#include "chimerapintailcommand.h"
#include "pintail.h"
+
//**********************************************************************************************************************
vector<string> ChimeraPintailCommand::setParameters(){
try {
- CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
- CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
- CommandParameter pconservation("conservation", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pconservation);
- CommandParameter pquantile("quantile", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pquantile);
- CommandParameter pfilter("filter", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfilter);
- CommandParameter pwindow("window", "Number", "", "0", "", "", "",false,false); parameters.push_back(pwindow);
- CommandParameter pincrement("increment", "Number", "", "25", "", "", "",false,false); parameters.push_back(pincrement);
- CommandParameter pmask("mask", "String", "", "", "", "", "",false,false); parameters.push_back(pmask);
- CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
-
+ CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(ptemplate);
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","chimera-accnos",false,true,true); parameters.push_back(pfasta);
+ CommandParameter pconservation("conservation", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(pconservation);
+ CommandParameter pquantile("quantile", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(pquantile);
+ CommandParameter pfilter("filter", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pfilter);
+ CommandParameter pwindow("window", "Number", "", "0", "", "", "","","",false,false); parameters.push_back(pwindow);
+ CommandParameter pincrement("increment", "Number", "", "25", "", "", "","",false,false); parameters.push_back(pincrement);
+ CommandParameter pmask("mask", "String", "", "", "", "", "","",false,false); parameters.push_back(pmask);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
+ CommandParameter psave("save", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(psave);
+
vector<string> myArray;
for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
return myArray;
try {
string helpString = "";
helpString += "The chimera.pintail command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
- helpString += "This command was created using the algorythms described in the 'At Least 1 in 20 16S rRNA Sequence Records Currently Held in the Public Repositories is Estimated To Contain Substantial Anomalies' paper by Kevin E. Ashelford 1, Nadia A. Chuzhanova 3, John C. Fry 1, Antonia J. Jones 2 and Andrew J. Weightman 1.\n";
+ helpString += "This command was created using the algorithms described in the 'At Least 1 in 20 16S rRNA Sequence Records Currently Held in the Public Repositories is Estimated To Contain Substantial Anomalies' paper by Kevin E. Ashelford 1, Nadia A. Chuzhanova 3, John C. Fry 1, Antonia J. Jones 2 and Andrew J. Weightman 1.\n";
helpString += "The chimera.pintail command parameters are fasta, reference, filter, mask, processors, window, increment, conservation and quantile.\n";
helpString += "The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required unless you have a valid current fasta file. \n";
helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
#ifdef USE_MPI
helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
#endif
+ helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
helpString += "The window parameter allows you to specify the window size for searching for chimeras, default=300. \n";
helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=25.\n";
helpString += "The conservation parameter allows you to enter a frequency file containing the highest bases frequency at each place in the alignment.\n";
}
}
//**********************************************************************************************************************
+string ChimeraPintailCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "chimera") { pattern = "[filename],[tag],pintail.chimeras-[filename],pintail.chimeras"; }
+ else if (type == "accnos") { pattern = "[filename],[tag],pintail.accnos-[filename],pintail.accnos"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ChimeraPintailCommand", "getOutputPattern");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
ChimeraPintailCommand::ChimeraPintailCommand(){
try {
abort = true; calledHelp = true;
vector<string> tempOutNames;
outputTypes["chimera"] = tempOutNames;
outputTypes["accnos"] = tempOutNames;
+
}
catch(exception& e) {
m->errorOut(e, "ChimeraPintailCommand", "ChimeraPintailCommand");
ChimeraPintailCommand::ChimeraPintailCommand(string option) {
try {
abort = false; calledHelp = false;
+ rdb = ReferenceDB::getInstance();
//allow user to run help
if(option == "help") { help(); abort = true; calledHelp = true; }
vector<string> tempOutNames;
outputTypes["chimera"] = tempOutNames;
outputTypes["accnos"] = tempOutNames;
+
//if the user changes the input directory command factory will send this info to us in the output parameter
inputDir = validParameter.validFile(parameters, "inputdir", false);
//erase from file list
fastaFileNames.erase(fastaFileNames.begin()+i);
i--;
+ }else {
+ m->setFastaFile(fastaFileNames[i]);
}
}
}
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
m->setProcessors(temp);
- convert(temp, processors);
+ m->mothurConvert(temp, processors);
temp = validParameter.validFile(parameters, "window", false); if (temp == "not found") { temp = "0"; }
- convert(temp, window);
+ m->mothurConvert(temp, window);
temp = validParameter.validFile(parameters, "increment", false); if (temp == "not found") { temp = "25"; }
- convert(temp, increment);
+ m->mothurConvert(temp, increment);
+
+ temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; }
+ save = m->isTrue(temp);
+ rdb->save = save;
+ if (save) { //clear out old references
+ rdb->clearMemory();
+ }
+
+ //this has to go after save so that if the user sets save=t and provides no reference we abort
+ templatefile = validParameter.validFile(parameters, "reference", true);
+ if (templatefile == "not found") {
+ //check for saved reference sequences
+ if (rdb->referenceSeqs.size() != 0) {
+ templatefile = "saved";
+ }else {
+ m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
+ m->mothurOutEndLine();
+ abort = true;
+ }
+ }else if (templatefile == "not open") { abort = true; }
+ else { if (save) { rdb->setSavedReference(templatefile); } }
+
maskfile = validParameter.validFile(parameters, "mask", false);
if (maskfile == "not found") { maskfile = ""; }
//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"){ outputDir = ""; }
- templatefile = validParameter.validFile(parameters, "reference", true);
- if (templatefile == "not open") { abort = true; }
- else if (templatefile == "not found") { templatefile = ""; m->mothurOut("reference is a required parameter for the chimera.pintail command."); m->mothurOutEndLine(); abort = true; }
-
consfile = validParameter.validFile(parameters, "conservation", true);
if (consfile == "not open") { abort = true; }
else if (consfile == "not found") {
if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine(); }
//check for quantile to save the time
+ string baseName = templatefile;
+ if (templatefile == "saved") { baseName = rdb->getSavedReference(); }
+
string tempQuan = "";
if ((!filter) && (maskfile == "")) {
- tempQuan = inputDir + m->getRootName(m->getSimpleName(templatefile)) + "pintail.quan";
+ tempQuan = inputDir + m->getRootName(m->getSimpleName(baseName)) + "pintail.quan";
}else if ((!filter) && (maskfile != "")) {
- tempQuan = inputDir + m->getRootName(m->getSimpleName(templatefile)) + "pintail.masked.quan";
+ tempQuan = inputDir + m->getRootName(m->getSimpleName(baseName)) + "pintail.masked.quan";
}else if ((filter) && (maskfile != "")) {
- tempQuan = inputDir + m->getRootName(m->getSimpleName(templatefile)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "masked.quan";
+ tempQuan = inputDir + m->getRootName(m->getSimpleName(baseName)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "masked.quan";
}else if ((filter) && (maskfile == "")) {
- tempQuan = inputDir + m->getRootName(m->getSimpleName(templatefile)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "quan";
+ tempQuan = inputDir + m->getRootName(m->getSimpleName(baseName)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "quan";
}
ifstream FileTest(tempQuan.c_str());
string tryPath = m->getDefaultPath();
string tempQuan = "";
if ((!filter) && (maskfile == "")) {
- tempQuan = tryPath + m->getRootName(m->getSimpleName(templatefile)) + "pintail.quan";
+ tempQuan = tryPath + m->getRootName(m->getSimpleName(baseName)) + "pintail.quan";
}else if ((!filter) && (maskfile != "")) {
- tempQuan = tryPath + m->getRootName(m->getSimpleName(templatefile)) + "pintail.masked.quan";
+ tempQuan = tryPath + m->getRootName(m->getSimpleName(baseName)) + "pintail.masked.quan";
}else if ((filter) && (maskfile != "")) {
- tempQuan = tryPath + m->getRootName(m->getSimpleName(templatefile)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "masked.quan";
+ tempQuan = tryPath + m->getRootName(m->getSimpleName(baseName)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "masked.quan";
}else if ((filter) && (maskfile == "")) {
- tempQuan = tryPath + m->getRootName(m->getSimpleName(templatefile)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "quan";
+ tempQuan = tryPath + m->getRootName(m->getSimpleName(baseName)) + "pintail.filtered." + m->getSimpleName(m->getRootName(fastaFileNames[s])) + "quan";
}
ifstream FileTest2(tempQuan.c_str());
}
}
}
-
chimera = new Pintail(fastaFileNames[s], templatefile, filter, processors, maskfile, consfile, quanfile, window, increment, outputDir);
if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]); }//if user entered a file with a path then preserve it
string outputFileName, accnosFileName;
- if (maskfile != "") {
- outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + m->getSimpleName(m->getRootName(maskfile)) + ".pintail.chimeras";
- accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + m->getSimpleName(m->getRootName(maskfile)) + ".pintail.accnos";
- }else {
- outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "pintail.chimeras";
- accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "pintail.accnos";
- }
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]));
+ if (maskfile != "") { variables["[tag]"] = m->getSimpleName(m->getRootName(maskfile)); }
+ outputFileName = getOutputFileName("chimera", variables);
+ accnosFileName = getOutputFileName("accnos", variables);
+
- if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } return 0; }
+ if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } return 0; }
if (chimera->getUnaligned()) {
m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
#ifdef USE_MPI
int pid, numSeqsPerProcessor;
int tag = 2001;
- vector<unsigned long int> MPIPos;
+ vector<unsigned long long> MPIPos;
MPI_Status status;
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
- if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } delete chimera; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } delete chimera; return 0; }
if (pid == 0) { //you are the root process
//do your part
driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
- if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } delete chimera; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); m->mothurRemove(outputFileName); m->mothurRemove(accnosFileName); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } delete chimera; return 0; }
}else{ //you are a child process
MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
//do your part
driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
- if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } delete chimera; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } delete chimera; return 0; }
}
//close files
MPI_File_close(&outMPIAccnos);
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
- vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
-
- for (int i = 0; i < (positions.size()-1); i++) {
- lines.push_back(new linePair(positions[i], positions[(i+1)]));
- }
-
+
//break up file
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ vector<unsigned long long> positions = m->divideFile(fastaFileNames[s], processors);
+
+ for (int i = 0; i < (positions.size()-1); i++) {
+ lines.push_back(new linePair(positions[i], positions[(i+1)]));
+ }
+
if(processors == 1){
numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
- if (m->control_pressed) { outputTypes.clear(); remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFileName); m->mothurRemove(accnosFileName); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
}else{
processIDS.resize(0);
//append output files
for(int i=1;i<processors;i++){
m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
- remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
+ m->mothurRemove((outputFileName + toString(processIDS[i]) + ".temp"));
}
//append output files
for(int i=1;i<processors;i++){
m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
- remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
+ m->mothurRemove((accnosFileName + toString(processIDS[i]) + ".temp"));
}
if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(accnosFileName.c_str());
- for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } outputTypes.clear();
+ m->mothurRemove(outputFileName);
+ m->mothurRemove(accnosFileName);
+ for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } outputTypes.clear();
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
delete chimera;
return 0;
}
#else
+ lines.push_back(new linePair(0, 1000));
numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
- if (m->control_pressed) { outputTypes.clear(); remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFileName); m->mothurRemove(accnosFileName); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); delete chimera; return 0; }
#endif
#endif
}
delete candidateSeq;
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- unsigned long int pos = inFASTA.tellg();
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ unsigned long long pos = inFASTA.tellg();
if ((pos == -1) || (pos >= filePos->end)) { break; }
#else
if (inFASTA.eof()) { break; }
#endif
//report progress
- if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine(); }
+ if((count) % 100 == 0){ m->mothurOutJustToScreen("Processing sequence: " + toString(count) + "\n"); }
}
//report progress
- if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine(); }
+ if((count) % 100 != 0){ m->mothurOutJustToScreen("Processing sequence: " + toString(count) + "\n"); }
out.close();
out2.close();
}
//**********************************************************************************************************************
#ifdef USE_MPI
-int ChimeraPintailCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long int>& MPIPos){
+int ChimeraPintailCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, vector<unsigned long long>& MPIPos){
try {
MPI_Status status;
delete candidateSeq;
//report progress
- if((i+1) % 100 == 0){ cout << "Processing sequence: " << (i+1) << endl; m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n"); }
+ if((i+1) % 100 == 0){ cout << "Processing sequence: " << (i+1) << endl; }
}
//report progress
- if(num % 100 != 0){ cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n"); }
+ if(num % 100 != 0){ cout << "Processing sequence: " << num << endl; }
return 0;
int ChimeraPintailCommand::createProcesses(string outputFileName, string filename, string accnos) {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
int process = 0;
int num = 0;
string tempFile = outputFileName + toString(processIDS[i]) + ".num.temp";
m->openInputFile(tempFile, in);
if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
- in.close(); remove(tempFile.c_str());
+ in.close(); m->mothurRemove(tempFile);
}
return num;
projects / mothur.git / blobdiff