//**********************************************************************************************************************
vector<string> SensSpecCommand::setParameters(){
try {
- CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
- CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none",false,false); parameters.push_back(pphylip);
- //CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName",false,false); parameters.push_back(pname);
- CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none",false,false); parameters.push_back(pcolumn);
- CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
- CommandParameter pcutoff("cutoff", "Number", "", "-1.00", "", "", "",false,false); parameters.push_back(pcutoff);
- CommandParameter pprecision("precision", "Number", "", "100", "", "", "",false,false); parameters.push_back(pprecision);
- CommandParameter phard("hard", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(phard);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none","sensspec",false,true,true); parameters.push_back(plist);
+ CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none","",false,false); parameters.push_back(pphylip);
+ CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none","",false,false); parameters.push_back(pcolumn);
+ CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
+ CommandParameter pcutoff("cutoff", "Number", "", "-1.00", "", "", "","",false,false); parameters.push_back(pcutoff);
+ CommandParameter pprecision("precision", "Number", "", "100", "", "", "","",false,false); parameters.push_back(pprecision);
+ CommandParameter phard("hard", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(phard);
+ 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); }
}
}
//**********************************************************************************************************************
+string SensSpecCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "sensspec") { pattern = "[filename],sensspec"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SensSpecCommand", "getOutputPattern");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
SensSpecCommand::SensSpecCommand(){
try {
abort = true; calledHelp = true;
SensSpecCommand::SensSpecCommand(string option) {
try {
- abort = false; calledHelp = false;
+ abort = false; calledHelp = false;
+ allLines = 1;
//allow user to run help
if(option == "help") { help(); abort = true; calledHelp = true; }
path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["column"] = inputDir + it->second; }
- }
-
- //it = parameters.find("name");
- //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["name"] = inputDir + it->second; }
- //}
-
+ }
}
//check for required parameters
listFile = validParameter.validFile(parameters, "list", true);
else if(!m->isTrue(temp)) { hard = 0; }
else if(m->isTrue(temp)) { hard = 1; }
-// temp = validParameter.validFile(parameters, "name", true);
-// if (temp == "not found") { nameFile = ""; }
-// else if(temp == "not open") { abort = true; }
-// else { nameFile = temp; }
-// cout << "name:\t" << nameFile << endl;
-
temp = validParameter.validFile(parameters, "cutoff", false); if (temp == "not found") { temp = "-1.00"; }
- convert(temp, cutoff);
+ m->mothurConvert(temp, cutoff);
// cout << cutoff << endl;
temp = validParameter.validFile(parameters, "precision", false); if (temp == "not found") { temp = "100"; }
- convert(temp, precision);
+ m->mothurConvert(temp, precision);
// cout << precision << endl;
- lineLabel = validParameter.validFile(parameters, "label", false); if (lineLabel == "not found") { lineLabel = ""; }
+ string label = validParameter.validFile(parameters, "label", false);
+ if (label == "not found") { label = ""; }
+ else {
+ if(label != "all") { m->splitAtDash(label, labels); allLines = 0; }
+ else { allLines = 1; }
+ }
- sensSpecFileName = outputDir + m->getRootName(m->getSimpleName(listFile)) + ".sensspec";
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(listFile));
+ sensSpecFileName = getOutputFileName("sensspec",variables);
}
}
catch(exception& e) {
int SensSpecCommand::execute(){
try{
if (abort == true) { if (calledHelp) { return 0; } return 2; }
-
+
+ int startTime = time(NULL);
+
+ //create list file with only unique names, saves time and memory by removing redundant names from list file that are not in the distance file.
+ string newListFile = preProcessList();
+ if (newListFile != "") { listFile = newListFile; }
+
setUpOutput();
outputNames.push_back(sensSpecFileName); outputTypes["sensspec"].push_back(sensSpecFileName);
if(format == "phylip") { processPhylip(); }
else if(format == "column") { processColumn(); }
+ //remove temp file if created
+ if (newListFile != "") { m->mothurRemove(newListFile); }
+
+ if (m->control_pressed) { m->mothurRemove(sensSpecFileName); return 0; }
+
+ m->mothurOut("It took " + toString(time(NULL) - startTime) + " to run sens.spec."); m->mothurOutEndLine();
+
m->mothurOutEndLine();
- m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();
m->mothurOut(sensSpecFileName); m->mothurOutEndLine();
m->mothurOutEndLine();
//***************************************************************************************************************
-void SensSpecCommand::processPhylip(){
+int SensSpecCommand::processPhylip(){
try{
//probably need some checking to confirm that the names in the distance matrix are the same as those in the list file
-
- ifstream inputListFile;
- m->openInputFile(listFile, inputListFile);
-
string origCutoff = "";
bool getCutoff = 0;
if(cutoff == -1.00) { getCutoff = 1; }
- else { origCutoff = toString(cutoff); cutoff += (0.49 / double(precision)); }
+ else { origCutoff = toString(cutoff); cutoff += (0.49 / double(precision)); }
- string label;
- int numOTUs;
-
map<string, int> seqMap;
string seqList;
- while(inputListFile){
- inputListFile >> label >> numOTUs;
- for(int i=0;i<numOTUs;i++){
- inputListFile >> seqList;
- int seqListLength = seqList.length();
- string seqName = "";
- for(int j=0;j<seqListLength;j++){
-
- if(seqList[j] == ','){
- seqMap[seqName] = i;
- seqName = "";
- }
- else{
- seqName += seqList[j];
- }
-
- }
- seqMap[seqName] = i;
+ InputData input(listFile, "list");
+ ListVector* list = input.getListVector();
+ string lastLabel = list->getLabel();
+
+ //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
+ set<string> processedLabels;
+ set<string> userLabels = labels;
+
+
+ while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
+
+ if(m->control_pressed){
+ for (int i = 0; i < outputNames.size(); i++){ m->mothurRemove(outputNames[i]); } delete list; return 0;
+ }
+
+ if(allLines == 1 || labels.count(list->getLabel()) == 1){
+
+ processedLabels.insert(list->getLabel());
+ userLabels.erase(list->getLabel());
+
+ //process
+ fillSeqMap(seqMap, list);
+ process(seqMap, list->getLabel(), getCutoff, origCutoff);
+ }
+
+ if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
+ string saveLabel = list->getLabel();
+
+ delete list;
+ list = input.getListVector(lastLabel);
+
+ processedLabels.insert(list->getLabel());
+ userLabels.erase(list->getLabel());
+
+ //process
+ fillSeqMap(seqMap, list);
+ process(seqMap, list->getLabel(), getCutoff, origCutoff);
+
+ //restore real lastlabel to save below
+ list->setLabel(saveLabel);
+ }
+
+ lastLabel = list->getLabel();
+
+ delete list;
+ list = input.getListVector();
+ }
+
+
+ //output error messages about any remaining user labels
+ set<string>::iterator it;
+ bool needToRun = false;
+ for (it = userLabels.begin(); it != userLabels.end(); it++) {
+ m->mothurOut("Your file does not include the label " + *it);
+ if (processedLabels.count(lastLabel) != 1) {
+ m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
+ needToRun = true;
+ }else {
+ m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
}
- m->gobble(inputListFile);
+ }
+
+ //run last label if you need to
+ if (needToRun == true) {
+ if (list != NULL) { delete list; }
+ list = input.getListVector(lastLabel);
+
+ //process
+ fillSeqMap(seqMap, list);
+ process(seqMap, list->getLabel(), getCutoff, origCutoff);
+
+ delete list;
+ }
- int lNumSeqs = seqMap.size();
- int pNumSeqs = 0;
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SensSpecCommand", "processPhylip");
+ exit(1);
+ }
+}
+
+//***************************************************************************************************************
- ifstream phylipFile;
- m->openInputFile(distFile, phylipFile);
- phylipFile >> pNumSeqs;
- if(pNumSeqs != lNumSeqs){ cout << "numSeq mismatch!" << endl; }
-
- string seqName;
- double distance;
- vector<int> otuIndices(lNumSeqs, -1);
+int SensSpecCommand::fillSeqMap(map<string, int>& seqMap, ListVector*& list){
+ try {
+ //for each otu
+ for(int i=0;i<list->getNumBins();i++){
+
+ if (m->control_pressed) { return 0; }
+
+ string seqList = list->get(i);
+ int seqListLength = seqList.length();
+ string seqName = "";
+
+ //parse bin by name, mapping each name to its otu number
+ for(int j=0;j<seqListLength;j++){
- truePositives = 0;
- falsePositives = 0;
- trueNegatives = 0;
- falseNegatives = 0;
-
- if(getCutoff == 1){
- if(label != "unique"){
- origCutoff = label;
- convert(label, cutoff);
- if(hard == 0){ cutoff += (0.49 / double(precision)); }
+ if(seqList[j] == ','){
+ seqMap[seqName] = i;
+ seqName = "";
}
else{
- origCutoff = "unique";
- cutoff = 0.0000;
+ seqName += seqList[j];
}
+
}
-
- cout << label << endl;
+ seqMap[seqName] = i;
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SensSpecCommand", "fillSeqMap");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+int SensSpecCommand::fillSeqPairSet(set<string>& seqPairSet, ListVector*& list){
+ try {
+ int numSeqs = 0;
+
+ //for each otu
+ for(int i=0;i<list->getNumBins();i++){
- for(int i=0;i<lNumSeqs;i++){
- phylipFile >> seqName;
- otuIndices[i] = seqMap[seqName];
+ if (m->control_pressed) { return 0; }
+
+ vector<string> seqNameVector;
+ string bin = list->get(i);
+ m->splitAtComma(bin, seqNameVector);
+
+ numSeqs += seqNameVector.size();
+
+ for(int j=0;j<seqNameVector.size();j++){
+ string seqPairString = "";
+ for(int k=0;k<j;k++){
+ if(seqNameVector[j] < seqNameVector[k]) { seqPairString = seqNameVector[j] + '\t' + seqNameVector[k]; }
+ else { seqPairString = seqNameVector[k] + '\t' + seqNameVector[j]; }
+ seqPairSet.insert(seqPairString);
+ }
+ }
+ }
+
+ return numSeqs;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SensSpecCommand", "fillSeqPairSet");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+int SensSpecCommand::process(map<string, int>& seqMap, string label, bool& getCutoff, string& origCutoff){
+ try {
+
+ int lNumSeqs = seqMap.size();
+ int pNumSeqs = 0;
+
+ ifstream phylipFile;
+ m->openInputFile(distFile, phylipFile);
+ phylipFile >> pNumSeqs;
+ if(pNumSeqs != lNumSeqs){ m->mothurOut("numSeq mismatch!\n"); /*m->control_pressed = true;*/ }
+
+ string seqName;
+ double distance;
+ vector<int> otuIndices(lNumSeqs, -1);
+
+ truePositives = 0;
+ falsePositives = 0;
+ trueNegatives = 0;
+ falseNegatives = 0;
+
+ if(getCutoff == 1){
+ if(label != "unique"){
+ origCutoff = label;
+ convert(label, cutoff);
+ if(hard == 0){ cutoff += (0.49 / double(precision)); }
+ }
+ else{
+ origCutoff = "unique";
+ cutoff = 0.0000;
+ }
+ }
+
+ m->mothurOut(label); m->mothurOutEndLine();
+
+ for(int i=0;i<pNumSeqs;i++){
+
+ if (m->control_pressed) { return 0; }
+
+ phylipFile >> seqName;
+ otuIndices[i] = seqMap[seqName];
+
+ for(int j=0;j<i;j++){
+ phylipFile >> distance;
- for(int j=0;j<i;j++){
- phylipFile >> distance;
-
- if(distance <= cutoff){
- if(otuIndices[i] == otuIndices[j]) { truePositives++; }
- else { falseNegatives++; }
- }
- else{
- if(otuIndices[i] == otuIndices[j]) { falsePositives++; }
- else { trueNegatives++; }
- }
+ if(distance <= cutoff){
+ if(otuIndices[i] == otuIndices[j]) { truePositives++; }
+ else { falseNegatives++; }
+ }
+ else{
+ if(otuIndices[i] == otuIndices[j]) { falsePositives++; }
+ else { trueNegatives++; }
}
}
- phylipFile.close();
+ }
+ phylipFile.close();
+
+ outputStatistics(label, origCutoff);
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SensSpecCommand", "process");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+int SensSpecCommand::process(set<string>& seqPairSet, string label, bool& getCutoff, string& origCutoff, int numSeqs){
+ try {
+ int numDists = (numSeqs * (numSeqs-1) / 2);
+
+ ifstream columnFile;
+ m->openInputFile(distFile, columnFile);
+ string seqNameA, seqNameB, seqPairString;
+ double distance;
+
+ truePositives = 0;
+ falsePositives = 0;
+ trueNegatives = numDists;
+ falseNegatives = 0;
+
+ if(getCutoff == 1){
+ if(label != "unique"){
+ origCutoff = label;
+ convert(label, cutoff);
+ if(hard == 0){ cutoff += (0.49 / double(precision)); }
+ }
+ else{
+ origCutoff = "unique";
+ cutoff = 0.0000;
+ }
+ }
+
+ m->mothurOut(label); m->mothurOutEndLine();
+
+ while(columnFile){
+ columnFile >> seqNameA >> seqNameB >> distance;
+ if(seqNameA < seqNameB) { seqPairString = seqNameA + '\t' + seqNameB; }
+ else { seqPairString = seqNameB + '\t' + seqNameA; }
+
+ set<string>::iterator it = seqPairSet.find(seqPairString);
- outputStatistics(label, origCutoff);
+ if(distance <= cutoff){
+ if(it != seqPairSet.end()){
+ truePositives++;
+ seqPairSet.erase(it);
+ }
+ else{
+ falseNegatives++;
+ }
+ trueNegatives--;
+ }
+ else if(it != seqPairSet.end()){
+ falsePositives++;
+ trueNegatives--;
+ seqPairSet.erase(it);
+ }
+
+ m->gobble(columnFile);
}
- inputListFile.close();
-
+ falsePositives += seqPairSet.size();
+
+ outputStatistics(label, origCutoff);
+
+
+ return 0;
}
catch(exception& e) {
- m->errorOut(e, "SensSpecCommand", "processPhylip");
+ m->errorOut(e, "SensSpecCommand", "process");
exit(1);
}
}
-
//***************************************************************************************************************
-void SensSpecCommand::processColumn(){
+int SensSpecCommand::processColumn(){
try{
- ifstream inputListFile;
- m->openInputFile(listFile, inputListFile);
-
string origCutoff = "";
bool getCutoff = 0;
if(cutoff == -1.00) { getCutoff = 1; }
else { origCutoff = toString(cutoff); cutoff += (0.49 / double(precision)); }
set<string> seqPairSet;
+ int numSeqs = 0;
- string label, seqList;
- int numOTUs;
- int numSeqs;
+ InputData input(listFile, "list");
+ ListVector* list = input.getListVector();
+ string lastLabel = list->getLabel();
- while(inputListFile){
- numSeqs = 0;
+ //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
+ set<string> processedLabels;
+ set<string> userLabels = labels;
+
+
+ while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
- inputListFile >> label >> numOTUs;
- for(int i=0;i<numOTUs;i++){
-
- vector<string> seqNameVector;
+ if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete list; return 0; }
+
+ if(allLines == 1 || labels.count(list->getLabel()) == 1){
- inputListFile >> seqList;
- int seqListLength = seqList.length();
- string seqName = "";
- for(int j=0;j<seqListLength;j++){
-
- if(seqList[j] == ','){
- seqNameVector.push_back(seqName);
- seqName = "";
- }
- else{
- seqName += seqList[j];
- }
- }
- seqNameVector.push_back(seqName);
-
- numSeqs += seqNameVector.size();
+ processedLabels.insert(list->getLabel());
+ userLabels.erase(list->getLabel());
- int numSeqsInOTU = seqNameVector.size();
- for(int j=0;j<numSeqsInOTU;j++){
- string seqPairString = "";
- for(int k=0;k<j;k++){
- if(seqNameVector[j] < seqNameVector[k]) { seqPairString = seqNameVector[j] + '\t' + seqNameVector[k]; }
- else { seqPairString = seqNameVector[k] + '\t' + seqNameVector[j]; }
- seqPairSet.insert(seqPairString);
- }
- }
+ //process
+ numSeqs = fillSeqPairSet(seqPairSet, list);
+ process(seqPairSet, list->getLabel(), getCutoff, origCutoff, numSeqs);
}
- m->gobble(inputListFile);
- int numDists = (numSeqs * (numSeqs-1) / 2);
-
- ifstream columnFile;
- m->openInputFile(distFile, columnFile);
- string seqNameA, seqNameB, seqPairString;
- double distance;
-
- truePositives = 0;
- falsePositives = 0;
- trueNegatives = numDists;
- falseNegatives = 0;
-
- if(getCutoff == 1){
- if(label != "unique"){
- origCutoff = label;
- convert(label, cutoff);
- if(hard == 0){ cutoff += (0.49 / double(precision)); }
- }
- else{
- origCutoff = "unique";
- cutoff = 0.0000;
- }
- }
-
- cout << label << endl;
+ if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
+ string saveLabel = list->getLabel();
+
+ delete list;
+ list = input.getListVector(lastLabel);
+
+ processedLabels.insert(list->getLabel());
+ userLabels.erase(list->getLabel());
+
+ //process
+ numSeqs = fillSeqPairSet(seqPairSet, list);
+ process(seqPairSet, list->getLabel(), getCutoff, origCutoff, numSeqs);
+
+ //restore real lastlabel to save below
+ list->setLabel(saveLabel);
+ }
- while(columnFile){
- columnFile >> seqNameA >> seqNameB >> distance;
- if(seqNameA < seqNameB) { seqPairString = seqNameA + '\t' + seqNameB; }
- else { seqPairString = seqNameB + '\t' + seqNameA; }
-
- set<string>::iterator it = seqPairSet.find(seqPairString);
+ lastLabel = list->getLabel();
- if(distance <= cutoff){
- if(it != seqPairSet.end()){
- truePositives++;
- seqPairSet.erase(it);
- }
- else{
- falseNegatives++;
- }
- trueNegatives--;
- }
- else if(it != seqPairSet.end()){
- falsePositives++;
- trueNegatives--;
- seqPairSet.erase(it);
- }
-
- m->gobble(columnFile);
+ delete list;
+ list = input.getListVector();
+ }
+
+
+ //output error messages about any remaining user labels
+ set<string>::iterator it;
+ bool needToRun = false;
+ for (it = userLabels.begin(); it != userLabels.end(); it++) {
+ m->mothurOut("Your file does not include the label " + *it);
+ if (processedLabels.count(lastLabel) != 1) {
+ m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
+ needToRun = true;
+ }else {
+ m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
}
- falsePositives += seqPairSet.size();
+ }
+
+ //run last label if you need to
+ if (needToRun == true) {
+ if (list != NULL) { delete list; }
+ list = input.getListVector(lastLabel);
- outputStatistics(label, origCutoff);
+ //process
+ numSeqs = fillSeqPairSet(seqPairSet, list);
+ delete list;
+ process(seqPairSet, list->getLabel(), getCutoff, origCutoff, numSeqs);
}
+
+ return 0;
}
catch(exception& e) {
m->errorOut(e, "SensSpecCommand", "processColumn");
exit(1);
}
}
+//***************************************************************************************************************
+
+string SensSpecCommand::preProcessList(){
+ try {
+ set<string> uniqueNames;
+ //get unique names from distance file
+ if (format == "phylip") {
+
+ ifstream phylipFile;
+ m->openInputFile(distFile, phylipFile);
+ string numTest;
+ int pNumSeqs;
+ phylipFile >> numTest;
+
+ if (!m->isContainingOnlyDigits(numTest)) { m->mothurOut("[ERROR]: expected a number and got " + numTest + ", quitting."); m->mothurOutEndLine(); exit(1); }
+ else {
+ m->mothurConvert(numTest, pNumSeqs);
+ }
+ phylipFile >> pNumSeqs; m->gobble(phylipFile);
+
+ string seqName;
+ double distance;
+
+ for(int i=0;i<pNumSeqs;i++){
+
+ if (m->control_pressed) { return ""; }
+
+ phylipFile >> seqName;
+ uniqueNames.insert(seqName);
+
+ for(int j=0;j<i;j++){
+ phylipFile >> distance;
+ }
+ m->gobble(phylipFile);
+ }
+ phylipFile.close();
+ }else {
+ ifstream columnFile;
+ m->openInputFile(distFile, columnFile);
+ string seqNameA, seqNameB;
+ double distance;
+
+ while(columnFile){
+ if (m->control_pressed) { return ""; }
+ columnFile >> seqNameA >> seqNameB >> distance;
+ uniqueNames.insert(seqNameA); uniqueNames.insert(seqNameB);
+ m->gobble(columnFile);
+ }
+ columnFile.close();
+ }
+
+ //read list file, if numSeqs > unique names then remove redundant names
+ string newListFile = listFile + ".temp";
+ ofstream out;
+ m->openOutputFile(newListFile, out);
+ ifstream in;
+ m->openInputFile(listFile, in);
+
+ bool wroteSomething = false;
+
+ while(!in.eof()){
+
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(newListFile); return ""; }
+
+ //read in list vector
+ ListVector list(in);
+
+ //listfile is already unique
+ if (list.getNumSeqs() == uniqueNames.size()) { in.close(); out.close(); m->mothurRemove(newListFile); return ""; }
+
+ //make a new list vector
+ ListVector newList;
+ newList.setLabel(list.getLabel());
+
+ //for each bin
+ for (int i = 0; i < list.getNumBins(); i++) {
+
+ //parse out names that are in accnos file
+ string binnames = list.get(i);
+ vector<string> bnames;
+ m->splitAtComma(binnames, bnames);
+
+ string newNames = "";
+ for (int i = 0; i < bnames.size(); i++) {
+ string name = bnames[i];
+ //if that name is in the .accnos file, add it
+ if (uniqueNames.count(name) != 0) { newNames += name + ","; }
+ }
+
+ //if there are names in this bin add to new list
+ if (newNames != "") {
+ newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
+ newList.push_back(newNames);
+ }
+ }
+
+ //print new listvector
+ if (newList.getNumBins() != 0) {
+ wroteSomething = true;
+ newList.print(out);
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething) { return newListFile; }
+ return "";
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SensSpecCommand", "preProcessList");
+ exit(1);
+ }
+}
+
//***************************************************************************************************************
projects / mothur.git / blobdiff