-/*
- * aligncommand.cpp
- * Mothur
- *
- * Created by Sarah Westcott on 5/15/09.
- * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
- *
- * This version of nast does everything I think that the greengenes nast server does and then some. I have added the
- * feature of allowing users to define their database, kmer size for searching, alignment penalty values and alignment
- * method. This latter feature is perhaps most significant. nastPlus enables a user to use either a Needleman-Wunsch
- * (non-affine gap penalty) or Gotoh (affine gap penalty) pairwise alignment algorithm. This is significant because it
- * allows for a global alignment and not the local alignment provided by bLAst. Furthermore, it has the potential to
- * provide a better alignment because of the banding method employed by blast (I'm not sure about this).
- *
- * to compile type:
- * make
- *
- * for basic instructions on how to run nastPlus type:
- * ./nastPlus
- */
-
-#include "aligncommand.h"
-#include "sequence.hpp"
-
-#include "alignment.hpp"
-#include "gotohoverlap.hpp"
-#include "needlemanoverlap.hpp"
-#include "blastalign.hpp"
-#include "noalign.hpp"
-
-#include "database.hpp"
-#include "kmerdb.hpp"
-#include "suffixdb.hpp"
-#include "blastdb.hpp"
-
-#include "nast.hpp"
-#include "nastreport.hpp"
-
-
-//**********************************************************************************************************************
-AlignCommand::AlignCommand(){
- try {
- globaldata = GlobalData::getInstance();
- candidateFileName = globaldata->inputFileName;
- templateFileName = globaldata->getTemplateFile();
- openInputFile(candidateFileName, in);
- convert(globaldata->getKSize(), kmerSize);
- convert(globaldata->getMatch(), match);
- convert(globaldata->getMismatch(), misMatch);
- convert(globaldata->getGapopen(), gapOpen);
- convert(globaldata->getGapextend(), gapExtend);
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the AlignCommand class Function AlignCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the AlignCommand class function AlignCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
-AlignCommand::~AlignCommand(){
-
-}
-
-//**********************************************************************************************************************
-
-int AlignCommand::execute(){
- try {
- srand( (unsigned)time( NULL ) ); //needed to assign names to temporary files
-
- Database* templateDB;
- if(globaldata->getSearch() == "kmer") { templateDB = new KmerDB(templateFileName, kmerSize); }
- else if(globaldata->getSearch() == "suffix") { templateDB = new SuffixDB(templateFileName); }
- else if(globaldata->getSearch() == "blast") { templateDB = new BlastDB(templateFileName, gapOpen, gapExtend, match, misMatch); }
- else { cout << globaldata->getSearch() << " is not a valid search option. I will run the command using suffix." << endl;
- templateDB = new SuffixDB(templateFileName); }
-
- Alignment* alignment;
- if(globaldata->getAlign() == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, 3000); }
- else if(globaldata->getAlign() == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, 3000); }
- else if(globaldata->getAlign() == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }
- else if(globaldata->getAlign() == "noalign") { alignment = new NoAlign(); }
- else { cout << globaldata->getAlign() << " is not a valid alignment option. I will run the command using blast." << endl;
- alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }
-
- int numFastaSeqs=count(istreambuf_iterator<char>(in),istreambuf_iterator<char>(), '>');
- in.seekg(0);
-
- string candidateAligngmentFName = candidateFileName.substr(0,candidateFileName.find_last_of(".")+1) + globaldata->getSearch() + '.' + globaldata->getAlign() + ".nast.align";
- ofstream candidateAlignmentFile;
- openOutputFile(candidateAligngmentFName, candidateAlignmentFile);
-
- string candidateReportFName = candidateFileName.substr(0,candidateFileName.find_last_of(".")+1) + globaldata->getSearch() + '.' + globaldata->getAlign() + ".nast.report";
- NastReport report(candidateReportFName);
-
- cout << "We are going to align the " << numFastaSeqs << " sequences in " << candidateFileName << "..." << endl;
- cout.flush();
-
- int start = time(NULL);
-
- for(int i=0;i<numFastaSeqs;i++){
-
- Sequence* candidateSeq = new Sequence(in);
- report.setCandidate(candidateSeq);
-
- Sequence* templateSeq = templateDB->findClosestSequence(candidateSeq);
- report.setTemplate(templateSeq);
- report.setSearchParameters(globaldata->getSearch(), templateDB->getSearchScore());
-
- Nast nast(alignment, candidateSeq, templateSeq);
- report.setAlignmentParameters(globaldata->getAlign(), alignment);
- report.setNastParameters(nast);
-
- candidateAlignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
- candidateAlignmentFile.flush();
-
- if((i+1) % 100 == 0){
- cout << "It has taken " << time(NULL) - start << " secs to align " << i+1 << " sequences" << endl;
- }
- report.print();
-
- delete candidateSeq;
- }
-
- cout << "It took " << time(NULL) - start << " secs to align " << numFastaSeqs << " sequences" << endl;
- cout << endl;
-
- delete templateDB;
- delete alignment;
-
-
- return 0;
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the AlignCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the AlignCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
\ No newline at end of file
+/*\r
+ * aligncommand.cpp\r
+ * Mothur\r
+ *\r
+ * Created by Sarah Westcott on 5/15/09.\r
+ * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.\r
+ *\r
+ * This version of nast does everything I think that the greengenes nast server does and then some. I have added the \r
+ * feature of allowing users to define their database, kmer size for searching, alignment penalty values and alignment \r
+ * method. This latter feature is perhaps most significant. nastPlus enables a user to use either a Needleman-Wunsch \r
+ * (non-affine gap penalty) or Gotoh (affine gap penalty) pairwise alignment algorithm. This is significant because it\r
+ * allows for a global alignment and not the local alignment provided by bLAst. Furthermore, it has the potential to\r
+ * provide a better alignment because of the banding method employed by blast (I'm not sure about this).\r
+ *\r
+ */\r
+\r
+#include "aligncommand.h"\r
+#include "sequence.hpp"\r
+\r
+#include "gotohoverlap.hpp"\r
+#include "needlemanoverlap.hpp"\r
+#include "blastalign.hpp"\r
+#include "noalign.hpp"\r
+\r
+#include "nast.hpp"\r
+#include "nastreport.hpp"\r
+\r
+\r
+//**********************************************************************************************************************\r
+\r
+AlignCommand::AlignCommand(string option) {\r
+ try {\r
+ \r
+ abort = false;\r
+ \r
+ //allow user to run help\r
+ if(option == "help") { help(); abort = true; }\r
+ \r
+ else {\r
+ \r
+ //valid paramters for this command\r
+ string AlignArray[] = {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors","flip","threshold","outputdir","inputdir"};\r
+ vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));\r
+ \r
+ OptionParser parser(option);\r
+ map<string, string> parameters = parser.getParameters(); \r
+ \r
+ ValidParameters validParameter;\r
+ map<string, string>::iterator it;\r
+ \r
+ //check to make sure all parameters are valid for command\r
+ for (it = parameters.begin(); it != parameters.end(); it++) { \r
+ if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }\r
+ }\r
+\r
+ //if the user changes the output directory command factory will send this info to us in the output parameter \r
+ outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }\r
+ \r
+\r
+ //if the user changes the input directory command factory will send this info to us in the output parameter \r
+ string inputDir = validParameter.validFile(parameters, "inputdir", false); \r
+ \r
+ if (inputDir == "not found"){ inputDir = ""; }\r
+ else {\r
+ string path;\r
+\r
+ it = parameters.find("template");\r
+\r
+ //user has given a template file\r
+ if(it != parameters.end()){ \r
+ path = hasPath(it->second);\r
+ //if the user has not given a path then, add inputdir. else leave path alone.\r
+ if (path == "") { parameters["template"] = inputDir + it->second; }\r
+ }\r
+ }\r
+\r
+ //check for required parameters\r
+ templateFileName = validParameter.validFile(parameters, "template", true);\r
+ \r
+ if (templateFileName == "not found") { \r
+ m->mothurOut("template is a required parameter for the align.seqs command."); \r
+ m->mothurOutEndLine();\r
+ abort = true; \r
+ }else if (templateFileName == "not open") { abort = true; } \r
+ \r
+ candidateFileName = validParameter.validFile(parameters, "candidate", false);\r
+ if (candidateFileName == "not found") { m->mothurOut("candidate is a required parameter for the align.seqs command."); m->mothurOutEndLine(); abort = true; }\r
+ else { \r
+ splitAtDash(candidateFileName, candidateFileNames);\r
+ \r
+ //go through files and make sure they are good, if not, then disregard them\r
+ for (int i = 0; i < candidateFileNames.size(); i++) {\r
+ if (inputDir != "") {\r
+ string path = hasPath(candidateFileNames[i]);\r
+ //if the user has not given a path then, add inputdir. else leave path alone.\r
+ if (path == "") { candidateFileNames[i] = inputDir + candidateFileNames[i]; }\r
+ }\r
+ \r
+ int ableToOpen;\r
+ ifstream in;\r
+ \r
+ #ifdef USE_MPI \r
+ int pid;\r
+ MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running\r
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are\r
+ \r
+ if (pid == 0) {\r
+ #endif\r
+\r
+ ableToOpen = openInputFile(candidateFileNames[i], in);\r
+ in.close();\r
+ \r
+ #ifdef USE_MPI \r
+ for (int j = 1; j < processors; j++) {\r
+ MPI_Send(&ableToOpen, 1, MPI_INT, j, 2001, MPI_COMM_WORLD); \r
+ }\r
+ }else{\r
+ MPI_Status status;\r
+ MPI_Recv(&ableToOpen, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);\r
+ }\r
+ \r
+ #endif\r
+\r
+ if (ableToOpen == 1) { \r
+ m->mothurOut(candidateFileNames[i] + " will be disregarded."); m->mothurOutEndLine(); \r
+ //erase from file list\r
+ candidateFileNames.erase(candidateFileNames.begin()+i);\r
+ i--;\r
+ }\r
+ \r
+ }\r
+ \r
+ //make sure there is at least one valid file left\r
+ if (candidateFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }\r
+ }\r
+ \r
+ //check for optional parameter and set defaults\r
+ // ...at some point should added some additional type checking...\r
+ string temp;\r
+ temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; }\r
+ convert(temp, kmerSize); \r
+ \r
+ temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; }\r
+ convert(temp, match); \r
+ \r
+ temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; }\r
+ convert(temp, misMatch); \r
+ \r
+ temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; }\r
+ convert(temp, gapOpen); \r
+ \r
+ temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; }\r
+ convert(temp, gapExtend); \r
+ \r
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }\r
+ convert(temp, processors); \r
+ \r
+ temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "f"; }\r
+ flip = isTrue(temp); \r
+ \r
+ temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found"){ temp = "0.50"; }\r
+ convert(temp, threshold); \r
+ \r
+ search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; }\r
+ \r
+ align = validParameter.validFile(parameters, "align", false); if (align == "not found"){ align = "needleman"; }\r
+ }\r
+ \r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "AlignCommand");\r
+ exit(1);\r
+ }\r
+}\r
+\r
+//**********************************************************************************************************************\r
+\r
+AlignCommand::~AlignCommand(){ \r
+\r
+ if (abort == false) {\r
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();\r
+ delete templateDB;\r
+ delete alignment;\r
+ }\r
+}\r
+\r
+//**********************************************************************************************************************\r
+\r
+void AlignCommand::help(){\r
+ try {\r
+ m->mothurOut("The align.seqs command reads a file containing sequences and creates an alignment file and a report file.\n");\r
+ m->mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen and gapextend.\n");\r
+ m->mothurOut("The template and candidate parameters are required. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");\r
+ m->mothurOut("The search parameter allows you to specify the method to find most similar template. Your options are: suffix, kmer and blast. The default is kmer.\n");\r
+ m->mothurOut("The align parameter allows you to specify the alignment method to use. Your options are: gotoh, needleman, blast and noalign. The default is needleman.\n");\r
+ m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n");\r
+ m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n");\r
+ m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n");\r
+ m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n");\r
+ m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n");\r
+ m->mothurOut("The flip parameter is used to specify whether or not you want mothur to try the reverse complement if a sequence falls below the threshold. The default is false.\n");\r
+ m->mothurOut("The threshold is used to specify a cutoff at which an alignment is deemed 'bad' and the reverse complement may be tried. The default threshold is 0.50, meaning 50% of the bases are removed in the alignment.\n");\r
+ m->mothurOut("If the flip parameter is set to true the reverse complement of the sequence is aligned and the better alignment is reported.\n");\r
+ m->mothurOut("The default for the threshold parameter is 0.50, meaning at least 50% of the bases must remain or the sequence is reported as potentially reversed.\n");\r
+ m->mothurOut("The align.seqs command should be in the following format: \n");\r
+ m->mothurOut("align.seqs(template=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) \n");\r
+ m->mothurOut("Example align.seqs(candidate=candidate.fasta, template=core.filtered, align=kmer, search=gotoh, ksize=8, match=2.0, mismatch=3.0, gapopen=-2.0, gapextend=-1.0)\n");\r
+ m->mothurOut("Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).\n\n");\r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "help");\r
+ exit(1);\r
+ }\r
+}\r
+\r
+\r
+//**********************************************************************************************************************\r
+\r
+int AlignCommand::execute(){\r
+ try {\r
+ if (abort == true) { return 0; }\r
+\r
+ templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch);\r
+ int longestBase = templateDB->getLongestBase();\r
+ \r
+ if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase); }\r
+ else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); }\r
+ else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }\r
+ else if(align == "noalign") { alignment = new NoAlign(); }\r
+ else {\r
+ m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");\r
+ m->mothurOutEndLine();\r
+ alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);\r
+ }\r
+ vector<string> outputNames;\r
+ \r
+ for (int s = 0; s < candidateFileNames.size(); s++) {\r
+ if (m->control_pressed) { return 0; }\r
+ \r
+ m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();\r
+ \r
+ if (outputDir == "") { outputDir += hasPath(candidateFileNames[s]); }\r
+ string alignFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align";\r
+ string reportFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align.report";\r
+ string accnosFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "flip.accnos";\r
+ bool hasAccnos = true;\r
+ \r
+ int numFastaSeqs = 0;\r
+ for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();\r
+ int start = time(NULL);\r
+ \r
+#ifdef USE_MPI \r
+ int pid, end, numSeqsPerProcessor; \r
+ int tag = 2001;\r
+ vector<long> MPIPos;\r
+ MPIWroteAccnos = false;\r
+ \r
+ MPI_Status status; \r
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are\r
+ MPI_Comm_size(MPI_COMM_WORLD, &processors); \r
+\r
+ MPI_File inMPI;\r
+ MPI_File outMPIAlign;\r
+ MPI_File outMPIReport;\r
+ MPI_File outMPIAccnos;\r
+ \r
+ int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY; \r
+ int inMode=MPI_MODE_RDONLY; \r
+ \r
+ //char* outAlignFilename = new char[alignFileName.length()];\r
+ //memcpy(outAlignFilename, alignFileName.c_str(), alignFileName.length());\r
+ \r
+ char outAlignFilename[1024];\r
+ strcpy(outAlignFilename, alignFileName.c_str());\r
+\r
+ //char* outReportFilename = new char[reportFileName.length()];\r
+ //memcpy(outReportFilename, reportFileName.c_str(), reportFileName.length());\r
+ \r
+ char outReportFilename[1024];\r
+ strcpy(outReportFilename, reportFileName.c_str());\r
+\r
+ //char* outAccnosFilename = new char[accnosFileName.length()];\r
+ //memcpy(outAccnosFilename, accnosFileName.c_str(), accnosFileName.length());\r
+ \r
+ char outAccnosFilename[1024];\r
+ strcpy(outAccnosFilename, accnosFileName.c_str());\r
+\r
+ //char* inFileName = new char[candidateFileNames[s].length()];\r
+ //memcpy(inFileName, candidateFileNames[s].c_str(), candidateFileNames[s].length());\r
+ \r
+ char inFileName[1024];\r
+ strcpy(inFileName, candidateFileNames[s].c_str());\r
+ \r
+ MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer\r
+ MPI_File_open(MPI_COMM_WORLD, outAlignFilename, outMode, MPI_INFO_NULL, &outMPIAlign);\r
+ MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport);\r
+ MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);\r
+ \r
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }\r
+ \r
+ if (pid == 0) { //you are the root process \r
+ \r
+ MPIPos = setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs\r
+ \r
+ //send file positions to all processes\r
+ for(int i = 1; i < processors; i++) { \r
+ MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);\r
+ MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);\r
+ }\r
+ \r
+ //figure out how many sequences you have to align\r
+ numSeqsPerProcessor = numFastaSeqs / processors;\r
+ int startIndex = pid * numSeqsPerProcessor;\r
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }\r
+ \r
+ \r
+ //align your part\r
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);\r
+ \r
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }\r
+\r
+ for (int i = 1; i < processors; i++) {\r
+ bool tempResult;\r
+ MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);\r
+ if (tempResult != 0) { MPIWroteAccnos = true; }\r
+ }\r
+ }else{ //you are a child process\r
+ MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);\r
+ MPIPos.resize(numFastaSeqs+1);\r
+ MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);\r
+\r
+ \r
+ //figure out how many sequences you have to align\r
+ numSeqsPerProcessor = numFastaSeqs / processors;\r
+ int startIndex = pid * numSeqsPerProcessor;\r
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }\r
+ \r
+ \r
+ //align your part\r
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);\r
+ \r
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }\r
+\r
+ MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD); \r
+ }\r
+ \r
+ //close files \r
+ MPI_File_close(&inMPI);\r
+ MPI_File_close(&outMPIAlign);\r
+ MPI_File_close(&outMPIReport);\r
+ MPI_File_close(&outMPIAccnos);\r
+ \r
+ //delete accnos file if blank\r
+ if (pid == 0) {\r
+ //delete accnos file if its blank else report to user\r
+ if (MPIWroteAccnos) { \r
+ m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");\r
+ if (!flip) {\r
+ m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); \r
+ }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }\r
+ m->mothurOutEndLine();\r
+ }else { \r
+ //MPI_Info info;\r
+ //MPI_File_delete(outAccnosFilename, info);\r
+ hasAccnos = false; \r
+ remove(accnosFileName.c_str()); \r
+ }\r
+ }\r
+ \r
+#else\r
+ \r
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)\r
+ if(processors == 1){\r
+ ifstream inFASTA;\r
+ openInputFile(candidateFileNames[s], inFASTA);\r
+ numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');\r
+ inFASTA.close();\r
+ \r
+ lines.push_back(new linePair(0, numFastaSeqs));\r
+ \r
+ driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);\r
+ \r
+ if (m->control_pressed) { \r
+ remove(accnosFileName.c_str()); \r
+ remove(alignFileName.c_str()); \r
+ remove(reportFileName.c_str()); \r
+ return 0; \r
+ }\r
+ \r
+ //delete accnos file if its blank else report to user\r
+ if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }\r
+ else { \r
+ m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");\r
+ if (!flip) {\r
+ m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); \r
+ }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }\r
+ m->mothurOutEndLine();\r
+ }\r
+ }\r
+ else{\r
+ vector<int> positions;\r
+ processIDS.resize(0);\r
+ \r
+ ifstream inFASTA;\r
+ openInputFile(candidateFileNames[s], inFASTA);\r
+ \r
+ string input;\r
+ while(!inFASTA.eof()){\r
+ input = getline(inFASTA);\r
+ if (input.length() != 0) {\r
+ if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }\r
+ }\r
+ }\r
+ inFASTA.close();\r
+ \r
+ numFastaSeqs = positions.size();\r
+ \r
+ int numSeqsPerProcessor = numFastaSeqs / processors;\r
+ \r
+ for (int i = 0; i < processors; i++) {\r
+ long int startPos = positions[ i * numSeqsPerProcessor ];\r
+ if(i == processors - 1){\r
+ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;\r
+ }\r
+ lines.push_back(new linePair(startPos, numSeqsPerProcessor));\r
+ }\r
+ \r
+ createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); \r
+ \r
+ rename((alignFileName + toString(processIDS[0]) + ".temp").c_str(), alignFileName.c_str());\r
+ rename((reportFileName + toString(processIDS[0]) + ".temp").c_str(), reportFileName.c_str());\r
+ \r
+ //append alignment and report files\r
+ for(int i=1;i<processors;i++){\r
+ appendAlignFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);\r
+ remove((alignFileName + toString(processIDS[i]) + ".temp").c_str());\r
+ \r
+ appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);\r
+ remove((reportFileName + toString(processIDS[i]) + ".temp").c_str());\r
+ }\r
+ \r
+ vector<string> nonBlankAccnosFiles;\r
+ //delete blank accnos files generated with multiple processes\r
+ for(int i=0;i<processors;i++){ \r
+ if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {\r
+ nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");\r
+ }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str()); }\r
+ }\r
+ \r
+ //append accnos files\r
+ if (nonBlankAccnosFiles.size() != 0) { \r
+ rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());\r
+ \r
+ for (int h=1; h < nonBlankAccnosFiles.size(); h++) {\r
+ appendAlignFiles(nonBlankAccnosFiles[h], accnosFileName);\r
+ remove(nonBlankAccnosFiles[h].c_str());\r
+ }\r
+ m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");\r
+ if (!flip) {\r
+ m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); \r
+ }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }\r
+ m->mothurOutEndLine();\r
+ }else{ hasAccnos = false; }\r
+ \r
+ if (m->control_pressed) { \r
+ remove(accnosFileName.c_str()); \r
+ remove(alignFileName.c_str()); \r
+ remove(reportFileName.c_str()); \r
+ return 0; \r
+ }\r
+ }\r
+ #else\r
+ ifstream inFASTA;\r
+ openInputFile(candidateFileNames[s], inFASTA);\r
+ numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');\r
+ inFASTA.close();\r
+ \r
+ lines.push_back(new linePair(0, numFastaSeqs));\r
+ \r
+ driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);\r
+ \r
+ if (m->control_pressed) { \r
+ remove(accnosFileName.c_str()); \r
+ remove(alignFileName.c_str()); \r
+ remove(reportFileName.c_str()); \r
+ return 0; \r
+ }\r
+ \r
+ //delete accnos file if its blank else report to user\r
+ if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }\r
+ else { \r
+ m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");\r
+ if (!flip) {\r
+ m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); \r
+ }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }\r
+ m->mothurOutEndLine();\r
+ }\r
+ \r
+ #endif\r
+\r
+#endif \r
+\r
+\r
+ #ifdef USE_MPI\r
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); \r
+ \r
+ if (pid == 0) { //only one process should output to screen\r
+ #endif\r
+\r
+ outputNames.push_back(alignFileName);\r
+ outputNames.push_back(reportFileName);\r
+ if (hasAccnos) { outputNames.push_back(accnosFileName); }\r
+ \r
+ #ifdef USE_MPI\r
+ }\r
+ #endif\r
+\r
+ m->mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences.");\r
+ m->mothurOutEndLine();\r
+ m->mothurOutEndLine();\r
+ }\r
+ \r
+ \r
+ m->mothurOutEndLine();\r
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();\r
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }\r
+ m->mothurOutEndLine();\r
+\r
+ return 0;\r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "execute");\r
+ exit(1);\r
+ }\r
+}\r
+\r
+//**********************************************************************************************************************\r
+\r
+int AlignCommand::driver(linePair* line, string alignFName, string reportFName, string accnosFName, string filename){\r
+ try {\r
+ ofstream alignmentFile;\r
+ openOutputFile(alignFName, alignmentFile);\r
+ \r
+ ofstream accnosFile;\r
+ openOutputFile(accnosFName, accnosFile);\r
+ \r
+ NastReport report(reportFName);\r
+ \r
+ ifstream inFASTA;\r
+ openInputFile(filename, inFASTA);\r
+\r
+ inFASTA.seekg(line->start);\r
+ \r
+ for(int i=0;i<line->numSeqs;i++){\r
+ \r
+ if (m->control_pressed) { return 0; }\r
+ \r
+ Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA);\r
+ \r
+ int origNumBases = candidateSeq->getNumBases();\r
+ string originalUnaligned = candidateSeq->getUnaligned();\r
+ int numBasesNeeded = origNumBases * threshold;\r
+ \r
+ if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file\r
+ if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {\r
+ alignment->resize(candidateSeq->getUnaligned().length()+1);\r
+ }\r
+ \r
+ Sequence temp = templateDB->findClosestSequence(candidateSeq);\r
+ Sequence* templateSeq = &temp;\r
+ \r
+ float searchScore = templateDB->getSearchScore();\r
+ \r
+ Nast* nast = new Nast(alignment, candidateSeq, templateSeq);\r
+ Sequence* copy;\r
+ \r
+ Nast* nast2;\r
+ bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below\r
+ //since nast does not make a copy of hte sequence passed, and it is used by the reporter below\r
+ //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place\r
+ //so this bool tells you if you need to delete it\r
+ \r
+ //if there is a possibility that this sequence should be reversed\r
+ if (candidateSeq->getNumBases() < numBasesNeeded) {\r
+ \r
+ string wasBetter = "";\r
+ //if the user wants you to try the reverse\r
+ if (flip) {\r
+ //get reverse compliment\r
+ copy = new Sequence(candidateSeq->getName(), originalUnaligned);\r
+ copy->reverseComplement();\r
+ \r
+ //rerun alignment\r
+ Sequence temp2 = templateDB->findClosestSequence(copy);\r
+ Sequence* templateSeq2 = &temp2;\r
+ \r
+ searchScore = templateDB->getSearchScore();\r
+ \r
+ nast2 = new Nast(alignment, copy, templateSeq2);\r
+ \r
+ //check if any better\r
+ if (copy->getNumBases() > candidateSeq->getNumBases()) {\r
+ candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better\r
+ templateSeq = templateSeq2; \r
+ delete nast;\r
+ nast = nast2;\r
+ needToDeleteCopy = true;\r
+ }else{ \r
+ wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";\r
+ delete nast2;\r
+ delete copy; \r
+ }\r
+ }\r
+ \r
+ //create accnos file with names\r
+ accnosFile << candidateSeq->getName() << wasBetter << endl;\r
+ }\r
+ \r
+ report.setCandidate(candidateSeq);\r
+ report.setTemplate(templateSeq);\r
+ report.setSearchParameters(search, searchScore);\r
+ report.setAlignmentParameters(align, alignment);\r
+ report.setNastParameters(*nast);\r
+ \r
+ alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;\r
+ \r
+ report.print();\r
+ delete nast;\r
+ if (needToDeleteCopy) { delete copy; }\r
+ }\r
+ delete candidateSeq;\r
+ \r
+ //report progress\r
+ if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); }\r
+ }\r
+ //report progress\r
+ if((line->numSeqs) % 100 != 0){ m->mothurOut(toString(line->numSeqs)); m->mothurOutEndLine(); }\r
+ \r
+ alignmentFile.close();\r
+ inFASTA.close();\r
+ accnosFile.close();\r
+ \r
+ return 1;\r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "driver");\r
+ exit(1);\r
+ }\r
+}\r
+//**********************************************************************************************************************\r
+#ifdef USE_MPI\r
+int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector<long>& MPIPos){\r
+ try {\r
+ string outputString = "";\r
+ MPI_Status statusReport; \r
+ MPI_Status statusAlign; \r
+ MPI_Status statusAccnos; \r
+ MPI_Status status; \r
+ int pid;\r
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are\r
+ \r
+ NastReport report;\r
+ \r
+ if (pid == 0) {\r
+ outputString = report.getHeaders();\r
+ int length = outputString.length();\r
+ \r
+ char* buf = new char[length];\r
+ memcpy(buf, outputString.c_str(), length);\r
+ \r
+ MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport);\r
+\r
+ delete buf;\r
+ }\r
+ \r
+ for(int i=0;i<num;i++){\r
+ \r
+ if (m->control_pressed) { return 0; }\r
+\r
+ //read next sequence\r
+ int length = MPIPos[start+i+1] - MPIPos[start+i];\r
+\r
+ char* buf4 = new char[length];\r
+ memcpy(buf4, outputString.c_str(), length);\r
+\r
+ MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);\r
+ \r
+ string tempBuf = buf4;\r
+\r
+ delete buf4;\r
+\r
+ if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }\r
+ istringstream iss (tempBuf,istringstream::in);\r
+ \r
+ Sequence* candidateSeq = new Sequence(iss); \r
+ int origNumBases = candidateSeq->getNumBases();\r
+ string originalUnaligned = candidateSeq->getUnaligned();\r
+ int numBasesNeeded = origNumBases * threshold;\r
+ \r
+ if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file\r
+ if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {\r
+ alignment->resize(candidateSeq->getUnaligned().length()+1);\r
+ }\r
+ \r
+ Sequence temp = templateDB->findClosestSequence(candidateSeq);\r
+ Sequence* templateSeq = &temp;\r
+ \r
+ float searchScore = templateDB->getSearchScore();\r
+ \r
+ Nast* nast = new Nast(alignment, candidateSeq, templateSeq);\r
+ Sequence* copy;\r
+ \r
+ Nast* nast2;\r
+ bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below\r
+ //since nast does not make a copy of hte sequence passed, and it is used by the reporter below\r
+ //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place\r
+ //so this bool tells you if you need to delete it\r
+ \r
+ //if there is a possibility that this sequence should be reversed\r
+ if (candidateSeq->getNumBases() < numBasesNeeded) {\r
+ \r
+ string wasBetter = "";\r
+ //if the user wants you to try the reverse\r
+ if (flip) {\r
+ //get reverse compliment\r
+ copy = new Sequence(candidateSeq->getName(), originalUnaligned);\r
+ copy->reverseComplement();\r
+ \r
+ //rerun alignment\r
+ Sequence temp2 = templateDB->findClosestSequence(copy);\r
+ Sequence* templateSeq2 = &temp2;\r
+ \r
+ searchScore = templateDB->getSearchScore();\r
+ \r
+ nast2 = new Nast(alignment, copy, templateSeq2);\r
+ \r
+ //check if any better\r
+ if (copy->getNumBases() > candidateSeq->getNumBases()) {\r
+ candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better\r
+ templateSeq = templateSeq2; \r
+ delete nast;\r
+ nast = nast2;\r
+ needToDeleteCopy = true;\r
+ }else{ \r
+ wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";\r
+ delete nast2;\r
+ delete copy; \r
+ }\r
+ }\r
+ \r
+ //create accnos file with names\r
+ outputString = candidateSeq->getName() + wasBetter + "\n";\r
+ \r
+ //send results to parent\r
+ int length = outputString.length();\r
+\r
+ char* buf = new char[length];\r
+ memcpy(buf, outputString.c_str(), length);\r
+ \r
+ MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos);\r
+ delete buf;\r
+ MPIWroteAccnos = true;\r
+ }\r
+ \r
+ report.setCandidate(candidateSeq);\r
+ report.setTemplate(templateSeq);\r
+ report.setSearchParameters(search, searchScore);\r
+ report.setAlignmentParameters(align, alignment);\r
+ report.setNastParameters(*nast);\r
+ \r
+ outputString = ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n";\r
+ \r
+ //send results to parent\r
+ int length = outputString.length();\r
+ char* buf2 = new char[length];\r
+ memcpy(buf2, outputString.c_str(), length);\r
+ \r
+ MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign);\r
+ \r
+ delete buf2;\r
+\r
+ outputString = report.getReport();\r
+ \r
+ //send results to parent\r
+ length = outputString.length();\r
+ char* buf3 = new char[length];\r
+ memcpy(buf3, outputString.c_str(), length);\r
+ \r
+ MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport);\r
+ \r
+ delete buf3;\r
+ delete nast;\r
+ if (needToDeleteCopy) { delete copy; }\r
+ }\r
+ delete candidateSeq;\r
+ \r
+ //report progress\r
+ if((i+1) % 100 == 0){ cout << (toString(i+1)) << endl; }\r
+ }\r
+ //report progress\r
+ if((num) % 100 != 0){ cout << (toString(num)) << endl; }\r
+ \r
+ return 1;\r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "driverMPI");\r
+ exit(1);\r
+ }\r
+}\r
+#endif\r
+/**************************************************************************************************/\r
+\r
+int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) {\r
+ try {\r
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)\r
+ int process = 0;\r
+ int exitCommand = 1;\r
+ // processIDS.resize(0);\r
+ \r
+ //loop through and create all the processes you want\r
+ while (process != processors) {\r
+ int pid = fork();\r
+ \r
+ if (pid > 0) {\r
+ processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later\r
+ process++;\r
+ }else if (pid == 0){\r
+ exitCommand = driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename);\r
+ exit(0);\r
+ }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }\r
+ }\r
+ \r
+ //force parent to wait until all the processes are done\r
+ for (int i=0;i<processors;i++) { \r
+ int temp = processIDS[i];\r
+ wait(&temp);\r
+ }\r
+ \r
+ return exitCommand;\r
+#endif \r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "createProcesses");\r
+ exit(1);\r
+ }\r
+}\r
+\r
+/**************************************************************************************************/\r
+\r
+void AlignCommand::appendAlignFiles(string temp, string filename) {\r
+ try{\r
+ \r
+ ofstream output;\r
+ ifstream input;\r
+ openOutputFileAppend(filename, output);\r
+ openInputFile(temp, input);\r
+ \r
+ while(char c = input.get()){\r
+ if(input.eof()) { break; }\r
+ else { output << c; }\r
+ }\r
+ \r
+ input.close();\r
+ output.close();\r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "appendAlignFiles");\r
+ exit(1);\r
+ }\r
+}\r
+//**********************************************************************************************************************\r
+\r
+void AlignCommand::appendReportFiles(string temp, string filename) {\r
+ try{\r
+ \r
+ ofstream output;\r
+ ifstream input;\r
+ openOutputFileAppend(filename, output);\r
+ openInputFile(temp, input);\r
+\r
+ while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line\r
+ \r
+ while(char c = input.get()){\r
+ if(input.eof()) { break; }\r
+ else { output << c; }\r
+ }\r
+ \r
+ input.close();\r
+ output.close();\r
+ }\r
+ catch(exception& e) {\r
+ m->errorOut(e, "AlignCommand", "appendReportFiles");\r
+ exit(1);\r
+ }\r
+}\r
+//**********************************************************************************************************************\r
projects / mothur.git / blobdiff