}
//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);
+ inputDir = validParameter.validFile(parameters, "inputdir", false);
if (inputDir == "not found"){ inputDir = ""; }
else {
string path;
- it = parameters.find("fasta");
- //user has given a template file
- if(it != parameters.end()){
- path = hasPath(it->second);
- //if the user has not given a path then, add inputdir. else leave path alone.
- if (path == "") { parameters["fasta"] = inputDir + it->second; }
- }
-
it = parameters.find("template");
//user has given a template file
if(it != parameters.end()){
//check for required parameters
- fastafile = validParameter.validFile(parameters, "fasta", true);
- if (fastafile == "not open") { abort = true; }
- else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.pintail command."); m->mothurOutEndLine(); 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"){
- outputDir = "";
- outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
- }
+ fastafile = validParameter.validFile(parameters, "fasta", false);
+ if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.pintail command."); m->mothurOutEndLine(); abort = true; }
+ else {
+ splitAtDash(fastafile, fastaFileNames);
+
+ //go through files and make sure they are good, if not, then disregard them
+ for (int i = 0; i < fastaFileNames.size(); i++) {
+ if (inputDir != "") {
+ string path = hasPath(fastaFileNames[i]);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { fastaFileNames[i] = inputDir + fastaFileNames[i]; }
+ }
+
+ int ableToOpen;
+ ifstream in;
+
+ #ifdef USE_MPI
+ int pid;
+ MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+
+ if (pid == 0) {
+ #endif
- templatefile = validParameter.validFile(parameters, "template", true);
- if (templatefile == "not open") { abort = true; }
- else if (templatefile == "not found") { templatefile = ""; m->mothurOut("template is a required parameter for the chimera.pintail command."); m->mothurOutEndLine(); abort = true; }
+ ableToOpen = openInputFile(fastaFileNames[i], in);
+ in.close();
+
+ #ifdef USE_MPI
+ for (int j = 1; j < processors; j++) {
+ MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD);
+ }
+ }else{
+ MPI_Status status;
+ MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
+ }
+
+ #endif
+
+ if (ableToOpen == 1) {
+ m->mothurOut(fastaFileNames[i] + " will be disregarded."); m->mothurOutEndLine();
+ //erase from file list
+ fastaFileNames.erase(fastaFileNames.begin()+i);
+ i--;
+ }
+ }
+
+ //make sure there is at least one valid file left
+ if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
+ }
- consfile = validParameter.validFile(parameters, "conservation", true);
- if (consfile == "not open") { abort = true; }
- else if (consfile == "not found") { consfile = ""; }
+ string temp;
+ temp = validParameter.validFile(parameters, "filter", false); if (temp == "not found") { temp = "F"; }
+ filter = isTrue(temp);
- quanfile = validParameter.validFile(parameters, "quantile", true);
- if (quanfile == "not open") { abort = true; }
- else if (quanfile == "not found") { quanfile = ""; }
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found") { temp = "1"; }
+ convert(temp, processors);
+
+ temp = validParameter.validFile(parameters, "window", false); if (temp == "not found") { temp = "0"; }
+ convert(temp, window);
+
+ temp = validParameter.validFile(parameters, "increment", false); if (temp == "not found") { temp = "25"; }
+ convert(temp, increment);
maskfile = validParameter.validFile(parameters, "mask", false);
if (maskfile == "not found") { maskfile = ""; }
if (ableToOpen == 1) { abort = true; }
in.close();
}
-
- string temp;
- temp = validParameter.validFile(parameters, "filter", false); if (temp == "not found") { temp = "F"; }
- filter = isTrue(temp);
+
- temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found") { temp = "1"; }
- convert(temp, processors);
+ //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 = "";
+ outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
+ }
+
+ templatefile = validParameter.validFile(parameters, "template", true);
+ if (templatefile == "not open") { abort = true; }
+ else if (templatefile == "not found") { templatefile = ""; m->mothurOut("template is a required parameter for the chimera.pintail command."); m->mothurOutEndLine(); abort = true; }
- temp = validParameter.validFile(parameters, "window", false); if (temp == "not found") { temp = "0"; }
- convert(temp, window);
+ consfile = validParameter.validFile(parameters, "conservation", true);
+ if (consfile == "not open") { abort = true; }
+ else if (consfile == "not found") {
+ consfile = "";
+ //check for consfile
+ string tempConsFile = getRootName(inputDir + getSimpleName(templatefile)) + "freq";
+ ifstream FileTest(tempConsFile.c_str());
+ if(FileTest){ m->mothurOut("I found " + tempConsFile + " in your input file directory. I will use it to save time."); m->mothurOutEndLine(); consfile = tempConsFile; FileTest.close(); }
+ }
- temp = validParameter.validFile(parameters, "increment", false); if (temp == "not found") { temp = "25"; }
- convert(temp, increment);
+ quanfile = validParameter.validFile(parameters, "quantile", true);
+ if (quanfile == "not open") { abort = true; }
+ else if (quanfile == "not found") { quanfile = ""; }
}
}
catch(exception& e) {
m->mothurOut("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");
m->mothurOut("The chimera.pintail command parameters are fasta, template, filter, mask, processors, window, increment, conservation and quantile.\n");
m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
+ m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
m->mothurOut("The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n");
m->mothurOut("The mask parameter allows you to specify a file containing one sequence you wish to use as a mask for the your sequences, by default no mask is applied. You can apply an ecoli mask by typing, mask=default. \n");
#ifdef USE_MPI
m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
#endif
- m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=1/4 sequence length. \n");
+ m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=300. \n");
m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=25.\n");
m->mothurOut("The conservation parameter allows you to enter a frequency file containing the highest bases frequency at each place in the alignment.\n");
m->mothurOut("The quantile parameter allows you to enter a file containing quantiles for a template files sequences, if you use the filter the quantile file generated becomes unique to the fasta file you used.\n");
m->mothurOut("The chimera.pintail command should be in the following format: \n");
- m->mothurOut("chimera.seqs(fasta=yourFastaFile, filter=yourFilter, correction=yourCorrection, processors=yourProcessors, method=bellerophon) \n");
- m->mothurOut("Example: chimera.seqs(fasta=AD.align, filter=True, correction=true, method=bellerophon, window=200) \n");
+ m->mothurOut("chimera.pintail(fasta=yourFastaFile, template=yourTemplate) \n");
+ m->mothurOut("Example: chimera.pintail(fasta=AD.align, template=silva.bacteria.fasta) \n");
m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
}
catch(exception& e) {
if (abort == true) { return 0; }
- int start = time(NULL);
-
- chimera = new Pintail(fastafile, templatefile, filter, processors, maskfile, consfile, quanfile, window, increment, outputDir);
-
- //set user options
- if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine(); }
-
-
- string outputFileName, accnosFileName;
- if (maskfile != "") {
- outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + maskfile + ".pintail.chimeras";
- accnosFileName = outputDir + getRootName(getSimpleName(fastafile)) + maskfile + ".pintail.accnos";
- }else {
- outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pintail.chimeras";
- accnosFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pintail.accnos";
- }
- bool hasAccnos = true;
-
- if (m->control_pressed) { delete chimera; return 0; }
-
- if (chimera->getUnaligned()) {
- m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
- delete chimera;
- return 0;
- }
- templateSeqsLength = chimera->getLength();
-
- #ifdef USE_MPI
- int pid, end, numSeqsPerProcessor;
- int tag = 2001;
- vector<long> MPIPos;
- MPIWroteAccnos = false;
-
- MPI_Status status;
- MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
- MPI_Comm_size(MPI_COMM_WORLD, &processors);
-
- MPI_File inMPI;
- MPI_File outMPI;
- MPI_File outMPIAccnos;
-
- int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
- int inMode=MPI_MODE_RDONLY;
-
- //char* outFilename = new char[outputFileName.length()];
- //memcpy(outFilename, outputFileName.c_str(), outputFileName.length());
-
- char outFilename[1024];
- strcpy(outFilename, outputFileName.c_str());
-
- //char* outAccnosFilename = new char[accnosFileName.length()];
- //memcpy(outAccnosFilename, accnosFileName.c_str(), accnosFileName.length());
-
- char outAccnosFilename[1024];
- strcpy(outAccnosFilename, accnosFileName.c_str());
+ for (int s = 0; s < fastaFileNames.size(); s++) {
+
+ m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
- //char* inFileName = new char[fastafile.length()];
- //memcpy(inFileName, fastafile.c_str(), fastafile.length());
+ int start = time(NULL);
+
+ //set user options
+ 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 tempQuan = "";
+ if ((!filter) && (maskfile == "")) {
+ tempQuan = inputDir + getRootName(getSimpleName(templatefile)) + "pintail.quan";
+ }else if ((!filter) && (maskfile != "")) {
+ tempQuan = inputDir + getRootName(getSimpleName(templatefile)) + "pintail.masked.quan";
+ }else if ((filter) && (maskfile != "")) {
+ tempQuan = inputDir + getRootName(getSimpleName(templatefile)) + "pintail.filtered." + getSimpleName(getRootName(fastaFileNames[s])) + "masked.quan";
+ }else if ((filter) && (maskfile == "")) {
+ tempQuan = inputDir + getRootName(getSimpleName(templatefile)) + "pintail.filtered." + getSimpleName(getRootName(fastaFileNames[s])) + "quan";
+ }
- char inFileName[1024];
- strcpy(inFileName, fastafile.c_str());
-
- MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
- MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
- MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
+ ifstream FileTest(tempQuan.c_str());
+ if(FileTest){ m->mothurOut("I found " + tempQuan + " in your input file directory. I will use it to save time."); m->mothurOutEndLine(); quanfile = tempQuan; FileTest.close(); }
- //delete inFileName;
- //delete outFilename;
- //delete outAccnosFilename;
-
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); delete chimera; return 0; }
-
- if (pid == 0) { //you are the root process
-
- MPIPos = setFilePosFasta(fastafile, numSeqs); //fills MPIPos, returns numSeqs
-
- //send file positions to all processes
- MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
- MPI_Bcast(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
-
- //figure out how many sequences you have to align
- numSeqsPerProcessor = numSeqs / processors;
- if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
- int startIndex = pid * numSeqsPerProcessor;
+ chimera = new Pintail(fastaFileNames[s], templatefile, filter, processors, maskfile, consfile, quanfile, window, increment, outputDir);
- //align your part
- driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
-
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); remove(outputFileName.c_str()); remove(accnosFileName.c_str()); delete chimera; return 0; }
-
- for (int i = 1; i < processors; i++) {
- bool tempResult;
- MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
- if (tempResult != 0) { MPIWroteAccnos = true; }
- }
- }else{ //you are a child process
- MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
- MPIPos.resize(numSeqs+1);
- MPI_Bcast(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
-
- //figure out how many sequences you have to align
- numSeqsPerProcessor = numSeqs / processors;
- if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
- int startIndex = pid * numSeqsPerProcessor;
-
- //align your part
- driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
-
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); MPI_File_close(&outMPIAccnos); delete chimera; return 0; }
-
- MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
+ string outputFileName, accnosFileName;
+ if (maskfile != "") {
+ outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + maskfile + ".pintail.chimeras";
+ accnosFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + maskfile + ".pintail.accnos";
+ }else {
+ outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "pintail.chimeras";
+ accnosFileName = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "pintail.accnos";
}
+ bool hasAccnos = true;
- //close files
- MPI_File_close(&inMPI);
- MPI_File_close(&outMPI);
- MPI_File_close(&outMPIAccnos);
+ if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) { remove(outputNames[j].c_str()); } return 0; }
- //delete accnos file if blank
- if (pid == 0) {
- if (!MPIWroteAccnos) {
- //MPI_Info info;
- //MPI_File_delete(outAccnosFilename, info);
- hasAccnos = false;
- remove(accnosFileName.c_str());
- }
+ if (chimera->getUnaligned()) {
+ m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
+ delete chimera;
+ return 0;
}
+ templateSeqsLength = chimera->getLength();
+
+ #ifdef USE_MPI
+ int pid, end, numSeqsPerProcessor;
+ int tag = 2001;
+ vector<long> MPIPos;
+ MPIWroteAccnos = false;
+
+ MPI_Status status;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
- #else
-
- //break up file
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- if(processors == 1){
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
- numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
- inFASTA.close();
+ MPI_File inMPI;
+ MPI_File outMPI;
+ MPI_File outMPIAccnos;
- lines.push_back(new linePair(0, numSeqs));
+ int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+ int inMode=MPI_MODE_RDONLY;
- driver(lines[0], outputFileName, fastafile, accnosFileName);
+ char outFilename[1024];
+ strcpy(outFilename, outputFileName.c_str());
- if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(accnosFileName.c_str());
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- delete chimera;
- return 0;
- }
+ char outAccnosFilename[1024];
+ strcpy(outAccnosFilename, accnosFileName.c_str());
- //delete accnos file if its blank
- if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
-
- }else{
- vector<int> positions;
- processIDS.resize(0);
+ char inFileName[1024];
+ strcpy(inFileName, fastaFileNames[s].c_str());
+
+ MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
+ MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
+ MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
+ if (m->control_pressed) { 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 (pid == 0) { //you are the root process
+
+ MPIPos = setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
+
+ //send file positions to all processes
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
+ }
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = numSeqs / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
- string input;
- while(!inFASTA.eof()){
- input = getline(inFASTA);
- if (input.length() != 0) {
- if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
+ //do your part
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
+
+ if (m->control_pressed) { 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; }
+
+ for (int i = 1; i < processors; i++) {
+ bool tempResult;
+ MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
+ if (tempResult != 0) { MPIWroteAccnos = true; }
}
+ }else{ //you are a child process
+ MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ MPIPos.resize(numSeqs+1);
+ MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = numSeqs / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor; }
+
+ //do your part
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, MPIPos);
+
+ if (m->control_pressed) { 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; }
+
+ MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
}
- inFASTA.close();
-
- numSeqs = positions.size();
- int numSeqsPerProcessor = numSeqs / processors;
+ //close files
+ MPI_File_close(&inMPI);
+ MPI_File_close(&outMPI);
+ MPI_File_close(&outMPIAccnos);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
- for (int i = 0; i < processors; i++) {
- long int startPos = positions[ i * numSeqsPerProcessor ];
- if(i == processors - 1){
- numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
+ //delete accnos file if blank
+ if (pid == 0) {
+ if (!MPIWroteAccnos) {
+ hasAccnos = false;
+ remove(accnosFileName.c_str());
}
- lines.push_back(new linePair(startPos, numSeqsPerProcessor));
}
-
-
- createProcesses(outputFileName, fastafile, accnosFileName);
-
- rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
+
+ #else
+
+ //break up file
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ if(processors == 1){
+ ifstream inFASTA;
+ openInputFile(fastaFileNames[s], inFASTA);
+ getNumSeqs(inFASTA, numSeqs);
+ inFASTA.close();
- //append output files
- for(int i=1;i<processors;i++){
- appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
- remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
- }
-
- vector<string> nonBlankAccnosFiles;
- //delete blank accnos files generated with multiple processes
- for(int i=0;i<processors;i++){
- if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
- nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
- }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str()); }
- }
+ lines.push_back(new linePair(0, numSeqs));
+
+ driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
+
+ 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()); }
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
+ delete chimera;
+ return 0;
+ }
+
+ //delete accnos file if its blank
+ if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
+
+ }else{
+ vector<int> positions;
+ processIDS.resize(0);
+
+ ifstream inFASTA;
+ openInputFile(fastaFileNames[s], inFASTA);
+
+ string input;
+ while(!inFASTA.eof()){
+ input = getline(inFASTA);
+ if (input.length() != 0) {
+ if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
+ }
+ }
+ inFASTA.close();
+
+ numSeqs = positions.size();
+
+ int numSeqsPerProcessor = numSeqs / processors;
+
+ for (int i = 0; i < processors; i++) {
+ long int startPos = positions[ i * numSeqsPerProcessor ];
+ if(i == processors - 1){
+ numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
+ }
+ lines.push_back(new linePair(startPos, numSeqsPerProcessor));
+ }
+
+ createProcesses(outputFileName, fastaFileNames[s], accnosFileName);
- //append accnos files
- if (nonBlankAccnosFiles.size() != 0) {
- rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());
+ rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
+
+ //append output files
+ for(int i=1;i<processors;i++){
+ appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
+ remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
+ }
+
+ vector<string> nonBlankAccnosFiles;
+ //delete blank accnos files generated with multiple processes
+ for(int i=0;i<processors;i++){
+ if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
+ nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
+ }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str()); }
+ }
+
+ //append accnos files
+ if (nonBlankAccnosFiles.size() != 0) {
+ rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());
+
+ for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
+ appendFiles(nonBlankAccnosFiles[h], accnosFileName);
+ remove(nonBlankAccnosFiles[h].c_str());
+ }
+ }else{ hasAccnos = false; }
- for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
- appendFiles(nonBlankAccnosFiles[h], accnosFileName);
- remove(nonBlankAccnosFiles[h].c_str());
+ 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()); }
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
+ delete chimera;
+ return 0;
}
- }else{ hasAccnos = false; }
+ }
+
+ #else
+ ifstream inFASTA;
+ openInputFile(fastaFileNames[s], inFASTA);
+ getNumSeqs(inFASTA, numSeqs);
+ inFASTA.close();
+ lines.push_back(new linePair(0, numSeqs));
+
+ driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName);
if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(accnosFileName.c_str());
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- delete chimera;
- return 0;
+ 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;
}
- }
-
- #else
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
- numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
- inFASTA.close();
- lines.push_back(new linePair(0, numSeqs));
+
+ //delete accnos file if its blank
+ if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
+ #endif
- driver(lines[0], outputFileName, fastafile, accnosFileName);
+ #endif
+
+ delete chimera;
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- if (m->control_pressed) {
- remove(outputFileName.c_str());
- remove(accnosFileName.c_str());
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- delete chimera;
- return 0;
- }
+ outputNames.push_back(outputFileName);
+ if (hasAccnos) { outputNames.push_back(accnosFileName); }
- //delete accnos file if its blank
- if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
- #endif
-
- #endif
-
- delete chimera;
- for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
+ m->mothurOutEndLine();
+ m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
+ }
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
- m->mothurOut(outputFileName); m->mothurOutEndLine();
- if (hasAccnos) { m->mothurOut(accnosFileName); m->mothurOutEndLine(); }
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
m->mothurOutEndLine();
- m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
-
+
return 0;
}
projects / mothur.git / blobdiff