#include "needlemanoverlap.hpp"
//**********************************************************************************************************************
+
vector<string> TrimSeqsCommand::getValidParameters(){
try {
- string Array[] = {"fasta", "flip", "oligos", "maxambig", "maxhomop", "minlength", "maxlength", "qfile",
- "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
+ string Array[] = {"fasta", "flip", "oligos", "maxambig", "maxhomop", "group","minlength", "maxlength", "qfile",
+ "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage",
+ "keepfirst", "removelast",
+ "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
return myArray;
}
exit(1);
}
}
+
//**********************************************************************************************************************
+
TrimSeqsCommand::TrimSeqsCommand(){
try {
abort = true;
exit(1);
}
}
+
//**********************************************************************************************************************
+
vector<string> TrimSeqsCommand::getRequiredParameters(){
try {
string Array[] = {"fasta"};
exit(1);
}
}
+
//**********************************************************************************************************************
+
vector<string> TrimSeqsCommand::getRequiredFiles(){
try {
vector<string> myArray;
exit(1);
}
}
+
//***************************************************************************************************************
TrimSeqsCommand::TrimSeqsCommand(string option) {
else {
//valid paramters for this command
- string AlignArray[] = {"fasta", "flip", "oligos", "maxambig", "maxhomop", "minlength", "maxlength", "qfile",
- "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage", "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
+ string AlignArray[] = { "fasta", "flip", "oligos", "maxambig", "maxhomop", "group","minlength", "maxlength", "qfile",
+ "qthreshold", "qwindowaverage", "qstepsize", "qwindowsize", "qaverage", "rollaverage",
+ "keepfirst", "removelast",
+ "allfiles", "qtrim","tdiffs", "pdiffs", "bdiffs", "processors", "outputdir","inputdir"};
vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["qfile"] = inputDir + it->second; }
}
+
+ it = parameters.find("group");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["group"] = inputDir + it->second; }
+ }
}
else if(temp == "not open"){ abort = true; }
else { oligoFile = temp; }
+ temp = validParameter.validFile(parameters, "group", true);
+ if (temp == "not found"){ groupfile = ""; }
+ else if(temp == "not open"){ abort = true; }
+ else { groupfile = temp; }
+
temp = validParameter.validFile(parameters, "maxambig", false); if (temp == "not found") { temp = "-1"; }
convert(temp, maxAmbig);
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"; }
+ temp = validParameter.validFile(parameters, "qtrim", false); if (temp == "not found") { temp = "t"; }
qtrim = m->isTrue(temp);
temp = validParameter.validFile(parameters, "rollaverage", false); if (temp == "not found") { temp = "0"; }
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"; }
+ temp = validParameter.validFile(parameters, "qwindowsize", false); if (temp == "not found") { temp = "50"; }
convert(temp, qWindowSize);
- temp = validParameter.validFile(parameters, "qstepsize", false); if (temp == "not found") { temp = "10"; }
+ temp = validParameter.validFile(parameters, "qstepsize", false); if (temp == "not found") { temp = "1"; }
convert(temp, qWindowStep);
temp = validParameter.validFile(parameters, "qaverage", false); if (temp == "not found") { temp = "0"; }
convert(temp, qAverage);
+
+ temp = validParameter.validFile(parameters, "keepfirst", false); if (temp == "not found") { temp = "0"; }
+ convert(temp, keepFirst);
+
+ temp = validParameter.validFile(parameters, "removelast", false); if (temp == "not found") { temp = "0"; }
+ convert(temp, removeLast);
temp = validParameter.validFile(parameters, "allfiles", false); if (temp == "not found") { temp = "F"; }
allFiles = m->isTrue(temp);
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found") { temp = "1"; }
convert(temp, processors);
- if(allFiles && oligoFile == ""){
- m->mothurOut("You selected allfiles, but didn't enter an oligos file. Ignoring the allfiles request."); m->mothurOutEndLine();
+ if ((oligoFile != "") && (groupfile != "")) {
+ m->mothurOut("You given both a oligos file and a groupfile, only one is allowed."); m->mothurOutEndLine(); abort = true;
+ }
+
+
+ if(allFiles && (oligoFile == "") && (groupfile == "")){
+ m->mothurOut("You selected allfiles, but didn't enter an oligos or group file. Ignoring the allfiles request."); m->mothurOutEndLine();
}
if((qAverage != 0 && qThreshold != 0) && qFileName == ""){
m->mothurOut("You didn't provide a quality file name, quality criteria will be ignored."); m->mothurOutEndLine();
exit(1);
}
}
+
//**********************************************************************************************************************
void TrimSeqsCommand::help(){
try {
- m->mothurOut("The trim.seqs command reads a fastaFile and creates .....\n");
- m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim and allfiles.\n");
+ m->mothurOut("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");
+ m->mothurOut("The .trim.fasta contains sequences that meet your requirements, and the .scrap.fasta contains those which don't.\n");
+ m->mothurOut("The trim.seqs command parameters are fasta, flip, oligos, group, maxambig, maxhomop, minlength, maxlength, qfile, qthreshold, qaverage, diffs, qtrim, keepfirst, removelast and allfiles.\n");
m->mothurOut("The fasta parameter is required.\n");
+ m->mothurOut("The group parameter allows you to enter a group file for your fasta file.\n");
m->mothurOut("The flip parameter will output the reverse compliment of your trimmed sequence. The default is false.\n");
- m->mothurOut("The oligos parameter .... The default is "".\n");
- m->mothurOut("The maxambig parameter .... The default is -1.\n");
- m->mothurOut("The maxhomop parameter .... The default is 0.\n");
- m->mothurOut("The minlength parameter .... The default is 0.\n");
- m->mothurOut("The maxlength parameter .... The default is 0.\n");
+ m->mothurOut("The oligos parameter allows you to provide an oligos file.\n");
+ m->mothurOut("The maxambig parameter allows you to set the maximum number of ambigious bases allowed. The default is -1.\n");
+ m->mothurOut("The maxhomop parameter allows you to set a maximum homopolymer length. \n");
+ m->mothurOut("The minlength parameter allows you to set and minimum sequence length. \n");
+ m->mothurOut("The maxlength parameter allows you to set and maximum sequence length. \n");
m->mothurOut("The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs.\n");
m->mothurOut("The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n");
m->mothurOut("The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n");
- m->mothurOut("The qfile parameter .....\n");
- m->mothurOut("The qthreshold parameter .... The default is 0.\n");
- m->mothurOut("The qaverage parameter .... The default is 0.\n");
- m->mothurOut("The allfiles parameter .... The default is F.\n");
- m->mothurOut("The qtrim parameter .... The default is F.\n");
+ m->mothurOut("The qfile parameter allows you to provide a quality file.\n");
+ m->mothurOut("The qthreshold parameter allows you to set a minimum quality score allowed. \n");
+ m->mothurOut("The qaverage parameter allows you to set a minimum average quality score allowed. \n");
+ m->mothurOut("The qwindowsize parameter allows you to set a number of bases in a window. Default=50.\n");
+ m->mothurOut("The qwindowaverage parameter allows you to set a minimum average quality score allowed over a window. \n");
+ m->mothurOut("The rollaverage parameter allows you to set a minimum rolling average quality score allowed over a window. \n");
+ m->mothurOut("The qstepsize parameter allows you to set a number of bases to move the window over. Default=1.\n");
+ m->mothurOut("The allfiles parameter will create separate group and fasta file for each grouping. The default is F.\n");
+ m->mothurOut("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");
+ m->mothurOut("The keepfirst parameter trims the sequence to the first keepfirst number of bases after the barcode or primers are removed, before the sequence is checked to see if it meets the other requirements. \n");
+ m->mothurOut("The removelast removes the last removelast number of bases after the barcode or primers are removed, before the sequence is checked to see if it meets the other requirements.\n");
m->mothurOut("The trim.seqs command should be in the following format: \n");
m->mothurOut("trim.seqs(fasta=yourFastaFile, flip=yourFlip, oligos=yourOligos, maxambig=yourMaxambig, \n");
m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength) \n");
numFPrimers = 0; //this needs to be initialized
numRPrimers = 0;
+ vector<string> fastaFileNames;
+ vector<string> qualFileNames;
string trimSeqFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.fasta";
outputNames.push_back(trimSeqFile); outputTypes["fasta"].push_back(trimSeqFile);
string trimQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "trim.qual";
string scrapQualFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "scrap.qual";
if (qFileName != "") { outputNames.push_back(trimQualFile); outputNames.push_back(scrapQualFile); outputTypes["qual"].push_back(trimQualFile); outputTypes["qual"].push_back(scrapQualFile); }
- string groupFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "groups";
+ string groupFile = "";
+ if (groupfile == "") { groupFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "groups"; }
+ else{
+ groupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "trim.groups";
+ outputNames.push_back(groupFile); outputTypes["group"].push_back(groupFile);
+ groupMap = new GroupMap(groupfile);
+ groupMap->readMap();
+
+ if(allFiles){
+ for (int i = 0; i < groupMap->namesOfGroups.size(); i++) {
+ groupToIndex[groupMap->namesOfGroups[i]] = i;
+ groupVector.push_back(groupMap->namesOfGroups[i]);
+ fastaFileNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + groupMap->namesOfGroups[i] + ".fasta"));
+
+ //we append later, so we want to clear file
+ ofstream outRemove;
+ m->openOutputFile(fastaFileNames[i], outRemove);
+ outRemove.close();
+ if(qFileName != ""){
+ qualFileNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + groupMap->namesOfGroups[i] + ".qual"));
+ ofstream outRemove2;
+ m->openOutputFile(qualFileNames[i], outRemove2);
+ outRemove2.close();
+ }
+ }
+ }
+ comboStarts = fastaFileNames.size()-1;
+ }
- vector<string> fastaFileNames;
- vector<string> qualFileNames;
if(oligoFile != ""){
outputNames.push_back(groupFile); outputTypes["group"].push_back(groupFile);
getOligos(fastaFileNames, qualFileNames);
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
-
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{
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++){
- m->appendFiles((trimSeqFile + toString(processIDS[i]) + ".temp"), trimSeqFile);
- remove((trimSeqFile + toString(processIDS[i]) + ".temp").c_str());
- m->appendFiles((scrapSeqFile + toString(processIDS[i]) + ".temp"), scrapSeqFile);
- remove((scrapSeqFile + toString(processIDS[i]) + ".temp").c_str());
-
- 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++) {
- 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
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
-
-
+
+ if (m->control_pressed) { return 0; }
+
+ set<string> blanks;
for(int i=0;i<fastaFileNames.size();i++){
- if (m->isBlank(fastaFileNames[i])) { remove(fastaFileNames[i].c_str()); }
- else if (filesToRemove.count(fastaFileNames[i]) > 0) { remove(fastaFileNames[i].c_str()); }
+ if (m->isBlank(fastaFileNames[i])) { blanks.insert(fastaFileNames[i]); }
+ else if (filesToRemove.count(fastaFileNames[i]) > 0) { remove(fastaFileNames[i].c_str()); }
else {
ifstream inFASTA;
string seqName;
m->openInputFile(fastaFileNames[i], inFASTA);
ofstream outGroups;
string outGroupFilename = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + "groups";
+
+ //if the fastafile is on the blanks list then the groups file should be as well
+ if (blanks.count(fastaFileNames[i]) != 0) { blanks.insert(outGroupFilename); }
+
m->openOutputFile(outGroupFilename, outGroups);
outputNames.push_back(outGroupFilename); outputTypes["group"].push_back(outGroupFilename);
if(itCombo->second == i){ thisGroup = itCombo->first; combos.erase(itCombo); break; }
}
}else{ thisGroup = groupVector[i]; }
-
+
while(!inFASTA.eof()){
if(inFASTA.get() == '>'){
inFASTA >> seqName;
}
}
+ for (set<string>::iterator itBlanks = blanks.begin(); itBlanks != blanks.end(); itBlanks++) { remove((*(itBlanks)).c_str()); }
+
+ blanks.clear();
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->isBlank(qualFileNames[i])) { blanks.insert(qualFileNames[i]); }
+ else if (filesToRemove.count(qualFileNames[i]) > 0) { remove(qualFileNames[i].c_str()); }
}
}
+ for (set<string>::iterator itBlanks = blanks.begin(); itBlanks != blanks.end(); itBlanks++) { remove((*(itBlanks)).c_str()); }
if (m->control_pressed) {
for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
try {
ofstream outFASTA;
- int able = m->openOutputFile(trimFile, outFASTA);
+ m->openOutputFile(trimFile, outFASTA);
ofstream scrapFASTA;
m->openOutputFile(scrapFile, scrapFASTA);
}
ofstream outGroups;
- //vector<ofstream*> fastaFileNames;
- //vector<ofstream*> qualFileNames;
if (oligoFile != "") {
m->openOutputFile(groupFile, outGroups);
- for (int i = 0; i < fastaNames.size(); i++) {
-
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- fastaNames[i] = (fastaNames[i] + toString(getpid()) + ".temp");
- //fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
- //clear old file if it exists
- ofstream temp;
- m->openOutputFile(fastaNames[i], temp);
- temp.close();
- if(qFileName != ""){
- qualNames[i] = (qualNames[i] + toString(getpid()) + ".temp");
- //qualFileNames.push_back(new ofstream((qualNames[i] + toString(getpid()) + ".temp").c_str(), ios::ate));
- //clear old file if it exists
- ofstream temp2;
- m->openOutputFile(qualNames[i], temp2);
- temp2.close();
- }
- #else
- //fastaFileNames.push_back(new ofstream((fastaNames[i] + toString(i) + ".temp").c_str(), ios::ate));
- fastaNames[i] = (fastaNames[i] + toString(i) + ".temp");
- ofstream temp;
- m->openOutputFile(fastaNames[i], temp);
- temp.close();
- if(qFileName != ""){
- //qualFileNames.push_back(new ofstream((qualNames[i] + toString(i) + ".temp").c_str(), ios::ate));
- qualNames[i] = (qualNames[i] + toString(i) + ".temp");
- ofstream temp2;
- m->openOutputFile(qualNames[i], temp2);
- temp2.close();
- }
- #endif
- }
}
ifstream inFASTA;
ifstream qFile;
if(qFileName != "") { m->openInputFile(qFileName, qFile); qFile.seekg(qline->start); }
+
+ for (int i = 0; i < fastaNames.size(); i++) { //clears old file
+ ofstream temp;
+ m->openOutputFile(fastaNames[i], temp);
+ temp.close();
+ }
+ for (int i = 0; i < qualNames.size(); i++) { //clears old file
+ ofstream temp;
+ m->openOutputFile(qualNames[i], temp);
+ temp.close();
+ }
+
+
bool done = false;
int count = 0;
if (m->control_pressed) {
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()); }
string trashCode = "";
int currentSeqsDiffs = 0;
- if(qFileName != ""){
- 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(barcodes.size() != 0){
success = stripBarcode(currSeq, currQual, groupBar);
if(success > bdiffs) { trashCode += 'b'; }
else{ currentSeqsDiffs += success; }
}
-
+
if(numFPrimers != 0){
success = stripForward(currSeq, currQual, groupPrime);
if(success > pdiffs) { trashCode += 'f'; }
}
if (currentSeqsDiffs > tdiffs) { trashCode += 't'; }
-
+
if(numRPrimers != 0){
success = stripReverse(currSeq, currQual);
if(!success) { trashCode += 'r'; }
}
+
+ if(keepFirst != 0){
+ success = keepFirstTrim(currSeq, currQual);
+ }
+
+ if(removeLast != 0){
+ success = removeLastTrim(currSeq, currQual);
+ if(!success) { trashCode += 'l'; }
+ }
+
+
+ if(qFileName != ""){
+ 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 { success = 1; }
+
+ //you don't want to trim, if it fails above then scrap it
+ if ((!qtrim) && (origLength != currSeq.getNumBases())) { success = 0; }
+
+ if(!success) { trashCode += 'q'; }
+ }
if(minLength > 0 || maxLength > 0){
success = cullLength(currSeq);
if(!success) { trashCode += 'n'; }
}
- if(flip){ currSeq.reverseComplement(); currQual.flipQScores(); } // should go last
+ if(flip){ // should go last
+ currSeq.reverseComplement();
+ currQual.flipQScores();
+ }
if(trashCode.length() == 0){
currSeq.setAligned(currSeq.getUnaligned());
}
}
}
+
+ if (groupfile != "") {
+ string thisGroup = groupMap->getGroup(currSeq.getName());
+
+ if (thisGroup != "not found") {
+ outGroups << currSeq.getName() << '\t' << thisGroup << endl;
+ if (allFiles) {
+ ofstream outTemp;
+ m->openOutputFileAppend(fastaNames[groupToIndex[thisGroup]], outTemp);
+ currSeq.printSequence(outTemp);
+ outTemp.close();
+ if(qFileName != ""){
+ ofstream outTemp2;
+ m->openOutputFileAppend(qualNames[groupToIndex[thisGroup]], outTemp2);
+ currQual.printQScores(outTemp2);
+ outTemp2.close();
+ }
+ }
+ }else{
+ m->mothurOut(currSeq.getName() + " is not in your groupfile, adding to group XXX."); m->mothurOutEndLine();
+ outGroups << currSeq.getName() << '\t' << "XXX" << endl;
+ if (allFiles) {
+ m->mothurOut("[ERROR]: " + currSeq.getName() + " will not be added to any .group.fasta or .group.qual file."); m->mothurOutEndLine();
+ }
+ }
+ }
}
else{
currSeq.setName(currSeq.getName() + '|' + trashCode);
if (oligoFile != "") { outGroups.close(); }
if(qFileName != "") { qFile.close(); scrapQual.close(); outQual.close(); }
- //for(int i=0;i<fastaFileNames.size();i++){
- // fastaFileNames[i]->close();
- // delete fastaFileNames[i];
- //}
-
- //if(qFileName != ""){
- //for(int i=0;i<qualFileNames.size();i++){
- //qualFileNames[i]->close();
- //delete qualFileNames[i];
- //}
- //}
-
return count;
}
catch(exception& e) {
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;
+ int process = 1;
int exitCommand = 1;
processIDS.clear();
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){
+ for (int i = 0; i < fastaNames.size(); i++) {
+ fastaNames[i] = (fastaNames[i] + toString(getpid()) + ".temp");
+ //clear old file if it exists
+ ofstream temp;
+ m->openOutputFile(fastaNames[i], temp);
+ temp.close();
+ if(qFileName != ""){
+ qualNames[i] = (qualNames[i] + toString(getpid()) + ".temp");
+ //clear old file if it exists
+ ofstream temp2;
+ m->openOutputFile(qualNames[i], temp2);
+ temp2.close();
+ }
+ }
+
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); }
+ }else {
+ m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
+ for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
+ exit(0);
+ }
+ }
+
+ //parent do my part
+ for (int i = 0; i < fastaNames.size(); i++) {
+ //clear old file if it exists
+ ofstream temp;
+ m->openOutputFile(fastaNames[i], temp);
+ temp.close();
+ if(qFileName != ""){
+ //clear old file if it exists
+ ofstream temp2;
+ m->openOutputFile(qualNames[i], temp2);
+ temp2.close();
+ }
}
+ driverCreateTrim(filename, qFileName, trimFile, scrapFile, trimQFile, scrapQFile, groupFile, fastaNames, qualNames, lines[0], qLines[0]);
+
+
//force parent to wait until all the processes are done
- for (int i=0;i<processors;i++) {
+ for (int i=0;i<processIDS.size();i++) {
int temp = processIDS[i];
wait(&temp);
}
+ //append files
+ for(int i=0;i<processIDS.size();i++){
+
+ m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
+
+ m->appendFiles((trimFile + toString(processIDS[i]) + ".temp"), trimFile);
+ remove((trimFile + toString(processIDS[i]) + ".temp").c_str());
+ m->appendFiles((scrapFile + toString(processIDS[i]) + ".temp"), scrapFile);
+ remove((scrapFile + toString(processIDS[i]) + ".temp").c_str());
+
+ m->mothurOut("Done with fasta files"); m->mothurOutEndLine();
+
+ if(qFileName != ""){
+ m->appendFiles((trimQFile + toString(processIDS[i]) + ".temp"), trimQFile);
+ remove((trimQFile + toString(processIDS[i]) + ".temp").c_str());
+ m->appendFiles((scrapQFile + toString(processIDS[i]) + ".temp"), scrapQFile);
+ remove((scrapQFile + toString(processIDS[i]) + ".temp").c_str());
+
+ m->mothurOut("Done with quality files"); m->mothurOutEndLine();
+ }
+
+ m->appendFiles((groupFile + toString(processIDS[i]) + ".temp"), groupFile);
+ remove((groupFile + toString(processIDS[i]) + ".temp").c_str());
+
+ m->mothurOut("Done with group file"); m->mothurOutEndLine();
+
+ for (int j = 0; j < fastaNames.size(); j++) {
+ m->appendFiles((fastaNames[j] + toString(processIDS[i]) + ".temp"), fastaNames[j]);
+ remove((fastaNames[j] + toString(processIDS[i]) + ".temp").c_str());
+ }
+
+ if(qFileName != ""){
+ for (int j = 0; j < qualNames.size(); j++) {
+ m->appendFiles((qualNames[j] + toString(processIDS[i]) + ".temp"), qualNames[j]);
+ remove((qualNames[j] + toString(processIDS[i]) + ".temp").c_str());
+ }
+ }
+
+ if (allFiles) { m->mothurOut("Done with allfiles"); m->mothurOutEndLine(); }
+ }
+
return exitCommand;
#endif
}
exit(1);
}
}
+
//***************************************************************************************************************
void TrimSeqsCommand::getOligos(vector<string>& outFASTAVec, vector<string>& outQualVec){
//int indexPrimer = 0;
while(!inOligos.eof()){
- inOligos >> type;
+ inOligos >> type; m->gobble(inOligos);
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
map<string, int>::iterator itPrime = primers.find(oligo);
if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine(); }
- primers[oligo]=index; index++;
- groupVector.push_back(group);
+ primers[oligo]=index; index++;
+ groupVector.push_back(group);
- if(allFiles){
- 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(allFiles){
+ outFASTAVec.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
if(qFileName != ""){
- filesToRemove.insert((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ 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"));
+ outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
+ if(qFileName != ""){
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ }
}
- }else {
- outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
- outputTypes["fasta"].push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + group + ".fasta"));
- if(qFileName != ""){
- outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
- outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
- }
}
- }
-
+
}
else if(type == "REVERSE"){
Sequence oligoRC("reverse", oligo);
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]=index; index++;
- groupVector.push_back(group);
+ barcodes[oligo]=index; index++;
+ groupVector.push_back(group);
+
+ if(allFiles){
+ outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + 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"));
+ outputTypes["qual"].push_back((outputDir + m->getRootName(m->getSimpleName(qFileName)) + group + ".qual"));
+ }
+ }
- if(allFiles){
- outputNames.push_back((outputDir + m->getRootName(m->getSimpleName(fastaFile)) + 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"));
- outputTypes["qual"].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(); }
}
m->gobble(inOligos);
oligo = oligo.substr(0,alnLength);
temp = temp.substr(0,alnLength);
- int newStart=0;
int numDiff = countDiffs(oligo, temp);
// cout << oligo << '\t' << temp << '\t' << numDiff << endl;
seq.setUnaligned(rawSequence.substr(oligo.length()));
if(qual.getName() != ""){
qual.trimQScores(oligo.length(), -1);
-
}
success = 0;
break;
oligo = oligo.substr(0,alnLength);
temp = temp.substr(0,alnLength);
- int newStart=0;
int numDiff = countDiffs(oligo, temp);
// cout << oligo << '\t' << temp << '\t' << numDiff << endl;
//***************************************************************************************************************
+bool TrimSeqsCommand::keepFirstTrim(Sequence& sequence, QualityScores& qscores){
+ try {
+ bool success = 1;
+ if(qscores.getName() != ""){
+ qscores.trimQScores(-1, keepFirst);
+ }
+ sequence.trim(keepFirst);
+ return success;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "keepFirstTrim", "countDiffs");
+ exit(1);
+ }
+
+}
+
+//***************************************************************************************************************
+
+bool TrimSeqsCommand::removeLastTrim(Sequence& sequence, QualityScores& qscores){
+ try {
+ bool success = 0;
+
+ int length = sequence.getNumBases() - removeLast;
+
+ if(length > 0){
+ if(qscores.getName() != ""){
+ qscores.trimQScores(-1, length);
+ }
+ sequence.trim(length);
+ success = 1;
+ }
+ else{
+ success = 0;
+ }
+
+ return success;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "removeLastTrim", "countDiffs");
+ exit(1);
+ }
+
+}
+
+//***************************************************************************************************************
+
bool TrimSeqsCommand::cullLength(Sequence& seq){
try {
}
}
+
//***************************************************************************************************************
int TrimSeqsCommand::countDiffs(string oligo, string seq){
}
}
-//***************************************************************************************************************
-
-//bool TrimSeqsCommand::stripQualThreshold(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; } }
-//
-// 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);
-// }
-//}
-
-//***************************************************************************************************************
-
-//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