#include "needlemanoverlap.hpp"
//**********************************************************************************************************************
-
-vector<string> TrimFlowsCommand::getValidParameters(){
+vector<string> TrimFlowsCommand::setParameters(){
try {
- string Array[] = {"flow", "totalflows", "minflows",
- "fasta", "minlength", "maxlength", "maxhomop", "signal", "noise"
- "oligos", "pdiffs", "bdiffs", "tdiffs",
- "allfiles", "processors",
-
-
-
-// "group",
- "outputdir","inputdir"
+ 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", "", "720", "", "", "",false,false); parameters.push_back(pmaxflows);
+ CommandParameter pminflows("minflows", "Number", "", "360", "", "", "",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);
- };
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "TrimFlowsCommand", "getValidParameters");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
-vector<string> TrimFlowsCommand::getRequiredParameters(){
- try {
- string Array[] = {"flow"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
return myArray;
}
catch(exception& e) {
- m->errorOut(e, "TrimFlowsCommand", "getRequiredParameters");
+ m->errorOut(e, "TrimFlowsCommand", "setParameters");
exit(1);
}
}
-
//**********************************************************************************************************************
-
-vector<string> TrimFlowsCommand::getRequiredFiles(){
+string TrimFlowsCommand::getHelpString(){
try {
- vector<string> myArray;
- return myArray;
+ string helpString = "";
+ helpString += "The trim.flows command reads a flowgram file and creates .....\n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
+ helpString += "For more details please check out the wiki http://www.mothur.org/wiki/Trim.flows.\n";
+ return helpString;
}
catch(exception& e) {
- m->errorOut(e, "TrimFlowsCommand", "getRequiredFiles");
+ m->errorOut(e, "TrimFlowsCommand", "getHelpString");
exit(1);
}
}
TrimFlowsCommand::TrimFlowsCommand(){
try {
- abort = true;
- //initialize outputTypes
+ abort = true; calledHelp = true;
+ setParameters();
vector<string> tempOutNames;
outputTypes["flow"] = tempOutNames;
outputTypes["fasta"] = tempOutNames;
exit(1);
}
}
-
-//***************************************************************************************************************
-
-TrimFlowsCommand::~TrimFlowsCommand(){ /* do nothing */ }
-
-//***************************************************************************************************************
-
-void TrimFlowsCommand::help(){
- try {
- m->mothurOut("The trim.flows command reads a flowgram file and creates .....\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
- m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/Trim.flows.\n\n");
-
- }
- catch(exception& e) {
- m->errorOut(e, "TrimFlowsCommand", "help");
- exit(1);
- }
-}
-
//**********************************************************************************************************************
TrimFlowsCommand::TrimFlowsCommand(string option) {
try {
- abort = false;
+ abort = false; calledHelp = false;
comboStarts = 0;
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string AlignArray[] = {"flow", "totalflows", "minflows",
- "fasta", "minlength", "maxlength", "maxhomop", "signal", "noise",
- "oligos", "pdiffs", "bdiffs", "tdiffs",
- "allfiles", "processors",
-
- // "group",
- "outputdir","inputdir"
-
- };
-
- vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
+
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
if (path == "") { parameters["oligos"] = 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; }
-// }
}
//check for required parameters
flowFileName = validParameter.validFile(parameters, "flow", true);
- if (flowFileName == "not found") { m->mothurOut("flow is a required parameter for the trim.flows command."); m->mothurOutEndLine(); abort = true; }
- else if (flowFileName == "not open") { abort = true; }
+ if (flowFileName == "not found") {
+ flowFileName = m->getFlowFile();
+ if (flowFileName != "") { m->mothurOut("Using " + flowFileName + " as input file for the flow parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("No valid current flow file. You must provide a flow file."); m->mothurOutEndLine();
+ abort = true;
+ }
+ }else if (flowFileName == "not open") { flowFileName = ""; abort = true; }
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
// ...at some point should added some additional type checking...
string temp;
- temp = validParameter.validFile(parameters, "minflows", false); if (temp == "not found") { temp = "360"; }
+ temp = validParameter.validFile(parameters, "minflows", false); if (temp == "not found") { temp = "450"; }
convert(temp, minFlows);
- temp = validParameter.validFile(parameters, "totalflows", false); if (temp == "not found") { temp = "720"; }
- convert(temp, totalFlows);
+ temp = validParameter.validFile(parameters, "maxflows", false); if (temp == "not found") { temp = "450"; }
+ convert(temp, maxFlows);
temp = validParameter.validFile(parameters, "oligos", true);
if (temp == "not found") { oligoFileName = ""; }
else if(temp == "not open") { abort = true; }
- else { oligoFileName = temp; }
+ else { oligoFileName = temp; m->setOligosFile(oligoFileName); }
temp = validParameter.validFile(parameters, "fasta", false); if (temp == "not found"){ fasta = 0; }
else if(m->isTrue(temp)) { fasta = 1; }
temp = validParameter.validFile(parameters, "noise", false); if (temp == "not found"){ temp = "0.70"; }
convert(temp, noise);
-
- temp = validParameter.validFile(parameters, "minlength", false); if (temp == "not found"){ temp = "0"; }
- convert(temp, minLength);
-
- temp = validParameter.validFile(parameters, "maxlength", false); if (temp == "not found"){ temp = "0"; }
- convert(temp, maxLength);
-
+
temp = validParameter.validFile(parameters, "bdiffs", false); if (temp == "not found"){ temp = "0"; }
convert(temp, bdiffs);
convert(temp, tdiffs);
if(tdiffs == 0){ tdiffs = bdiffs + pdiffs; }
- temp = validParameter.validFile(parameters, "allfiles", false); if (temp == "not found") { temp = "T"; }
- allFiles = m->isTrue(temp);
-
- temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found") { temp = "1"; }
- convert(temp, processors);
-
- if(oligoFileName == ""){ allFiles = 0; }
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
+ m->setProcessors(temp);
+ convert(temp, processors);
+
+ flowOrder = validParameter.validFile(parameters, "order", false);
+ if (flowOrder == "not found"){ flowOrder = "TACG"; }
+ else if(flowOrder.length() != 4){
+ m->mothurOut("The value of the order option must be four bases long\n");
+ }
+
+ if(oligoFileName == "") { allFiles = 0; }
+ else { allFiles = 1; }
numFPrimers = 0;
numRPrimers = 0;
int TrimFlowsCommand::execute(){
try{
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
string trimFlowFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "trim.flow";
outputNames.push_back(trimFlowFileName); outputTypes["flow"].push_back(trimFlowFileName);
if (m->control_pressed) { return 0; }
+ string flowFilesFileName;
+ ofstream output;
+
if(allFiles){
- ofstream output;
- string flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
+ flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
m->openOutputFile(flowFilesFileName, output);
for(int i=0;i<barcodePrimerComboFileNames.size();i++){
if (pFile==NULL) perror ("Error opening file");
else{
fseek (pFile, 0, SEEK_END);
- size=ftell (pFile);
+ size=ftell(pFile);
fclose (pFile);
}
if(size < 10){
-// m->mothurOut("deleting: " + barcodePrimerComboFileNames[i][j] + '\n');
remove(barcodePrimerComboFileNames[i][j].c_str());
}
else{
-// m->mothurOut("saving: " + barcodePrimerComboFileNames[i][j] + '\n');
output << barcodePrimerComboFileNames[i][j] << endl;
+ outputNames.push_back(barcodePrimerComboFileNames[i][j]);
+ outputTypes["flow"].push_back(barcodePrimerComboFileNames[i][j]);
}
}
}
output.close();
}
+ else{
+ flowFilesFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + "flow.files";
+ m->openOutputFile(flowFilesFileName, output);
+
+ output << trimFlowFileName << endl;
+
+ output.close();
+ }
+ outputTypes["flow.files"].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();
scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
if(line->start == 4){
- scrapFlowFile << totalFlows << endl;
- trimFlowFile << totalFlows << endl;
- for(int i=0;i<barcodePrimerComboFileNames.size();i++){
- for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
- // barcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
- ofstream temp;
- m->openOutputFile(barcodePrimerComboFileNames[i][j], temp);
- temp << totalFlows << endl;
- temp.close();
- }
- }
+ scrapFlowFile << maxFlows << endl;
+ trimFlowFile << maxFlows << endl;
+ if(allFiles){
+ for(int i=0;i<barcodePrimerComboFileNames.size();i++){
+ for(int j=0;j<barcodePrimerComboFileNames[0].size();j++){
+ // barcodePrimerComboFileNames[i][j] += toString(getpid()) + ".temp";
+ ofstream temp;
+ m->openOutputFile(barcodePrimerComboFileNames[i][j], temp);
+ temp << maxFlows << endl;
+ temp.close();
+ }
+ }
+ }
}
- FlowData flowData(numFlows, signal, noise, maxHomoP);
+ FlowData flowData(numFlows, signal, noise, maxHomoP, flowOrder);
ofstream fastaFile;
if(fasta){ m->openOutputFile(fastaFileName, fastaFile); }
int count = 0;
bool moreSeqs = 1;
-
+
while(moreSeqs) {
int success = 1;
string trashCode = "";
flowData.getNext(flowFile);
- flowData.capFlows(totalFlows);
+ flowData.capFlows(maxFlows);
Sequence currSeq = flowData.getSequence();
if(!flowData.hasMinFlows(minFlows)){ //screen to see if sequence is of a minimum number of flows
trashCode += 'l';
}
- if(minLength > 0 || maxLength > 0){ //screen to see if sequence is above and below a specific number of bases
- int seqLength = currSeq.getNumBases();
- if(seqLength < minLength || seqLength > maxLength){
- success = 0;
- trashCode += 'l';
- }
- }
-
- int primerIndex, barcodeIndex;
+ int primerIndex = 0;
+ int barcodeIndex = 0;
if(barcodes.size() != 0){
success = stripBarcode(currSeq, barcodeIndex);
success = stripReverse(currSeq);
if(!success) { trashCode += 'r'; }
}
-
-
+
if(trashCode.length() == 0){
+
flowData.printFlows(trimFlowFile);
+
+ if(fasta) { currSeq.printSequence(fastaFile); }
- if(fasta){
- currSeq.printSequence(fastaFile);
- }
-
- if(barcodes.size() != 0 || primers.size() != 0){
-// string fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + barcodePrimerCombos[barcodeIndex][primerIndex] + ".flow";
+ if(allFiles){
ofstream output;
m->openOutputFileAppend(barcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
flowData.printFlows(output);
output.close();
- }
-
+ }
}
else{
flowData.printFlows(scrapFlowFile, trashCode);
}
oligosFile.close();
- //add in potential combos
+ if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ allFiles = 0; }
+
+ //add in potential combos
+ if(barcodeNameVector.size() == 0){
+ barcodes[""] = 0;
+ barcodeNameVector.push_back("");
+ }
+
+ if(primerNameVector.size() == 0){
+ primers[""] = 0;
+ primerNameVector.push_back("");
+ }
+
+
outFlowFileNames.resize(barcodeNameVector.size());
for(int i=0;i<outFlowFileNames.size();i++){
outFlowFileNames[i].assign(primerNameVector.size(), "");
}
if(allFiles){
+
for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
}
else{
- comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
+ if(barcodeName == ""){
+ comboGroupName = primerNameVector[itPrimer->second];
+ }
+ else{
+ comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
+ }
fileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + comboGroupName + ".flow";
}
- outputNames.push_back(fileName);
- outputTypes["flow"].push_back(fileName);
outFlowFileNames[itBar->second][itPrimer->second] = fileName;
ofstream temp;
if (barcodes.size() > 0) {
map<string,int>::iterator it=barcodes.begin();
- for(it;it!=barcodes.end();it++){
+ for(map<string,int>::iterator it=barcodes.begin();it!=barcodes.end();it++){
if(it->first.length() > maxLength){
maxLength = it->first.length();
}
oligo = oligo.substr(0,alnLength);
temp = temp.substr(0,alnLength);
- int newStart=0;
int numDiff = countDiffs(oligo, temp);
if(numDiff < minDiff){
string rawSequence = seq.getUnaligned();
int success = pdiffs + 1; //guilty until proven innocent
-
+
//can you find the primer
for(map<string,int>::iterator it=primers.begin();it!=primers.end();it++){
string oligo = it->first;
}
//if you found the barcode or if you don't want to allow for diffs
- if ((pdiffs == 0) || (success == 0)) { return success; }
+ if ((pdiffs == 0) || (success == 0)) { return success; }
else { //try aligning and see if you can find it
oligo = oligo.substr(0,alnLength);
temp = temp.substr(0,alnLength);
- int newStart=0;
int numDiff = countDiffs(oligo, temp);
if(numDiff < minDiff){
m->openInputFile(flowFileName, in);
in.seekg(spot);
- unsigned long int newSpot = spot;
string dummy = m->getline(in);
-
//there was not another sequence before the end of the file
unsigned long int sanityPos = in.tellg();
}else if (pid == 0){
vector<vector<string> > tempBarcodePrimerComboFileNames = barcodePrimerComboFileNames;
- 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(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();
+
+ }
}
}
-
driverCreateTrim(flowFileName,
(trimFlowFileName + toString(getpid()) + ".temp"),
(scrapFlowFileName + toString(getpid()) + ".temp"),
projects / mothur.git / blobdiff