-/*\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
- MPI_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs\r
- MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos \r
- \r
- //figure out how many sequences you have to align\r
- numSeqsPerProcessor = numFastaSeqs / processors;\r
- if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }\r
- int startIndex = pid * numSeqsPerProcessor;\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_Bcast(&numFastaSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs\r
- MPIPos.resize(numFastaSeqs+1);\r
- MPI_Bcast(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions\r
- \r
- //figure out how many sequences you have to align\r
- numSeqsPerProcessor = numFastaSeqs / processors;\r
- if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }\r
- int startIndex = pid * numSeqsPerProcessor;\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
+/*
+ * 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 "referencedb.h"
+
+//**********************************************************************************************************************
+vector<string> AlignCommand::setParameters(){
+ try {
+ CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(ptemplate);
+ CommandParameter pcandidate("fasta", "InputTypes", "", "", "none", "none", "none","fasta-alignreport-accnos",false,true,true); parameters.push_back(pcandidate);
+ CommandParameter psearch("search", "Multiple", "kmer-blast-suffix", "kmer", "", "", "","",false,false,true); parameters.push_back(psearch);
+ CommandParameter pksize("ksize", "Number", "", "8", "", "", "","",false,false); parameters.push_back(pksize);
+ CommandParameter pmatch("match", "Number", "", "1.0", "", "", "","",false,false); parameters.push_back(pmatch);
+ CommandParameter palign("align", "Multiple", "needleman-gotoh-blast-noalign", "needleman", "", "", "","",false,false,true); parameters.push_back(palign);
+ CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "","",false,false); parameters.push_back(pmismatch);
+ CommandParameter pgapopen("gapopen", "Number", "", "-5.0", "", "", "","",false,false); parameters.push_back(pgapopen);
+ CommandParameter pgapextend("gapextend", "Number", "", "-2.0", "", "", "","",false,false); parameters.push_back(pgapextend);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
+ CommandParameter pflip("flip", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pflip);
+ CommandParameter psave("save", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(psave);
+ CommandParameter pthreshold("threshold", "Number", "", "0.50", "", "", "","",false,false); parameters.push_back(pthreshold);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string AlignCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The align.seqs command reads a file containing sequences and creates an alignment file and a report file.";
+ helpString += "The align.seqs command parameters are reference, fasta, search, ksize, align, match, mismatch, gapopen, gapextend and processors.";
+ helpString += "The reference and fasta parameters are required. You may leave fasta blank if you have a valid fasta file. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta.";
+ helpString += "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.";
+ helpString += "The align parameter allows you to specify the alignment method to use. Your options are: gotoh, needleman, blast and noalign. The default is needleman.";
+ helpString += "The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 8.";
+ helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.";
+ helpString += "The mistmatch parameter allows you to specify the penalty for having different bases. The default is -1.0.";
+ helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -5.0.";
+ helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -2.0.";
+ helpString += "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.";
+ helpString += "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.";
+ helpString += "If the flip parameter is set to true the reverse complement of the sequence is aligned and the better alignment is reported.";
+ helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
+ helpString += "The 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.";
+ helpString += "The align.seqs command should be in the following format:";
+ helpString += "align.seqs(reference=yourTemplateFile, fasta=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty)";
+ helpString += "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)";
+ helpString += "Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "getHelpString");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string AlignCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "fasta") { pattern = "[filename],align"; } //makes file like: amazon.align
+ else if (type == "alignreport") { pattern = "[filename],align.report"; }
+ else if (type == "accnos") { pattern = "[filename],flip.accnos"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "getOutputPattern");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+AlignCommand::AlignCommand(){
+ try {
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["alignreport"] = tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "AlignCommand");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+AlignCommand::AlignCommand(string option) {
+ try {
+ abort = false; calledHelp = false;
+ ReferenceDB* rdb = ReferenceDB::getInstance();
+
+ //allow user to run help
+ if(option == "help") { help(); abort = true; calledHelp = true;}
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
+
+ else {
+ vector<string> myArray = setParameters();
+
+ OptionParser parser(option);
+ map<string, string> parameters = parser.getParameters();
+
+ ValidParameters validParameter("align.seqs");
+ map<string, string>::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; }
+ }
+
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["alignreport"] = tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+
+ //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("reference");
+
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["reference"] = inputDir + it->second; }
+ }
+ }
+
+ candidateFileName = validParameter.validFile(parameters, "fasta", false);
+ if (candidateFileName == "not found") {
+ //if there is a current fasta file, use it
+ string filename = m->getFastaFile();
+ if (filename != "") { candidateFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current fastafile and the candidate parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }else {
+ m->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++) {
+ //candidateFileNames[i] = m->getFullPathName(candidateFileNames[i]);
+
+ bool ignore = false;
+ if (candidateFileNames[i] == "current") {
+ candidateFileNames[i] = m->getFastaFile();
+ if (candidateFileNames[i] != "") { m->mothurOut("Using " + candidateFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
+ //erase from file list
+ candidateFileNames.erase(candidateFileNames.begin()+i);
+ i--;
+ }
+ }
+
+ if (!ignore) {
+
+ if (inputDir != "") {
+ string path = m->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;
+ ableToOpen = m->openInputFile(candidateFileNames[i], in, "noerror");
+ in.close();
+
+ //if you can't open it, try default location
+ if (ableToOpen == 1) {
+ if (m->getDefaultPath() != "") { //default path is set
+ string tryPath = m->getDefaultPath() + m->getSimpleName(candidateFileNames[i]);
+ m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
+ ifstream in2;
+ ableToOpen = m->openInputFile(tryPath, in2, "noerror");
+ in2.close();
+ candidateFileNames[i] = tryPath;
+ }
+ }
+
+ //if you can't open it, try output location
+ if (ableToOpen == 1) {
+ if (m->getOutputDir() != "") { //default path is set
+ string tryPath = m->getOutputDir() + m->getSimpleName(candidateFileNames[i]);
+ m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
+ ifstream in2;
+ ableToOpen = m->openInputFile(tryPath, in2, "noerror");
+ in2.close();
+ candidateFileNames[i] = tryPath;
+ }
+ }
+
+
+
+ if (ableToOpen == 1) {
+ m->mothurOut("Unable to open " + candidateFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
+ //erase from file list
+ candidateFileNames.erase(candidateFileNames.begin()+i);
+ i--;
+ }else {
+ m->setFastaFile(candidateFileNames[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"; }
+ m->mothurConvert(temp, kmerSize);
+
+ temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; }
+ m->mothurConvert(temp, match);
+
+ temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; }
+ m->mothurConvert(temp, misMatch);
+
+ temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-5.0"; }
+ m->mothurConvert(temp, gapOpen);
+
+ temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-2.0"; }
+ m->mothurConvert(temp, gapExtend);
+
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
+ m->setProcessors(temp);
+ m->mothurConvert(temp, processors);
+
+ temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "f"; }
+ flip = m->isTrue(temp);
+
+ temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; }
+ save = m->isTrue(temp);
+ rdb->save = save;
+ if (save) { //clear out old references
+ rdb->clearMemory();
+ }
+
+ //this has to go after save so that if the user sets save=t and provides no reference we abort
+ templateFileName = validParameter.validFile(parameters, "reference", true);
+ if (templateFileName == "not found") {
+ //check for saved reference sequences
+ if (rdb->referenceSeqs.size() != 0) {
+ templateFileName = "saved";
+ }else {
+ m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the align.seqs command.");
+ m->mothurOutEndLine();
+ abort = true;
+ }
+ }else if (templateFileName == "not open") { abort = true; }
+ else { if (save) { rdb->setSavedReference(templateFileName); } }
+
+ temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found"){ temp = "0.50"; }
+ m->mothurConvert(temp, threshold);
+
+ search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; }
+ if ((search != "suffix") && (search != "kmer") && (search != "blast")) { m->mothurOut("invalid search option: choices are kmer, suffix or blast."); m->mothurOutEndLine(); abort=true; }
+
+ align = validParameter.validFile(parameters, "align", false); if (align == "not found"){ align = "needleman"; }
+ if ((align != "needleman") && (align != "gotoh") && (align != "blast") && (align != "noalign")) { m->mothurOut("invalid align option: choices are needleman, gotoh, blast or noalign."); m->mothurOutEndLine(); abort=true; }
+
+ }
+
+ }
+ 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;
+ }
+}
+//**********************************************************************************************************************
+
+int AlignCommand::execute(){
+ try {
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
+
+ templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, rand());
+
+ for (int s = 0; s < candidateFileNames.size(); s++) {
+ if (m->control_pressed) { outputTypes.clear(); return 0; }
+
+ m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();
+
+ if (outputDir == "") { outputDir += m->hasPath(candidateFileNames[s]); }
+ map<string, string> variables; variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s]));
+ string alignFileName = getOutputFileName("fasta", variables);
+ string reportFileName = getOutputFileName("alignreport", variables);
+ string accnosFileName = getOutputFileName("accnos", variables);
+
+ 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, numSeqsPerProcessor;
+ int tag = 2001;
+ vector<unsigned long long> MPIPos;
+ MPIWroteAccnos = false;
+
+ MPI_Status status;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
+
+ MPI_File inMPI;
+ MPI_File outMPIAlign;
+ MPI_File outMPIReport;
+ MPI_File outMPIAccnos;
+
+ int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+ int inMode=MPI_MODE_RDONLY;
+
+ char outAlignFilename[1024];
+ strcpy(outAlignFilename, alignFileName.c_str());
+
+ char outReportFilename[1024];
+ strcpy(outReportFilename, reportFileName.c_str());
+
+ char outAccnosFilename[1024];
+ strcpy(outAccnosFilename, accnosFileName.c_str());
+
+ 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); outputTypes.clear(); return 0; }
+
+ if (pid == 0) { //you are the root process
+
+ MPIPos = m->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); outputTypes.clear(); 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); outputTypes.clear(); 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;
+ m->mothurRemove(accnosFileName);
+ }
+ }
+
+#else
+
+ vector<unsigned long long> positions;
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ positions = m->divideFile(candidateFileNames[s], processors);
+ for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(new linePair(positions[i], positions[(i+1)])); }
+ #else
+ if (processors == 1) {
+ lines.push_back(new linePair(0, 1000));
+ }else {
+ positions = m->setFilePosFasta(candidateFileNames[s], numFastaSeqs);
+ if (positions.size() < processors) { processors = positions.size(); }
+
+ //figure out how many sequences you have to process
+ int numSeqsPerProcessor = numFastaSeqs / processors;
+ for (int i = 0; i < processors; i++) {
+ int startIndex = i * numSeqsPerProcessor;
+ if(i == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; }
+ lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
+ }
+ }
+ #endif
+
+ if(processors == 1){
+ numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
+ }else{
+ numFastaSeqs = createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
+ }
+
+ if (m->control_pressed) { m->mothurRemove(accnosFileName); m->mothurRemove(alignFileName); m->mothurRemove(reportFileName); outputTypes.clear(); return 0; }
+
+ //delete accnos file if its blank else report to user
+ if (m->isBlank(accnosFileName)) { m->mothurRemove(accnosFileName); 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
+
+
+ #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); outputTypes["fasta"].push_back(alignFileName);
+ outputNames.push_back(reportFileName); outputTypes["alignreport"].push_back(reportFileName);
+ if (hasAccnos) { outputNames.push_back(accnosFileName); outputTypes["accnos"].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();
+ }
+
+ //set align file as new current fastafile
+ string currentFasta = "";
+ itTypes = outputTypes.find("fasta");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { currentFasta = (itTypes->second)[0]; m->setFastaFile(currentFasta); }
+ }
+
+ 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* filePos, string alignFName, string reportFName, string accnosFName, string filename){
+ try {
+ ofstream alignmentFile;
+ m->openOutputFile(alignFName, alignmentFile);
+
+ ofstream accnosFile;
+ m->openOutputFile(accnosFName, accnosFile);
+
+ NastReport report(reportFName);
+
+ ifstream inFASTA;
+ m->openInputFile(filename, inFASTA);
+
+ inFASTA.seekg(filePos->start);
+
+ bool done = false;
+ int count = 0;
+
+ //moved this into driver to avoid deep copies in windows paralellized version
+ Alignment* alignment;
+ 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);
+ }
+
+ while (!done) {
+
+ if (m->control_pressed) { break; }
+
+ Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
+ cout << candidateSeq->getAligned() << endl;
+ report.setCandidate(candidateSeq);
+
+ 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;
+ wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
+ }else{
+ wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
+ delete nast2;
+ delete copy;
+ }
+ }
+
+ //create accnos file with names
+ accnosFile << candidateSeq->getName() << wasBetter << endl;
+ }
+
+ 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; }
+
+ count++;
+ }
+ delete candidateSeq;
+
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ unsigned long long pos = inFASTA.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #else
+ if (inFASTA.eof()) { break; }
+ #endif
+
+ //report progress
+ if((count) % 100 == 0){ m->mothurOutJustToScreen(toString(count) + "\n"); }
+
+ }
+ //report progress
+ if((count) % 100 != 0){ m->mothurOutJustToScreen(toString(count) + "\n"); }
+
+ delete alignment;
+ alignmentFile.close();
+ inFASTA.close();
+ accnosFile.close();
+
+ return count;
+ }
+ 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<unsigned long long>& 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;
+ }
+
+ Alignment* alignment;
+ 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);
+ }
+
+
+ for(int i=0;i<num;i++){
+
+ if (m->control_pressed) { delete alignment; 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);
+ report.setCandidate(candidateSeq);
+
+ 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;
+ wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
+ }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.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 {
+ int num = 0;
+ processIDS.resize(0);
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ int process = 1;
+
+ //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){
+ num = driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename);
+
+ //pass numSeqs to parent
+ ofstream out;
+ string tempFile = alignFileName + toString(getpid()) + ".num.temp";
+ m->openOutputFile(tempFile, out);
+ out << num << endl;
+ out.close();
+
+ exit(0);
+ }else {
+ m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
+ for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
+ exit(0);
+ }
+ }
+
+ //do my part
+ num = driver(lines[0], alignFileName, reportFileName, accnosFName, filename);
+
+ //force parent to wait until all the processes are done
+ for (int i=0;i<processIDS.size();i++) {
+ int temp = processIDS[i];
+ wait(&temp);
+ }
+
+ vector<string> nonBlankAccnosFiles;
+ if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
+ else { m->mothurRemove(accnosFName); } //remove so other files can be renamed to it
+
+ for (int i = 0; i < processIDS.size(); i++) {
+ ifstream in;
+ string tempFile = alignFileName + toString(processIDS[i]) + ".num.temp";
+ m->openInputFile(tempFile, in);
+ if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
+ in.close(); m->mothurRemove(tempFile);
+
+ m->appendFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
+ m->mothurRemove((alignFileName + toString(processIDS[i]) + ".temp"));
+
+ appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);
+ m->mothurRemove((reportFileName + toString(processIDS[i]) + ".temp"));
+
+ if (!(m->isBlank(accnosFName + toString(processIDS[i]) + ".temp"))) {
+ nonBlankAccnosFiles.push_back(accnosFName + toString(processIDS[i]) + ".temp");
+ }else { m->mothurRemove((accnosFName + toString(processIDS[i]) + ".temp")); }
+
+ }
+
+ //append accnos files
+ if (nonBlankAccnosFiles.size() != 0) {
+ rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
+
+ for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
+ m->appendFiles(nonBlankAccnosFiles[h], accnosFName);
+ m->mothurRemove(nonBlankAccnosFiles[h]);
+ }
+ }else { //recreate the accnosfile if needed
+ ofstream out;
+ m->openOutputFile(accnosFName, out);
+ out.close();
+ }
+#else
+ //////////////////////////////////////////////////////////////////////////////////////////////////////
+ //Windows version shared memory, so be careful when passing variables through the alignData struct.
+ //Above fork() will clone, so memory is separate, but that's not the case with windows,
+ //////////////////////////////////////////////////////////////////////////////////////////////////////
+
+ vector<alignData*> pDataArray;
+ DWORD dwThreadIdArray[processors-1];
+ HANDLE hThreadArray[processors-1];
+
+ //Create processor worker threads.
+ for( int i=0; i<processors-1; i++ ){
+ //copy templateDb
+ //AlignmentDB* tempDB = new AlignmentDB(*templateDB);
+
+ // Allocate memory for thread data.
+ string extension = "";
+ if (i != 0) { extension = toString(i) + ".temp"; }
+
+ alignData* tempalign = new alignData(templateFileName, (alignFileName + extension), (reportFileName + extension), (accnosFName + extension), filename, align, search, kmerSize, m, lines[i]->start, lines[i]->end, flip, match, misMatch, gapOpen, gapExtend, threshold, i);
+ pDataArray.push_back(tempalign);
+ processIDS.push_back(i);
+
+ //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
+ //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
+ hThreadArray[i] = CreateThread(NULL, 0, MyAlignThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
+ }
+
+ //need to check for line ending error
+ ifstream inFASTA;
+ m->openInputFile(filename, inFASTA);
+ inFASTA.seekg(lines[processors-1]->start-1);
+ char c = inFASTA.peek();
+
+ if (c != '>') { //we need to move back
+ lines[processors-1]->start--;
+ }
+
+ //using the main process as a worker saves time and memory
+ //do my part - do last piece because windows is looking for eof
+ num = driver(lines[processors-1], (alignFileName + toString(processors-1) + ".temp"), (reportFileName + toString(processors-1) + ".temp"), (accnosFName + toString(processors-1) + ".temp"), filename);
+
+ //Wait until all threads have terminated.
+ WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
+
+ //Close all thread handles and free memory allocations.
+ for(int i=0; i < pDataArray.size(); i++){
+ if (pDataArray[i]->count != pDataArray[i]->end) {
+ m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
+ }
+ num += pDataArray[i]->count;
+ CloseHandle(hThreadArray[i]);
+ delete pDataArray[i];
+ }
+
+ vector<string> nonBlankAccnosFiles;
+ if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
+ else { m->mothurRemove(accnosFName); } //remove so other files can be renamed to it
+
+ for (int i = 1; i < processors; i++) {
+ m->appendFiles((alignFileName + toString(i) + ".temp"), alignFileName);
+ m->mothurRemove((alignFileName + toString(i) + ".temp"));
+
+ appendReportFiles((reportFileName + toString(i) + ".temp"), reportFileName);
+ m->mothurRemove((reportFileName + toString(i) + ".temp"));
+
+ if (!(m->isBlank(accnosFName + toString(i) + ".temp"))) {
+ nonBlankAccnosFiles.push_back(accnosFName + toString(i) + ".temp");
+ }else { m->mothurRemove((accnosFName + toString(i) + ".temp")); }
+ }
+
+ //append accnos files
+ if (nonBlankAccnosFiles.size() != 0) {
+ rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
+
+ for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
+ m->appendFiles(nonBlankAccnosFiles[h], accnosFName);
+ m->mothurRemove(nonBlankAccnosFiles[h]);
+ }
+ }else { //recreate the accnosfile if needed
+ ofstream out;
+ m->openOutputFile(accnosFName, out);
+ out.close();
+ }
+#endif
+
+ return num;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "createProcesses");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+
+void AlignCommand::appendReportFiles(string temp, string filename) {
+ try{
+
+ ofstream output;
+ ifstream input;
+ m->openOutputFileAppend(filename, output);
+ m->openInputFile(temp, input);
+
+ while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line
+
+ char buffer[4096];
+ while (!input.eof()) {
+ input.read(buffer, 4096);
+ output.write(buffer, input.gcount());
+ }
+
+ input.close();
+ output.close();
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "appendReportFiles");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
projects / mothur.git / blobdiff