//**********************************************************************************************************************
vector<string> TrimFlowsCommand::setParameters(){
try {
- CommandParameter pflow("flow", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pflow);
- CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(poligos);
- CommandParameter pmaxhomop("maxhomop", "Number", "", "9", "", "", "",false,false); parameters.push_back(pmaxhomop);
- CommandParameter pmaxflows("maxflows", "Number", "", "450", "", "", "",false,false); parameters.push_back(pmaxflows);
- CommandParameter pminflows("minflows", "Number", "", "450", "", "", "",false,false); parameters.push_back(pminflows);
- CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(ppdiffs);
- CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(pbdiffs);
- CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(ptdiffs);
- CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
- CommandParameter psignal("signal", "Number", "", "0.50", "", "", "",false,false); parameters.push_back(psignal);
- CommandParameter pnoise("noise", "Number", "", "0.70", "", "", "",false,false); parameters.push_back(pnoise);
- CommandParameter pallfiles("allfiles", "Boolean", "", "t", "", "", "",false,false); parameters.push_back(pallfiles);
- CommandParameter porder("order", "String", "", "", "", "", "",false,false); parameters.push_back(porder);
- CommandParameter pfasta("fasta", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfasta);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ CommandParameter pflow("flow", "InputTypes", "", "", "none", "none", "none","flow",false,true,true); parameters.push_back(pflow);
+ CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none","",false,false,true); parameters.push_back(poligos);
+ CommandParameter pmaxhomop("maxhomop", "Number", "", "9", "", "", "","",false,false); parameters.push_back(pmaxhomop);
+ CommandParameter pmaxflows("maxflows", "Number", "", "450", "", "", "","",false,false); parameters.push_back(pmaxflows);
+ CommandParameter pminflows("minflows", "Number", "", "450", "", "", "","",false,false); parameters.push_back(pminflows);
+ 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 pldiffs("ldiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pldiffs);
+ CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(psdiffs);
+ CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ptdiffs);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
+ CommandParameter psignal("signal", "Number", "", "0.50", "", "", "","",false,false); parameters.push_back(psignal);
+ CommandParameter pnoise("noise", "Number", "", "0.70", "", "", "","",false,false); parameters.push_back(pnoise);
+ CommandParameter pallfiles("allfiles", "Boolean", "", "t", "", "", "","",false,false); parameters.push_back(pallfiles);
+ CommandParameter porder("order", "String", "", "TACG", "", "", "","",false,false); parameters.push_back(porder);
+ CommandParameter pfasta("fasta", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pfasta);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
vector<string> myArray;
for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
exit(1);
}
}
-
+//**********************************************************************************************************************
+string TrimFlowsCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "flow") { pattern = "[filename],[tag],flow"; }
+ else if (type == "fasta") { pattern = "[filename],flow.fasta"; }
+ else if (type == "file") { pattern = "[filename],[tag],flow.files"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "TrimFlowsCommand", "getOutputPattern");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
TrimFlowsCommand::TrimFlowsCommand(){
vector<string> tempOutNames;
outputTypes["flow"] = tempOutNames;
outputTypes["fasta"] = tempOutNames;
+ outputTypes["file"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "TrimFlowsCommand", "TrimFlowsCommand");
vector<string> tempOutNames;
outputTypes["flow"] = tempOutNames;
outputTypes["fasta"] = tempOutNames;
+ outputTypes["file"] = tempOutNames;
//if the user changes the input directory command factory will send this info to us in the output parameter
string inputDir = validParameter.validFile(parameters, "inputdir", false);
temp = validParameter.validFile(parameters, "pdiffs", false); if (temp == "not found"){ temp = "0"; }
m->mothurConvert(temp, pdiffs);
- temp = validParameter.validFile(parameters, "tdiffs", false);
- if (temp == "not found"){ int tempTotal = pdiffs + bdiffs; temp = toString(tempTotal); }
+ temp = validParameter.validFile(parameters, "ldiffs", false); if (temp == "not found") { temp = "0"; }
+ m->mothurConvert(temp, ldiffs);
+
+ temp = validParameter.validFile(parameters, "sdiffs", false); if (temp == "not found") { temp = "0"; }
+ m->mothurConvert(temp, sdiffs);
+
+ temp = validParameter.validFile(parameters, "tdiffs", false); if (temp == "not found") { int tempTotal = pdiffs + bdiffs + ldiffs + sdiffs; temp = toString(tempTotal); }
m->mothurConvert(temp, tdiffs);
- if(tdiffs == 0){ tdiffs = bdiffs + pdiffs; }
+
+ if(tdiffs == 0){ tdiffs = bdiffs + pdiffs + ldiffs + sdiffs; }
+
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
m->setProcessors(temp);
if (abort == true) { if (calledHelp) { return 0; } return 2; }
- string trimFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "trim.flow";
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(flowFileName));
+ string fastaFileName = getOutputFileName("fasta",variables);
+ if(fasta){ outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName); }
+
+ variables["[tag]"] = "trim";
+ string trimFlowFileName = getOutputFileName("flow",variables);
outputNames.push_back(trimFlowFileName); outputTypes["flow"].push_back(trimFlowFileName);
- string scrapFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "scrap.flow";
+ variables["[tag]"] = "scrap";
+ string scrapFlowFileName = getOutputFileName("flow",variables);
outputNames.push_back(scrapFlowFileName); outputTypes["flow"].push_back(scrapFlowFileName);
- string fastaFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.fasta";
- if(fasta){
- outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName);
- }
+
vector<unsigned long long> flowFilePos;
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
flowFilePos = getFlowFileBreaks();
for (int i = 0; i < (flowFilePos.size()-1); i++) {
lines.push_back(new linePair(flowFilePos[i], flowFilePos[(i+1)]));
if(allFiles){
set<string> namesAlreadyProcessed;
- flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
+ variables["[tag]"] = "";
+ flowFilesFileName = getOutputFileName("file",variables);
m->openOutputFile(flowFilesFileName, output);
for(int i=0;i<barcodePrimerComboFileNames.size();i++){
for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
if (namesAlreadyProcessed.count(barcodePrimerComboFileNames[i][j]) == 0) {
- FILE * pFile;
- unsigned long long size;
-
- //get num bytes in file
- pFile = fopen (barcodePrimerComboFileNames[i][j].c_str(),"rb");
- if (pFile==NULL) perror ("Error opening file");
- else{
- fseek (pFile, 0, SEEK_END);
- size=ftell(pFile);
- fclose (pFile);
- }
-
- if(size < 10){
- m->mothurRemove(barcodePrimerComboFileNames[i][j]);
- }
- else{
- output << barcodePrimerComboFileNames[i][j] << endl;
- outputNames.push_back(barcodePrimerComboFileNames[i][j]);
- outputTypes["flow"].push_back(barcodePrimerComboFileNames[i][j]);
- }
- namesAlreadyProcessed.insert(barcodePrimerComboFileNames[i][j]);
+ if (barcodePrimerComboFileNames[i][j] != "") {
+ FILE * pFile;
+ unsigned long long size;
+
+ //get num bytes in file
+ pFile = fopen (barcodePrimerComboFileNames[i][j].c_str(),"rb");
+ if (pFile==NULL) perror ("Error opening file");
+ else{
+ fseek (pFile, 0, SEEK_END);
+ size=ftell(pFile);
+ fclose (pFile);
+ }
+
+ if(size < 10){
+ m->mothurRemove(barcodePrimerComboFileNames[i][j]);
+ }
+ else{
+ output << m->getFullPathName(barcodePrimerComboFileNames[i][j]) << endl;
+ outputNames.push_back(barcodePrimerComboFileNames[i][j]);
+ outputTypes["flow"].push_back(barcodePrimerComboFileNames[i][j]);
+ }
+ namesAlreadyProcessed.insert(barcodePrimerComboFileNames[i][j]);
+ }
}
}
}
output.close();
}
else{
- flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
+ variables["[tag]"] = "";
+ flowFilesFileName = getOutputFileName("file",variables);
m->openOutputFile(flowFilesFileName, output);
- output << trimFlowFileName << endl;
+ output << m->getFullPathName(trimFlowFileName) << endl;
output.close();
}
- outputTypes["flow.files"].push_back(flowFilesFileName);
+ outputTypes["file"].push_back(flowFilesFileName);
outputNames.push_back(flowFilesFileName);
-
-// set fasta file as new current fastafile
-// string current = "";
-// itTypes = outputTypes.find("fasta");
-// if (itTypes != outputTypes.end()) {
-// if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
-// }
-
+
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
if(allFiles){
for(int i=0;i<thisBarcodePrimerComboFileNames.size();i++){
for(int j=0;j<thisBarcodePrimerComboFileNames[0].size();j++){
- ofstream temp;
- m->openOutputFile(thisBarcodePrimerComboFileNames[i][j], temp);
- temp << maxFlows << endl;
- temp.close();
+ if (thisBarcodePrimerComboFileNames[i][j] != "") {
+ ofstream temp;
+ m->openOutputFile(thisBarcodePrimerComboFileNames[i][j], temp);
+ temp << maxFlows << endl;
+ temp.close();
+ }
}
}
}
int count = 0;
bool moreSeqs = 1;
- TrimOligos trimOligos(pdiffs, bdiffs, primers, barcodes, revPrimer);
+ TrimOligos trimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, revPrimer, linker, spacer);
while(moreSeqs) {
- //cout << "driver " << count << endl;
-
+
if (m->control_pressed) { break; }
int success = 1;
string trashCode = "";
flowData.getNext(flowFile);
- //cout << "driver good bit " << flowFile.good() << endl;
flowData.capFlows(maxFlows);
Sequence currSeq = flowData.getSequence();
-
if(!flowData.hasMinFlows(minFlows)){ //screen to see if sequence is of a minimum number of flows
success = 0;
trashCode += 'l';
int primerIndex = 0;
int barcodeIndex = 0;
+ if(numLinkers != 0){
+ success = trimOligos.stripLinker(currSeq);
+ if(success > ldiffs) { trashCode += 'k'; }
+ else{ currentSeqDiffs += success; }
+
+ }
+
+ if (m->debug) { m->mothurOut("[DEBUG]: " + currSeq.getName() + " " + currSeq.getUnaligned() + "\n"); }
+
if(barcodes.size() != 0){
success = trimOligos.stripBarcode(currSeq, barcodeIndex);
if(success > bdiffs) { trashCode += 'b'; }
else{ currentSeqDiffs += success; }
}
+ if(numSpacers != 0){
+ success = trimOligos.stripSpacer(currSeq);
+ if(success > sdiffs) { trashCode += 's'; }
+ else{ currentSeqDiffs += success; }
+
+ }
+
if(numFPrimers != 0){
success = trimOligos.stripForward(currSeq, primerIndex);
if(success > pdiffs) { trashCode += 'f'; }
}
if(trashCode.length() == 0){
-
- flowData.printFlows(trimFlowFile);
-
- if(fasta) { currSeq.printSequence(fastaFile); }
-
- if(allFiles){
- ofstream output;
- m->openOutputFileAppend(thisBarcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
- output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
-
- flowData.printFlows(output);
- output.close();
- }
+ string thisGroup = "";
+ if(barcodes.size() != 0){
+ thisGroup = barcodeNameVector[barcodeIndex];
+ if (primers.size() != 0) {
+ if (primerNameVector[primerIndex] != "") {
+ if(thisGroup != "") {
+ thisGroup += "." + primerNameVector[primerIndex];
+ }else {
+ thisGroup = primerNameVector[primerIndex];
+ }
+ }
+ }
+ }
+
+ int pos = thisGroup.find("ignore");
+ if (pos == string::npos) {
+ flowData.printFlows(trimFlowFile);
+
+ if(fasta) { currSeq.printSequence(fastaFile); }
+
+ if(allFiles){
+ ofstream output;
+ m->openOutputFileAppend(thisBarcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
+ output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
+
+ flowData.printFlows(output);
+ output.close();
+ }
+ }
}
else{
flowData.printFlows(scrapFlowFile, trashCode);
//report progress
if((count) % 10000 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
unsigned long long pos = flowFile.tellg();
if ((pos == -1) || (pos >= line->end)) { break; }
}
else if(type == "REVERSE"){
- Sequence oligoRC("reverse", oligo);
- oligoRC.reverseComplement();
- revPrimer.push_back(oligoRC.getUnaligned());
+ string oligoRC = reverseOligo(oligo);
+ revPrimer.push_back(oligoRC);
}
else if(type == "BARCODE"){
oligosFile >> group;
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(type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine();
string primerName = primerNameVector[itPrimer->second];
string barcodeName = barcodeNameVector[itBar->second];
-
- string comboGroupName = "";
- string fileName = "";
-
- if(primerName == ""){
- comboGroupName = barcodeNameVector[itBar->second];
- fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
- }
- else{
- if(barcodeName == ""){
- comboGroupName = primerNameVector[itPrimer->second];
- }
- else{
- comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
- }
- fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
- }
-
- outFlowFileNames[itBar->second][itPrimer->second] = fileName;
-
- ofstream temp;
- m->openOutputFile(fileName, temp);
- temp.close();
+
+ if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
+ else {
+ string comboGroupName = "";
+ string fileName = "";
+
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(flowFileName));
+
+ if(primerName == ""){
+ comboGroupName = barcodeNameVector[itBar->second];
+ variables["[tag]"] = comboGroupName;
+ fileName = getOutputFileName("flow", variables);
+ }
+ else{
+ if(barcodeName == ""){
+ comboGroupName = primerNameVector[itPrimer->second];
+ }
+ else{
+ comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
+ }
+ variables["[tag]"] = comboGroupName;
+ fileName = getOutputFileName("flow", variables);
+ }
+
+ outFlowFileNames[itBar->second][itPrimer->second] = fileName;
+
+ ofstream temp;
+ m->openOutputFile(fileName, temp);
+ temp.close();
+ }
}
}
}
numFPrimers = primers.size();
numRPrimers = revPrimer.size();
+ numLinkers = linker.size();
+ numSpacers = spacer.size();
}
catch(exception& e) {
exit(1);
}
}
+//********************************************************************/
+string TrimFlowsCommand::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, "TrimFlowsCommand", "reverseOligo");
+ exit(1);
+ }
+}
+
/**************************************************************************************************/
vector<unsigned long long> TrimFlowsCommand::getFlowFileBreaks() {
processIDS.clear();
int exitCommand = 1;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
int process = 1;
//loop through and create all the processes you want
if(allFiles){
for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
- tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
- ofstream temp;
- m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
- temp.close();
-
+ if (tempBarcodePrimerComboFileNames[i][j] != "") {
+ tempBarcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
+ ofstream temp;
+ m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
+ temp.close();
+ }
}
}
}
if(allFiles){
for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
- tempBarcodePrimerComboFileNames[i][j] += extension;
- ofstream temp;
- m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
- temp.close();
-
+ if (tempBarcodePrimerComboFileNames[i][j] != "") {
+ tempBarcodePrimerComboFileNames[i][j] += extension;
+ ofstream temp;
+ m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
+ temp.close();
+ }
}
}
}
if(allFiles){
for(int i=0;i<tempBarcodePrimerComboFileNames.size();i++){
for(int j=0;j<tempBarcodePrimerComboFileNames[0].size();j++){
- tempBarcodePrimerComboFileNames[i][j] += toString(processors-1) + ".temp";
- ofstream temp;
- m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
- temp.close();
+ if (tempBarcodePrimerComboFileNames[i][j] != "") {
+ tempBarcodePrimerComboFileNames[i][j] += toString(processors-1) + ".temp";
+ ofstream temp;
+ m->openOutputFile(tempBarcodePrimerComboFileNames[i][j], temp);
+ temp.close();
+ }
}
}
if(allFiles){
for (int j = 0; j < barcodePrimerComboFileNames.size(); j++) {
for (int k = 0; k < barcodePrimerComboFileNames[0].size(); k++) {
- m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
- m->mothurRemove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"));
+ if (barcodePrimerComboFileNames[j][k] != "") {
+ m->appendFiles((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"), barcodePrimerComboFileNames[j][k]);
+ m->mothurRemove((barcodePrimerComboFileNames[j][k] + toString(processIDS[i]) + ".temp"));
+ }
}
}
}
projects / mothur.git / blobdiff