
+ if(linker.size() != 0){
+ success = trimOligos->stripLinker(currSeq, currQual);
+ if(success > ldiffs) { trashCode += 'k'; }
+ else{ currentSeqsDiffs += success; }
+
+ }
+
+ if(numBarcodes != 0){
+ success = trimOligos->stripBarcode(currSeq, currQual, barcode);
+ if(success > bdiffs) { trashCode += 'b'; }
+ else{ currentSeqsDiffs += success; }
+ }
+
+ if(spacer.size() != 0){
+ success = trimOligos->stripSpacer(currSeq, currQual);
+ if(success > sdiffs) { trashCode += 's'; }
+ else{ currentSeqsDiffs += success; }
+
+ }
+
+ if(numPrimers != 0){
+ success = trimOligos->stripForward(currSeq, currQual, primer, true);
+ if(success > pdiffs) { trashCode += 'f'; }
+ else{ currentSeqsDiffs += success; }
+ }
+
+ if (currentSeqsDiffs > tdiffs) { trashCode += 't'; }
+
+ if(revPrimer.size() != 0){
+ success = trimOligos->stripReverse(currSeq, currQual);
+ if(!success) { trashCode += 'r'; }
+ }
+
+
+ return trashCode.length();
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ParseFastaQCommand", "findGroup");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int ParseFastaQCommand::findGroup(fastqRead2 thisRead, int& barcode, int& primer, string groupMode) {
+ try {
+ string trashCode = "";
+ primer = 0;
+
+ string group = groupMap->getGroup(thisRead.seq.getName());
+ if (group == "not found") { trashCode += "g"; } //scrap for group
+ else { //find file group
+ map<string, int>::iterator it = barcodes.find(group);
+ if (it != barcodes.end()) {
+ barcode = it->second;
+ }else { trashCode += "g"; }
+ }
+
+ return trashCode.length();
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ParseFastaQCommand", "findGroup");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+
+bool ParseFastaQCommand::readOligos(string oligoFile){
+ try {
+ ifstream inOligos;
+ m->openInputFile(oligoFile, inOligos);
+
+ string type, oligo, roligo, group;
+ bool hasPrimer = false; bool hasPairedBarcodes = false; pairedOligos = false;
+
+ int indexPrimer = 0;
+ int indexBarcode = 0;
+ int indexPairedPrimer = 0;
+ int indexPairedBarcode = 0;
+ set<string> uniquePrimers;
+ set<string> uniqueBarcodes;
+
+ while(!inOligos.eof()){
+
+ inOligos >> type;
+
+ if (m->debug) { m->mothurOut("[DEBUG]: reading type - " + type + ".\n"); }
+
+ if(type[0] == '#'){
+ while (!inOligos.eof()) { char c = inOligos.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+ m->gobble(inOligos);
+ }
+ else{
+ m->gobble(inOligos);
+ //make type case insensitive
+ for(int i=0;i<type.length();i++){ type[i] = toupper(type[i]); }
+
+ inOligos >> oligo;
+
+ if (m->debug) { m->mothurOut("[DEBUG]: reading - " + oligo + ".\n"); }
+
+ for(int i=0;i<oligo.length();i++){
+ oligo[i] = toupper(oligo[i]);
+ if(oligo[i] == 'U') { oligo[i] = 'T'; }
+ }
+
+ if(type == "FORWARD"){
+ group = "";
+
+ // get rest of line in case there is a primer name
+ while (!inOligos.eof()) {
+ char c = inOligos.get();
+ if (c == 10 || c == 13 || c == -1){ break; }
+ else if (c == 32 || c == 9){;} //space or tab
+ else { group += c; }
+ }
+
+ //check for repeat barcodes
+ map<string, int>::iterator itPrime = primers.find(oligo);
+ if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine(); }
+
+ if (m->debug) { if (group != "") { m->mothurOut("[DEBUG]: reading group " + group + ".\n"); }else{ m->mothurOut("[DEBUG]: no group for primer " + oligo + ".\n"); } }
+
+ primers[oligo]=indexPrimer; indexPrimer++;
+ primerNameVector.push_back(group);
+ }
+ else if (type == "PRIMER"){
+ m->gobble(inOligos);
+
+ inOligos >> roligo;
+
+ for(int i=0;i<roligo.length();i++){
+ roligo[i] = toupper(roligo[i]);
+ if(roligo[i] == 'U') { roligo[i] = 'T'; }
+ }
+ roligo = reverseOligo(roligo);
+
+ group = "";
+
+ // get rest of line in case there is a primer name
+ while (!inOligos.eof()) {
+ char c = inOligos.get();
+ if (c == 10 || c == 13 || c == -1){ break; }
+ else if (c == 32 || c == 9){;} //space or tab
+ else { group += c; }
+ }
+
+ oligosPair newPrimer(oligo, roligo);
+
+ if (m->debug) { m->mothurOut("[DEBUG]: primer pair " + newPrimer.forward + " " + newPrimer.reverse + ", and group = " + group + ".\n"); }
+
+ //check for repeat barcodes
+ string tempPair = oligo+roligo;
+ if (uniquePrimers.count(tempPair) != 0) { m->mothurOut("primer pair " + newPrimer.forward + " " + newPrimer.reverse + " is in your oligos file already."); m->mothurOutEndLine(); }
+ else { uniquePrimers.insert(tempPair); }
+
+ if (m->debug) { if (group != "") { m->mothurOut("[DEBUG]: reading group " + group + ".\n"); }else{ m->mothurOut("[DEBUG]: no group for primer pair " + newPrimer.forward + " " + newPrimer.reverse + ".\n"); } }
+
+ pairedPrimers[indexPairedPrimer]=newPrimer; indexPairedPrimer++;
+ primerNameVector.push_back(group);
+ hasPrimer = true;
+ }
+ else if(type == "REVERSE"){
+ //Sequence oligoRC("reverse", oligo);
+ //oligoRC.reverseComplement();
+ string oligoRC = reverseOligo(oligo);
+ revPrimer.push_back(oligoRC);
+ }
+ else if(type == "BARCODE"){
+ inOligos >> group;
+
+ //barcode lines can look like BARCODE atgcatgc groupName - for 454 seqs
+ //or BARCODE atgcatgc atgcatgc groupName - for illumina data that has forward and reverse info
+
+ string temp = "";
+ while (!inOligos.eof()) {
+ char c = inOligos.get();
+ if (c == 10 || c == 13 || c == -1){ break; }
+ else if (c == 32 || c == 9){;} //space or tab
+ else { temp += c; }
+ }
+
+ //then this is illumina data with 4 columns
+ if (temp != "") {
+ hasPairedBarcodes = true;
+ string reverseBarcode = group; //reverseOligo(group); //reverse barcode
+ group = temp;
+
+ for(int i=0;i<reverseBarcode.length();i++){
+ reverseBarcode[i] = toupper(reverseBarcode[i]);
+ if(reverseBarcode[i] == 'U') { reverseBarcode[i] = 'T'; }
+ }
+
+ reverseBarcode = reverseOligo(reverseBarcode);
+ oligosPair newPair(oligo, reverseBarcode);
+
+ if (m->debug) { m->mothurOut("[DEBUG]: barcode pair " + newPair.forward + " " + newPair.reverse + ", and group = " + group + ".\n"); }
+ //check for repeat barcodes
+ string tempPair = oligo+reverseBarcode;
+ if (uniqueBarcodes.count(tempPair) != 0) { m->mothurOut("barcode pair " + newPair.forward + " " + newPair.reverse + " is in your oligos file already, disregarding."); m->mothurOutEndLine(); }
+ else { uniqueBarcodes.insert(tempPair); }
+
+ pairedBarcodes[indexPairedBarcode]=newPair; indexPairedBarcode++;
+ barcodeNameVector.push_back(group);
+ }else {
+ //check for repeat barcodes
+ map<string, int>::iterator itBar = barcodes.find(oligo);
+ if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine(); }
+
+ barcodes[oligo]=indexBarcode; indexBarcode++;
+ barcodeNameVector.push_back(group);
+ }
+ }else if(type == "LINKER"){
+ linker.push_back(oligo);
+ }else if(type == "SPACER"){
+ spacer.push_back(oligo);
+ }
+ else{ m->mothurOut("[WARNING]: " + type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine(); }
+ }
+ m->gobble(inOligos);
+ }
+ inOligos.close();
+
+ if (hasPairedBarcodes || hasPrimer) {
+ pairedOligos = true;
+ if ((primers.size() != 0) || (barcodes.size() != 0) || (linker.size() != 0) || (spacer.size() != 0) || (revPrimer.size() != 0)) { m->control_pressed = true; m->mothurOut("[ERROR]: cannot mix paired primers and barcodes with non paired or linkers and spacers, quitting."); m->mothurOutEndLine(); return 0; }
+ }
+
+ //add in potential combos
+ if(barcodeNameVector.size() == 0){
+ barcodes[""] = 0;
+ barcodeNameVector.push_back("");
+ }
+
+ if(primerNameVector.size() == 0){
+ primers[""] = 0;
+ primerNameVector.push_back("");
+ }
+
+ fastqFileNames.resize(barcodeNameVector.size());
+ for(int i=0;i<fastqFileNames.size();i++){
+ fastqFileNames[i].assign(primerNameVector.size(), "");
+ }
+
+
+ set<string> uniqueNames; //used to cleanup outputFileNames
+ if (pairedOligos) {
+ for(map<int, oligosPair>::iterator itBar = pairedBarcodes.begin();itBar != pairedBarcodes.end();itBar++){
+ for(map<int, oligosPair>::iterator itPrimer = pairedPrimers.begin();itPrimer != pairedPrimers.end(); itPrimer++){
+
+ string primerName = primerNameVector[itPrimer->first];
+ string barcodeName = barcodeNameVector[itBar->first];
+
+ if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
+ else {
+ string comboGroupName = "";
+ string fastqFileName = "";
+
+ if(primerName == ""){
+ comboGroupName = barcodeNameVector[itBar->first];
+ }
+ else{
+ if(barcodeName == ""){
+ comboGroupName = primerNameVector[itPrimer->first];
+ }
+ else{
+ comboGroupName = barcodeNameVector[itBar->first] + "." + primerNameVector[itPrimer->first];
+ }
+ }
+
+
+ ofstream temp;
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
+ variables["[group]"] = comboGroupName;
+ fastqFileName = getOutputFileName("fastq", variables);
+ if (uniqueNames.count(fastqFileName) == 0) {
+ outputNames.push_back(fastqFileName);
+ outputTypes["fastq"].push_back(fastqFileName);
+ uniqueNames.insert(fastqFileName);
+ }
+
+ fastqFileNames[itBar->first][itPrimer->first] = fastqFileName;
+ m->openOutputFile(fastqFileName, temp); temp.close();
+
+ }
+ }
+ }
+ }else {
+ 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 = primerNameVector[itPrimer->second];
+ string barcodeName = barcodeNameVector[itBar->second];
+
+ if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
+ else {
+ string comboGroupName = "";
+ string fastqFileName = "";
+
+ if(primerName == ""){
+ comboGroupName = barcodeNameVector[itBar->second];
+ }
+ else{
+ if(barcodeName == ""){
+ comboGroupName = primerNameVector[itPrimer->second];
+ }
+ else{
+ comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
+ }
+ }
+
+
+ ofstream temp;
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
+ variables["[group]"] = comboGroupName;
+ fastqFileName = getOutputFileName("fastq", variables);
+ if (uniqueNames.count(fastqFileName) == 0) {
+ outputNames.push_back(fastqFileName);
+ outputTypes["fastq"].push_back(fastqFileName);
+ uniqueNames.insert(fastqFileName);
+ }
+
+ fastqFileNames[itBar->second][itPrimer->second] = fastqFileName;
+ m->openOutputFile(fastqFileName, temp); temp.close();
+
+ }
+ }
+ }
+ }
+
+ ofstream temp;
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
+ variables["[group]"] = "scrap";
+ noMatchFile = getOutputFileName("fastq", variables);
+ m->openOutputFile(noMatchFile, temp); temp.close();
+
+ return true;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ParseFastaQCommand", "getOligos");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+bool ParseFastaQCommand::readGroup(string groupfile){
+ try {
+ fastqFileNames.clear();
+
+ groupMap = new GroupMap();
+ groupMap->readMap(groupfile);
+
+ //like barcodeNameVector - no primer names
+ vector<string> groups = groupMap->getNamesOfGroups();
+
+ fastqFileNames.resize(groups.size());
+ for (int i = 0; i < fastqFileNames.size(); i++) {
+ for (int j = 0; j < 1; j++) {
+
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
+ variables["[group]"] = groups[i];
+ string thisFilename = getOutputFileName("fastq",variables);
+ outputNames.push_back(thisFilename);
+ outputTypes["fastq"].push_back(thisFilename);
+
+ ofstream temp;
+ m->openOutputFileBinary(thisFilename, temp); temp.close();
+ fastqFileNames[i].push_back(thisFilename);
+ barcodes[groups[i]] = i;
+ }
+ }
+
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
+ variables["[group]"] = "scrap";
+ noMatchFile = getOutputFileName("fastq",variables);
+ m->mothurRemove(noMatchFile);
+
+ return true;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ParseFastaQCommand", "readGroup");
+ exit(1);
+ }
+}
+//********************************************************************/
+string ParseFastaQCommand::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, "ParseFastaQCommand", "reverseOligo");
+ exit(1);
+ }
+}
+
+
+//**********************************************************************************************************************
projects / mothur.git / commitdiff