From 6de5adaae66b28aa60a75f123005cede410c156c Mon Sep 17 00:00:00 2001 From: westcott Date: Thu, 10 Jun 2010 17:53:15 +0000 Subject: [PATCH] committing so I can work on the other machine --- aligncommand.cpp | 1792 ++++++++++++++++++++++----------------------- chimeraslayer.cpp | 4 +- decalc.cpp | 2 + distancedb.cpp | 4 +- makefile | 4 +- readblast.cpp | 7 +- readtree.cpp | 42 +- treenode.cpp | 6 +- 8 files changed, 949 insertions(+), 912 deletions(-) diff --git a/aligncommand.cpp b/aligncommand.cpp index 03992dc..c5407ca 100644 --- a/aligncommand.cpp +++ b/aligncommand.cpp @@ -1,896 +1,896 @@ -/* - * 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). - * - */ - -#include "aligncommand.h" -#include "sequence.hpp" - -#include "gotohoverlap.hpp" -#include "needlemanoverlap.hpp" -#include "blastalign.hpp" -#include "noalign.hpp" - -#include "nast.hpp" -#include "nastreport.hpp" - - -//********************************************************************************************************************** - -AlignCommand::AlignCommand(string option) { - try { - - abort = false; - - //allow user to run help - if(option == "help") { help(); abort = true; } - - else { - - //valid paramters for this command - string AlignArray[] = {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors","flip","threshold","outputdir","inputdir"}; - vector myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string))); - - OptionParser parser(option); - map parameters = parser.getParameters(); - - ValidParameters validParameter; - map::iterator it; - - //check to make sure all parameters are valid for command - for (it = parameters.begin(); it != parameters.end(); it++) { - if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { 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 = ""; } - - - //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); - - if (inputDir == "not found"){ inputDir = ""; } - else { - string path; - - it = parameters.find("template"); - - //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["template"] = inputDir + it->second; } - } - } - - //check for required parameters - templateFileName = validParameter.validFile(parameters, "template", true); - - if (templateFileName == "not found") { - m->mothurOut("template is a required parameter for the align.seqs command."); - m->mothurOutEndLine(); - abort = true; - }else if (templateFileName == "not open") { abort = true; } - - candidateFileName = validParameter.validFile(parameters, "candidate", false); - if (candidateFileName == "not found") { m->mothurOut("candidate is a required parameter for the align.seqs command."); m->mothurOutEndLine(); abort = true; } - else { - splitAtDash(candidateFileName, candidateFileNames); - - //go through files and make sure they are good, if not, then disregard them - for (int i = 0; i < candidateFileNames.size(); i++) { - if (inputDir != "") { - string path = hasPath(candidateFileNames[i]); - //if the user has not given a path then, add inputdir. else leave path alone. - if (path == "") { candidateFileNames[i] = inputDir + candidateFileNames[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 - - ableToOpen = openInputFile(candidateFileNames[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(candidateFileNames[i] + " will be disregarded."); m->mothurOutEndLine(); - //erase from file list - candidateFileNames.erase(candidateFileNames.begin()+i); - i--; - } - - } - - //make sure there is at least one valid file left - if (candidateFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; } - } - - //check for optional parameter and set defaults - // ...at some point should added some additional type checking... - string temp; - temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; } - convert(temp, kmerSize); - - temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; } - convert(temp, match); - - temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; } - convert(temp, misMatch); - - temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; } - convert(temp, gapOpen); - - temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; } - convert(temp, gapExtend); - - temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; } - convert(temp, processors); - - temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "f"; } - flip = isTrue(temp); - - temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found"){ temp = "0.50"; } - convert(temp, threshold); - - search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; } - - align = validParameter.validFile(parameters, "align", false); if (align == "not found"){ align = "needleman"; } - } - - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "AlignCommand"); - exit(1); - } -} - -//********************************************************************************************************************** - -AlignCommand::~AlignCommand(){ - - if (abort == false) { - for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); - delete templateDB; - delete alignment; - } -} - -//********************************************************************************************************************** - -void AlignCommand::help(){ - try { - m->mothurOut("The align.seqs command reads a file containing sequences and creates an alignment file and a report file.\n"); - m->mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen and gapextend.\n"); - 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"); - 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"); - 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"); - m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n"); - m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n"); - m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n"); - m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n"); - m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n"); - 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"); - 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"); - 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"); - 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"); - m->mothurOut("The align.seqs command should be in the following format: \n"); - m->mothurOut("align.seqs(template=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) \n"); - 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"); - m->mothurOut("Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).\n\n"); - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "help"); - exit(1); - } -} - - -//********************************************************************************************************************** - -int AlignCommand::execute(){ - try { - if (abort == true) { return 0; } - - templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch); - int longestBase = templateDB->getLongestBase(); - - if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase); } - else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); } - else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); } - else if(align == "noalign") { alignment = new NoAlign(); } - else { - m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman."); - m->mothurOutEndLine(); - alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); - } - vector outputNames; - - for (int s = 0; s < candidateFileNames.size(); s++) { - if (m->control_pressed) { return 0; } - - m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine(); - - if (outputDir == "") { outputDir += hasPath(candidateFileNames[s]); } - string alignFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align"; - string reportFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align.report"; - string accnosFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "flip.accnos"; - bool hasAccnos = true; - - int numFastaSeqs = 0; - for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); - int start = time(NULL); - -#ifdef USE_MPI - int pid, end, numSeqsPerProcessor; - int tag = 2001; - vector 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 outMPIAlign; - MPI_File outMPIReport; - MPI_File outMPIAccnos; - - int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY; - int inMode=MPI_MODE_RDONLY; - - //char* outAlignFilename = new char[alignFileName.length()]; - //memcpy(outAlignFilename, alignFileName.c_str(), alignFileName.length()); - - char outAlignFilename[1024]; - strcpy(outAlignFilename, alignFileName.c_str()); - - //char* outReportFilename = new char[reportFileName.length()]; - //memcpy(outReportFilename, reportFileName.c_str(), reportFileName.length()); - - char outReportFilename[1024]; - strcpy(outReportFilename, reportFileName.c_str()); - - //char* outAccnosFilename = new char[accnosFileName.length()]; - //memcpy(outAccnosFilename, accnosFileName.c_str(), accnosFileName.length()); - - char outAccnosFilename[1024]; - strcpy(outAccnosFilename, accnosFileName.c_str()); - - //char* inFileName = new char[candidateFileNames[s].length()]; - //memcpy(inFileName, candidateFileNames[s].c_str(), candidateFileNames[s].length()); - - char inFileName[1024]; - strcpy(inFileName, candidateFileNames[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, outAlignFilename, outMode, MPI_INFO_NULL, &outMPIAlign); - MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport); - MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos); - - if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; } - - if (pid == 0) { //you are the root process - - MPIPos = setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs - - //send file positions to all processes - for(int i = 1; i < processors; i++) { - MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD); - MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD); - } - - //figure out how many sequences you have to align - numSeqsPerProcessor = numFastaSeqs / processors; - int startIndex = pid * numSeqsPerProcessor; - if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } - - - //align your part - driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos); - - if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); 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(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status); - MPIPos.resize(numFastaSeqs+1); - MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status); - - - //figure out how many sequences you have to align - numSeqsPerProcessor = numFastaSeqs / processors; - int startIndex = pid * numSeqsPerProcessor; - if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } - - - //align your part - driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos); - - if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; } - - MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD); - } - - //close files - MPI_File_close(&inMPI); - MPI_File_close(&outMPIAlign); - MPI_File_close(&outMPIReport); - MPI_File_close(&outMPIAccnos); - - //delete accnos file if blank - if (pid == 0) { - //delete accnos file if its blank else report to user - if (MPIWroteAccnos) { - m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); - if (!flip) { - m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); - }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } - m->mothurOutEndLine(); - }else { - //MPI_Info info; - //MPI_File_delete(outAccnosFilename, info); - hasAccnos = false; - remove(accnosFileName.c_str()); - } - } - -#else - - #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - if(processors == 1){ - ifstream inFASTA; - openInputFile(candidateFileNames[s], inFASTA); - numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); - inFASTA.close(); - - lines.push_back(new linePair(0, numFastaSeqs)); - - driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); - - if (m->control_pressed) { - remove(accnosFileName.c_str()); - remove(alignFileName.c_str()); - remove(reportFileName.c_str()); - return 0; - } - - //delete accnos file if its blank else report to user - if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; } - else { - m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); - if (!flip) { - m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); - }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } - m->mothurOutEndLine(); - } - } - else{ - vector positions; - processIDS.resize(0); - - ifstream inFASTA; - openInputFile(candidateFileNames[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(); - - numFastaSeqs = positions.size(); - - int numSeqsPerProcessor = numFastaSeqs / processors; - - for (int i = 0; i < processors; i++) { - long int startPos = positions[ i * numSeqsPerProcessor ]; - if(i == processors - 1){ - numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; - } - lines.push_back(new linePair(startPos, numSeqsPerProcessor)); - } - - createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); - - rename((alignFileName + toString(processIDS[0]) + ".temp").c_str(), alignFileName.c_str()); - rename((reportFileName + toString(processIDS[0]) + ".temp").c_str(), reportFileName.c_str()); - - //append alignment and report files - for(int i=1;i nonBlankAccnosFiles; - //delete blank accnos files generated with multiple processes - for(int i=0;imothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); - if (!flip) { - m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); - }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } - m->mothurOutEndLine(); - }else{ hasAccnos = false; } - - if (m->control_pressed) { - remove(accnosFileName.c_str()); - remove(alignFileName.c_str()); - remove(reportFileName.c_str()); - return 0; - } - } - #else - ifstream inFASTA; - openInputFile(candidateFileNames[s], inFASTA); - numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); - inFASTA.close(); - - lines.push_back(new linePair(0, numFastaSeqs)); - - driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); - - if (m->control_pressed) { - remove(accnosFileName.c_str()); - remove(alignFileName.c_str()); - remove(reportFileName.c_str()); - return 0; - } - - //delete accnos file if its blank else report to user - if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; } - else { - m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); - if (!flip) { - m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); - }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } - m->mothurOutEndLine(); - } - - #endif - -#endif - - - #ifdef USE_MPI - MPI_Comm_rank(MPI_COMM_WORLD, &pid); - - if (pid == 0) { //only one process should output to screen - #endif - - outputNames.push_back(alignFileName); - outputNames.push_back(reportFileName); - if (hasAccnos) { outputNames.push_back(accnosFileName); } - - #ifdef USE_MPI - } - #endif - - m->mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences."); - m->mothurOutEndLine(); - m->mothurOutEndLine(); - } - - - m->mothurOutEndLine(); - m->mothurOut("Output File Names: "); m->mothurOutEndLine(); - for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); } - m->mothurOutEndLine(); - - return 0; - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "execute"); - exit(1); - } -} - -//********************************************************************************************************************** - -int AlignCommand::driver(linePair* line, string alignFName, string reportFName, string accnosFName, string filename){ - try { - ofstream alignmentFile; - openOutputFile(alignFName, alignmentFile); - - ofstream accnosFile; - openOutputFile(accnosFName, accnosFile); - - NastReport report(reportFName); - - ifstream inFASTA; - openInputFile(filename, inFASTA); - - inFASTA.seekg(line->start); - - for(int i=0;inumSeqs;i++){ - - if (m->control_pressed) { return 0; } - - Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA); - - int origNumBases = candidateSeq->getNumBases(); - string originalUnaligned = candidateSeq->getUnaligned(); - int numBasesNeeded = origNumBases * threshold; - - if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file - if (candidateSeq->getUnaligned().length() > alignment->getnRows()) { - alignment->resize(candidateSeq->getUnaligned().length()+1); - } - - Sequence temp = templateDB->findClosestSequence(candidateSeq); - Sequence* templateSeq = &temp; - - float searchScore = templateDB->getSearchScore(); - - Nast* nast = new Nast(alignment, candidateSeq, templateSeq); - Sequence* copy; - - Nast* nast2; - bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below - //since nast does not make a copy of hte sequence passed, and it is used by the reporter below - //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place - //so this bool tells you if you need to delete it - - //if there is a possibility that this sequence should be reversed - if (candidateSeq->getNumBases() < numBasesNeeded) { - - string wasBetter = ""; - //if the user wants you to try the reverse - if (flip) { - //get reverse compliment - copy = new Sequence(candidateSeq->getName(), originalUnaligned); - copy->reverseComplement(); - - //rerun alignment - Sequence temp2 = templateDB->findClosestSequence(copy); - Sequence* templateSeq2 = &temp2; - - searchScore = templateDB->getSearchScore(); - - nast2 = new Nast(alignment, copy, templateSeq2); - - //check if any better - if (copy->getNumBases() > candidateSeq->getNumBases()) { - candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better - templateSeq = templateSeq2; - delete nast; - nast = nast2; - needToDeleteCopy = true; - }else{ - wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence."; - delete nast2; - delete copy; - } - } - - //create accnos file with names - accnosFile << candidateSeq->getName() << wasBetter << endl; - } - - report.setCandidate(candidateSeq); - report.setTemplate(templateSeq); - report.setSearchParameters(search, searchScore); - report.setAlignmentParameters(align, alignment); - report.setNastParameters(*nast); - - alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl; - - report.print(); - delete nast; - if (needToDeleteCopy) { delete copy; } - } - delete candidateSeq; - - //report progress - if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); } - } - //report progress - if((line->numSeqs) % 100 != 0){ m->mothurOut(toString(line->numSeqs)); m->mothurOutEndLine(); } - - alignmentFile.close(); - inFASTA.close(); - accnosFile.close(); - - return 1; - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "driver"); - exit(1); - } -} -//********************************************************************************************************************** -#ifdef USE_MPI -int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector& MPIPos){ - try { - string outputString = ""; - MPI_Status statusReport; - MPI_Status statusAlign; - MPI_Status statusAccnos; - MPI_Status status; - int pid; - MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are - - NastReport report; - - if (pid == 0) { - outputString = report.getHeaders(); - int length = outputString.length(); - - char* buf = new char[length]; - memcpy(buf, outputString.c_str(), length); - - MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport); - - delete buf; - } - - for(int i=0;icontrol_pressed) { return 0; } - - //read next sequence - int length = MPIPos[start+i+1] - MPIPos[start+i]; - - char* buf4 = new char[length]; - memcpy(buf4, outputString.c_str(), length); - - MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status); - - string tempBuf = buf4; - - delete buf4; - - if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); } - istringstream iss (tempBuf,istringstream::in); - - Sequence* candidateSeq = new Sequence(iss); - int origNumBases = candidateSeq->getNumBases(); - string originalUnaligned = candidateSeq->getUnaligned(); - int numBasesNeeded = origNumBases * threshold; - - if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file - if (candidateSeq->getUnaligned().length() > alignment->getnRows()) { - alignment->resize(candidateSeq->getUnaligned().length()+1); - } - - Sequence temp = templateDB->findClosestSequence(candidateSeq); - Sequence* templateSeq = &temp; - - float searchScore = templateDB->getSearchScore(); - - Nast* nast = new Nast(alignment, candidateSeq, templateSeq); - Sequence* copy; - - Nast* nast2; - bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below - //since nast does not make a copy of hte sequence passed, and it is used by the reporter below - //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place - //so this bool tells you if you need to delete it - - //if there is a possibility that this sequence should be reversed - if (candidateSeq->getNumBases() < numBasesNeeded) { - - string wasBetter = ""; - //if the user wants you to try the reverse - if (flip) { - //get reverse compliment - copy = new Sequence(candidateSeq->getName(), originalUnaligned); - copy->reverseComplement(); - - //rerun alignment - Sequence temp2 = templateDB->findClosestSequence(copy); - Sequence* templateSeq2 = &temp2; - - searchScore = templateDB->getSearchScore(); - - nast2 = new Nast(alignment, copy, templateSeq2); - - //check if any better - if (copy->getNumBases() > candidateSeq->getNumBases()) { - candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better - templateSeq = templateSeq2; - delete nast; - nast = nast2; - needToDeleteCopy = true; - }else{ - wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence."; - delete nast2; - delete copy; - } - } - - //create accnos file with names - outputString = candidateSeq->getName() + wasBetter + "\n"; - - //send results to parent - int length = outputString.length(); - - char* buf = new char[length]; - memcpy(buf, outputString.c_str(), length); - - MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos); - delete buf; - MPIWroteAccnos = true; - } - - report.setCandidate(candidateSeq); - report.setTemplate(templateSeq); - report.setSearchParameters(search, searchScore); - report.setAlignmentParameters(align, alignment); - report.setNastParameters(*nast); - - outputString = ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n"; - - //send results to parent - int length = outputString.length(); - char* buf2 = new char[length]; - memcpy(buf2, outputString.c_str(), length); - - MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign); - - delete buf2; - - outputString = report.getReport(); - - //send results to parent - length = outputString.length(); - char* buf3 = new char[length]; - memcpy(buf3, outputString.c_str(), length); - - MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport); - - delete buf3; - delete nast; - if (needToDeleteCopy) { delete copy; } - } - delete candidateSeq; - - //report progress - if((i+1) % 100 == 0){ cout << (toString(i+1)) << endl; } - } - //report progress - if((num) % 100 != 0){ cout << (toString(num)) << endl; } - - return 1; - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "driverMPI"); - exit(1); - } -} -#endif -/**************************************************************************************************/ - -int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) { - try { -#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - int process = 0; - int exitCommand = 1; - // processIDS.resize(0); - - //loop through and create all the processes you want - while (process != processors) { - int pid = fork(); - - if (pid > 0) { - processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later - process++; - }else if (pid == 0){ - exitCommand = driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename); - exit(0); - }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); } - } - - //force parent to wait until all the processes are done - for (int i=0;ierrorOut(e, "AlignCommand", "createProcesses"); - exit(1); - } -} - -/**************************************************************************************************/ - -void AlignCommand::appendAlignFiles(string temp, string filename) { - try{ - - ofstream output; - ifstream input; - openOutputFileAppend(filename, output); - openInputFile(temp, input); - - while(char c = input.get()){ - if(input.eof()) { break; } - else { output << c; } - } - - input.close(); - output.close(); - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "appendAlignFiles"); - exit(1); - } -} -//********************************************************************************************************************** - -void AlignCommand::appendReportFiles(string temp, string filename) { - try{ - - ofstream output; - ifstream input; - openOutputFileAppend(filename, output); - openInputFile(temp, input); - - while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line - - while(char c = input.get()){ - if(input.eof()) { break; } - else { output << c; } - } - - input.close(); - output.close(); - } - catch(exception& e) { - m->errorOut(e, "AlignCommand", "appendReportFiles"); - exit(1); - } -} -//********************************************************************************************************************** +/* + * 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). + * + */ + +#include "aligncommand.h" +#include "sequence.hpp" + +#include "gotohoverlap.hpp" +#include "needlemanoverlap.hpp" +#include "blastalign.hpp" +#include "noalign.hpp" + +#include "nast.hpp" +#include "nastreport.hpp" + + +//********************************************************************************************************************** + +AlignCommand::AlignCommand(string option) { + try { + + abort = false; + + //allow user to run help + if(option == "help") { help(); abort = true; } + + else { + + //valid paramters for this command + string AlignArray[] = {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors","flip","threshold","outputdir","inputdir"}; + vector myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string))); + + OptionParser parser(option); + map parameters = parser.getParameters(); + + ValidParameters validParameter; + map::iterator it; + + //check to make sure all parameters are valid for command + for (it = parameters.begin(); it != parameters.end(); it++) { + if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { 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 = ""; } + + + //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); + + if (inputDir == "not found"){ inputDir = ""; } + else { + string path; + + it = parameters.find("template"); + + //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["template"] = inputDir + it->second; } + } + } + + //check for required parameters + templateFileName = validParameter.validFile(parameters, "template", true); + + if (templateFileName == "not found") { + m->mothurOut("template is a required parameter for the align.seqs command."); + m->mothurOutEndLine(); + abort = true; + }else if (templateFileName == "not open") { abort = true; } + + candidateFileName = validParameter.validFile(parameters, "candidate", false); + if (candidateFileName == "not found") { m->mothurOut("candidate is a required parameter for the align.seqs command."); m->mothurOutEndLine(); abort = true; } + else { + splitAtDash(candidateFileName, candidateFileNames); + + //go through files and make sure they are good, if not, then disregard them + for (int i = 0; i < candidateFileNames.size(); i++) { + if (inputDir != "") { + string path = hasPath(candidateFileNames[i]); + //if the user has not given a path then, add inputdir. else leave path alone. + if (path == "") { candidateFileNames[i] = inputDir + candidateFileNames[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 + + ableToOpen = openInputFile(candidateFileNames[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(candidateFileNames[i] + " will be disregarded."); m->mothurOutEndLine(); + //erase from file list + candidateFileNames.erase(candidateFileNames.begin()+i); + i--; + } + + } + + //make sure there is at least one valid file left + if (candidateFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; } + } + + //check for optional parameter and set defaults + // ...at some point should added some additional type checking... + string temp; + temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; } + convert(temp, kmerSize); + + temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; } + convert(temp, match); + + temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; } + convert(temp, misMatch); + + temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; } + convert(temp, gapOpen); + + temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; } + convert(temp, gapExtend); + + temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; } + convert(temp, processors); + + temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "f"; } + flip = isTrue(temp); + + temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found"){ temp = "0.50"; } + convert(temp, threshold); + + search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; } + + align = validParameter.validFile(parameters, "align", false); if (align == "not found"){ align = "needleman"; } + } + + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "AlignCommand"); + exit(1); + } +} + +//********************************************************************************************************************** + +AlignCommand::~AlignCommand(){ + + if (abort == false) { + for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); + delete templateDB; + delete alignment; + } +} + +//********************************************************************************************************************** + +void AlignCommand::help(){ + try { + m->mothurOut("The align.seqs command reads a file containing sequences and creates an alignment file and a report file.\n"); + m->mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen and gapextend.\n"); + 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"); + 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"); + 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"); + m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.\n"); + m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n"); + m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.\n"); + m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n"); + m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -1.0.\n"); + 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"); + 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"); + 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"); + 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"); + m->mothurOut("The align.seqs command should be in the following format: \n"); + m->mothurOut("align.seqs(template=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) \n"); + 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"); + m->mothurOut("Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).\n\n"); + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "help"); + exit(1); + } +} + + +//********************************************************************************************************************** + +int AlignCommand::execute(){ + try { + if (abort == true) { return 0; } + + templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch); + int longestBase = templateDB->getLongestBase(); + + if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase); } + else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); } + else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); } + else if(align == "noalign") { alignment = new NoAlign(); } + else { + m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman."); + m->mothurOutEndLine(); + alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); + } + vector outputNames; + + for (int s = 0; s < candidateFileNames.size(); s++) { + if (m->control_pressed) { return 0; } + + m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine(); + + if (outputDir == "") { outputDir += hasPath(candidateFileNames[s]); } + string alignFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align"; + string reportFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align.report"; + string accnosFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "flip.accnos"; + bool hasAccnos = true; + + int numFastaSeqs = 0; + for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear(); + int start = time(NULL); + +#ifdef USE_MPI + int pid, end, numSeqsPerProcessor; + int tag = 2001; + vector 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 outMPIAlign; + MPI_File outMPIReport; + MPI_File outMPIAccnos; + + int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY; + int inMode=MPI_MODE_RDONLY; + + //char* outAlignFilename = new char[alignFileName.length()]; + //memcpy(outAlignFilename, alignFileName.c_str(), alignFileName.length()); + + char outAlignFilename[1024]; + strcpy(outAlignFilename, alignFileName.c_str()); + + //char* outReportFilename = new char[reportFileName.length()]; + //memcpy(outReportFilename, reportFileName.c_str(), reportFileName.length()); + + char outReportFilename[1024]; + strcpy(outReportFilename, reportFileName.c_str()); + + //char* outAccnosFilename = new char[accnosFileName.length()]; + //memcpy(outAccnosFilename, accnosFileName.c_str(), accnosFileName.length()); + + char outAccnosFilename[1024]; + strcpy(outAccnosFilename, accnosFileName.c_str()); + + //char* inFileName = new char[candidateFileNames[s].length()]; + //memcpy(inFileName, candidateFileNames[s].c_str(), candidateFileNames[s].length()); + + char inFileName[1024]; + strcpy(inFileName, candidateFileNames[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, outAlignFilename, outMode, MPI_INFO_NULL, &outMPIAlign); + MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport); + MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos); + + if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; } + + if (pid == 0) { //you are the root process + + MPIPos = setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs + + //send file positions to all processes + for(int i = 1; i < processors; i++) { + MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD); + MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD); + } + + //figure out how many sequences you have to align + numSeqsPerProcessor = numFastaSeqs / processors; + int startIndex = pid * numSeqsPerProcessor; + if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } + + + //align your part + driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos); + + if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); 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(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status); + MPIPos.resize(numFastaSeqs+1); + MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status); + + + //figure out how many sequences you have to align + numSeqsPerProcessor = numFastaSeqs / processors; + int startIndex = pid * numSeqsPerProcessor; + if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; } + + + //align your part + driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos); + + if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; } + + MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD); + } + + //close files + MPI_File_close(&inMPI); + MPI_File_close(&outMPIAlign); + MPI_File_close(&outMPIReport); + MPI_File_close(&outMPIAccnos); + + //delete accnos file if blank + if (pid == 0) { + //delete accnos file if its blank else report to user + if (MPIWroteAccnos) { + m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); + if (!flip) { + m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); + }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } + m->mothurOutEndLine(); + }else { + //MPI_Info info; + //MPI_File_delete(outAccnosFilename, info); + hasAccnos = false; + remove(accnosFileName.c_str()); + } + } + +#else + + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + if(processors == 1){ + ifstream inFASTA; + openInputFile(candidateFileNames[s], inFASTA); + numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); + inFASTA.close(); + + lines.push_back(new linePair(0, numFastaSeqs)); + + driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); + + if (m->control_pressed) { + remove(accnosFileName.c_str()); + remove(alignFileName.c_str()); + remove(reportFileName.c_str()); + return 0; + } + + //delete accnos file if its blank else report to user + if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; } + else { + m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); + if (!flip) { + m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); + }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } + m->mothurOutEndLine(); + } + } + else{ + vector positions; + processIDS.resize(0); + + ifstream inFASTA; + openInputFile(candidateFileNames[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(); + + numFastaSeqs = positions.size(); + + int numSeqsPerProcessor = numFastaSeqs / processors; + + for (int i = 0; i < processors; i++) { + long int startPos = positions[ i * numSeqsPerProcessor ]; + if(i == processors - 1){ + numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; + } + lines.push_back(new linePair(startPos, numSeqsPerProcessor)); + } + + createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); + + rename((alignFileName + toString(processIDS[0]) + ".temp").c_str(), alignFileName.c_str()); + rename((reportFileName + toString(processIDS[0]) + ".temp").c_str(), reportFileName.c_str()); + + //append alignment and report files + for(int i=1;i nonBlankAccnosFiles; + //delete blank accnos files generated with multiple processes + for(int i=0;imothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); + if (!flip) { + m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); + }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } + m->mothurOutEndLine(); + }else{ hasAccnos = false; } + + if (m->control_pressed) { + remove(accnosFileName.c_str()); + remove(alignFileName.c_str()); + remove(reportFileName.c_str()); + return 0; + } + } + #else + ifstream inFASTA; + openInputFile(candidateFileNames[s], inFASTA); + numFastaSeqs=count(istreambuf_iterator(inFASTA),istreambuf_iterator(), '>'); + inFASTA.close(); + + lines.push_back(new linePair(0, numFastaSeqs)); + + driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]); + + if (m->control_pressed) { + remove(accnosFileName.c_str()); + remove(alignFileName.c_str()); + remove(reportFileName.c_str()); + return 0; + } + + //delete accnos file if its blank else report to user + if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; } + else { + m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + "."); + if (!flip) { + m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well."); + }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); } + m->mothurOutEndLine(); + } + + #endif + +#endif + + + #ifdef USE_MPI + MPI_Comm_rank(MPI_COMM_WORLD, &pid); + + if (pid == 0) { //only one process should output to screen + #endif + + outputNames.push_back(alignFileName); + outputNames.push_back(reportFileName); + if (hasAccnos) { outputNames.push_back(accnosFileName); } + + #ifdef USE_MPI + } + #endif + + m->mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences."); + m->mothurOutEndLine(); + m->mothurOutEndLine(); + } + + + m->mothurOutEndLine(); + m->mothurOut("Output File Names: "); m->mothurOutEndLine(); + for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); } + m->mothurOutEndLine(); + + return 0; + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "execute"); + exit(1); + } +} + +//********************************************************************************************************************** + +int AlignCommand::driver(linePair* line, string alignFName, string reportFName, string accnosFName, string filename){ + try { + ofstream alignmentFile; + openOutputFile(alignFName, alignmentFile); + + ofstream accnosFile; + openOutputFile(accnosFName, accnosFile); + + NastReport report(reportFName); + + ifstream inFASTA; + openInputFile(filename, inFASTA); + + inFASTA.seekg(line->start); + + for(int i=0;inumSeqs;i++){ + + if (m->control_pressed) { return 0; } + + Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA); + + int origNumBases = candidateSeq->getNumBases(); + string originalUnaligned = candidateSeq->getUnaligned(); + int numBasesNeeded = origNumBases * threshold; + + if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file + if (candidateSeq->getUnaligned().length() > alignment->getnRows()) { + alignment->resize(candidateSeq->getUnaligned().length()+1); + } + + Sequence temp = templateDB->findClosestSequence(candidateSeq); + Sequence* templateSeq = &temp; + + float searchScore = templateDB->getSearchScore(); + + Nast* nast = new Nast(alignment, candidateSeq, templateSeq); + Sequence* copy; + + Nast* nast2; + bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below + //since nast does not make a copy of hte sequence passed, and it is used by the reporter below + //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place + //so this bool tells you if you need to delete it + + //if there is a possibility that this sequence should be reversed + if (candidateSeq->getNumBases() < numBasesNeeded) { + + string wasBetter = ""; + //if the user wants you to try the reverse + if (flip) { + //get reverse compliment + copy = new Sequence(candidateSeq->getName(), originalUnaligned); + copy->reverseComplement(); + + //rerun alignment + Sequence temp2 = templateDB->findClosestSequence(copy); + Sequence* templateSeq2 = &temp2; + + searchScore = templateDB->getSearchScore(); + + nast2 = new Nast(alignment, copy, templateSeq2); + + //check if any better + if (copy->getNumBases() > candidateSeq->getNumBases()) { + candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better + templateSeq = templateSeq2; + delete nast; + nast = nast2; + needToDeleteCopy = true; + }else{ + wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence."; + delete nast2; + delete copy; + } + } + + //create accnos file with names + accnosFile << candidateSeq->getName() << wasBetter << endl; + } + + report.setCandidate(candidateSeq); + report.setTemplate(templateSeq); + report.setSearchParameters(search, searchScore); + report.setAlignmentParameters(align, alignment); + report.setNastParameters(*nast); + + alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl; + + report.print(); + delete nast; + if (needToDeleteCopy) { delete copy; } + } + delete candidateSeq; + + //report progress + if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); } + } + //report progress + if((line->numSeqs) % 100 != 0){ m->mothurOut(toString(line->numSeqs)); m->mothurOutEndLine(); } + + alignmentFile.close(); + inFASTA.close(); + accnosFile.close(); + + return 1; + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "driver"); + exit(1); + } +} +//********************************************************************************************************************** +#ifdef USE_MPI +int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector& MPIPos){ + try { + string outputString = ""; + MPI_Status statusReport; + MPI_Status statusAlign; + MPI_Status statusAccnos; + MPI_Status status; + int pid; + MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are + + NastReport report; + + if (pid == 0) { + outputString = report.getHeaders(); + int length = outputString.length(); + + char* buf = new char[length]; + memcpy(buf, outputString.c_str(), length); + + MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport); + + delete buf; + } + + for(int i=0;icontrol_pressed) { return 0; } + + //read next sequence + int length = MPIPos[start+i+1] - MPIPos[start+i]; + + char* buf4 = new char[length]; + memcpy(buf4, outputString.c_str(), length); + + MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status); + + string tempBuf = buf4; + + delete buf4; + + if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); } + istringstream iss (tempBuf,istringstream::in); + + Sequence* candidateSeq = new Sequence(iss); + int origNumBases = candidateSeq->getNumBases(); + string originalUnaligned = candidateSeq->getUnaligned(); + int numBasesNeeded = origNumBases * threshold; + + if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file + if (candidateSeq->getUnaligned().length() > alignment->getnRows()) { + alignment->resize(candidateSeq->getUnaligned().length()+1); + } + + Sequence temp = templateDB->findClosestSequence(candidateSeq); + Sequence* templateSeq = &temp; + + float searchScore = templateDB->getSearchScore(); + + Nast* nast = new Nast(alignment, candidateSeq, templateSeq); + Sequence* copy; + + Nast* nast2; + bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below + //since nast does not make a copy of hte sequence passed, and it is used by the reporter below + //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place + //so this bool tells you if you need to delete it + + //if there is a possibility that this sequence should be reversed + if (candidateSeq->getNumBases() < numBasesNeeded) { + + string wasBetter = ""; + //if the user wants you to try the reverse + if (flip) { + //get reverse compliment + copy = new Sequence(candidateSeq->getName(), originalUnaligned); + copy->reverseComplement(); + + //rerun alignment + Sequence temp2 = templateDB->findClosestSequence(copy); + Sequence* templateSeq2 = &temp2; + + searchScore = templateDB->getSearchScore(); + + nast2 = new Nast(alignment, copy, templateSeq2); + + //check if any better + if (copy->getNumBases() > candidateSeq->getNumBases()) { + candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better + templateSeq = templateSeq2; + delete nast; + nast = nast2; + needToDeleteCopy = true; + }else{ + wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence."; + delete nast2; + delete copy; + } + } + + //create accnos file with names + outputString = candidateSeq->getName() + wasBetter + "\n"; + + //send results to parent + int length = outputString.length(); + + char* buf = new char[length]; + memcpy(buf, outputString.c_str(), length); + + MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos); + delete buf; + MPIWroteAccnos = true; + } + + report.setCandidate(candidateSeq); + report.setTemplate(templateSeq); + report.setSearchParameters(search, searchScore); + report.setAlignmentParameters(align, alignment); + report.setNastParameters(*nast); + + outputString = ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n"; + + //send results to parent + int length = outputString.length(); + char* buf2 = new char[length]; + memcpy(buf2, outputString.c_str(), length); + + MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign); + + delete buf2; + + outputString = report.getReport(); + + //send results to parent + length = outputString.length(); + char* buf3 = new char[length]; + memcpy(buf3, outputString.c_str(), length); + + MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport); + + delete buf3; + delete nast; + if (needToDeleteCopy) { delete copy; } + } + delete candidateSeq; + + //report progress + if((i+1) % 100 == 0){ cout << (toString(i+1)) << endl; } + } + //report progress + if((num) % 100 != 0){ cout << (toString(num)) << endl; } + + return 1; + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "driverMPI"); + exit(1); + } +} +#endif +/**************************************************************************************************/ + +int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) { + try { +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + int process = 0; + int exitCommand = 1; + // processIDS.resize(0); + + //loop through and create all the processes you want + while (process != processors) { + int pid = fork(); + + if (pid > 0) { + processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later + process++; + }else if (pid == 0){ + exitCommand = driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename); + exit(0); + }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); } + } + + //force parent to wait until all the processes are done + for (int i=0;ierrorOut(e, "AlignCommand", "createProcesses"); + exit(1); + } +} + +/**************************************************************************************************/ + +void AlignCommand::appendAlignFiles(string temp, string filename) { + try{ + + ofstream output; + ifstream input; + openOutputFileAppend(filename, output); + openInputFile(temp, input); + + while(char c = input.get()){ + if(input.eof()) { break; } + else { output << c; } + } + + input.close(); + output.close(); + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "appendAlignFiles"); + exit(1); + } +} +//********************************************************************************************************************** + +void AlignCommand::appendReportFiles(string temp, string filename) { + try{ + + ofstream output; + ifstream input; + openOutputFileAppend(filename, output); + openInputFile(temp, input); + + while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line + + while(char c = input.get()){ + if(input.eof()) { break; } + else { output << c; } + } + + input.close(); + output.close(); + } + catch(exception& e) { + m->errorOut(e, "AlignCommand", "appendReportFiles"); + exit(1); + } +} +//********************************************************************************************************************** diff --git a/chimeraslayer.cpp b/chimeraslayer.cpp index 2c4f7fb..5295eac 100644 --- a/chimeraslayer.cpp +++ b/chimeraslayer.cpp @@ -286,13 +286,13 @@ int ChimeraSlayer::getChimeras(Sequence* query) { //referenceSeqs, numWanted, matchScore, misMatchPenalty, divR, minSimilarity maligner = new Maligner(templateSeqs, numWanted, match, misMatch, divR, minSim, minCov, searchMethod, databaseLeft, databaseRight); slayer = new Slayer(window, increment, minSim, divR, iters, minSNP); - + if (m->control_pressed) { return 0; } string chimeraFlag = maligner->getResults(query, decalc); if (m->control_pressed) { return 0; } vector Results = maligner->getOutput(); - + //found in testing realigning only made things worse if (realign) { ChimeraReAligner realigner(templateSeqs, match, misMatch); diff --git a/decalc.cpp b/decalc.cpp index 857bcf2..f5aa850 100644 --- a/decalc.cpp +++ b/decalc.cpp @@ -824,6 +824,8 @@ vector DeCalculator::findClosest(Sequence* querySeq, vector dists.size()) { m->mothurOut("numwanted is larger than the number of template sequences, adjusting numwanted."); m->mothurOutEndLine(); numWanted = dists.size(); } + //cout << numWanted << endl; for (int i = 0; i < numWanted; i++) { //cout << dists[i].seq->getName() << '\t' << dists[i].dist << endl; diff --git a/distancedb.cpp b/distancedb.cpp index 0c1aeeb..5c49a93 100644 --- a/distancedb.cpp +++ b/distancedb.cpp @@ -48,8 +48,8 @@ vector DistanceDB::findClosestSequences(Sequence* query, int numWanted){ bool templateSameLength = true; string sequence = query->getAligned(); vector dists; - - if (numWanted > data.size()) { m->mothurOut("numwanted is larger than the number of template sequences, using 10."); m->mothurOutEndLine(); numWanted = 10; } + + if (numWanted > data.size()) { m->mothurOut("numwanted is larger than the number of template sequences, using "+ toString(data.size()) + "."); m->mothurOutEndLine(); numWanted = data.size(); } if (sequence.length() != templateSeqsLength) { templateSameLength = false; } diff --git a/makefile b/makefile index 8c33659..3e83112 100644 --- a/makefile +++ b/makefile @@ -463,7 +463,7 @@ mothur : \ ./logsd.o\ ./geom.o\ ./setlogfilecommand.o\ - -o mothur + -o ../Release/mothur clean : rm \ @@ -1708,7 +1708,7 @@ install : mothur ./classifyotucommand.o : classifyotucommand.cpp $(CC) $(CC_OPTIONS) classifyotucommand.cpp -c $(INCLUDE) -o ./classifyotucommand.o -# Item # 209 -- splitmatrix -- +# Item # 209 -- weightedlinkage -- ./weightedlinkage.o : weightedlinkage.cpp $(CC) $(CC_OPTIONS) weightedlinkage.cpp -c $(INCLUDE) -o ./weightedlinkage.o diff --git a/readblast.cpp b/readblast.cpp index 49e3708..9979a75 100644 --- a/readblast.cpp +++ b/readblast.cpp @@ -311,11 +311,15 @@ int ReadBlast::readNames(NameAssignment* nameMap) { ifstream in; openInputFile(blastfile, in); + ofstream outName; + openOutputFile("tempOutNames", outName); + //read first line in >> prevName; + for (int i = 0; i < 11; i++) { in >> hold; } gobble(in); - + //save name in nameMap nameMap->push_back(prevName); @@ -324,6 +328,7 @@ int ReadBlast::readNames(NameAssignment* nameMap) { //read line in >> name; + for (int i = 0; i < 11; i++) { in >> hold; } gobble(in); diff --git a/readtree.cpp b/readtree.cpp index 51f44e2..1df49b5 100644 --- a/readtree.cpp +++ b/readtree.cpp @@ -296,6 +296,7 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) { int rc = readNewickInt(f, n, T); if (rc == -1) { return -1; } //reports an error in reading + if(f.peek()==')'){ readSpecialChar(f,')',"right parenthesis"); //to pass over labels in trees @@ -316,23 +317,52 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) { } //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating - /* if(f.peek() == ','){ + while(f.peek() == ','){ //force this node to be left child and read new rc T->tree[n].setChildren(lc,rc); T->tree[lc].setParent(n); T->tree[rc].setParent(n); + T->printTree(); cout << endl; + lc = n; n++; + readSpecialChar(f,',',"comma"); - - }*/ - + int rc = readNewickInt(f, n, T); + cout << n << '\t' << lc << '\t' << rc << endl; + if (rc == -1) { return -1; } //reports an error in reading + + if(f.peek()==')'){ + string k; + k = f.peek(); + cout << "in here " << k << endl; + readSpecialChar(f,')',"right parenthesis"); + //to pass over labels in trees + c=filehandle.get(); + while((c!=',') && (c != -1) && (c!= ':') && (c!=';')){ c=filehandle.get(); } + filehandle.putback(c); + + if(f.peek() == ':'){ + readSpecialChar(f,':',"colon"); + + if(n >= numNodes){ m->mothurOut("Error: Too many nodes in input tree\n"); readOk = -1; return -1; } + + T->tree[n].setBranchLength(readBranchLength(f)); + }else{ + T->tree[n].setBranchLength(0.0); + } + + break; + } + } + cout << n << '\t' << lc << '\t' << rc << endl; T->tree[n].setChildren(lc,rc); T->tree[lc].setParent(n); T->tree[rc].setParent(n); - + T->printTree(); cout << endl; + return n++; }else{ f.putback(c); @@ -342,7 +372,7 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) { name += d; d=f.get(); } - + cout << name << endl; int blen = 0; if(d == ':') { blen = 1; } diff --git a/treenode.cpp b/treenode.cpp index c00ade0..0131f59 100644 --- a/treenode.cpp +++ b/treenode.cpp @@ -61,9 +61,9 @@ int Node::getIndex() { return vectorIndex; } //to be used by printTree in the Tree class to print the leaf info void Node::printNode() { try{ - m->mothurOut(toString(parent) + " " + toString(lchild) + " " + toString(rchild) + " "); + m->mothurOut(name + " " + toString(parent) + " " + toString(lchild) + " " + toString(rchild) + " "); - for (int i = 0; i < group.size(); i++) { m->mothurOut( group[i] + " "); } + /*for (int i = 0; i < group.size(); i++) { m->mothurOut( group[i] + " "); } //there is a branch length if (branchLength != -1) { @@ -78,7 +78,7 @@ void Node::printNode() { m->mothurOut(" |"); for(it=pcount.begin();it!=pcount.end();it++){ m->mothurOut(" " + it->first + ":" + toString(it->second)); - } + }*/ m->mothurOutEndLine(); -- 2.39.2