aligncommand.cpp

   1 /*
   2  *  aligncommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/15/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  *      This version of nast does everything I think that the greengenes nast server does and then some.  I have added the
   9  *      feature of allowing users to define their database, kmer size for searching, alignment penalty values and alignment
  10  *      method.  This latter feature is perhaps most significant.  nastPlus enables a user to use either a Needleman-Wunsch
  11  *      (non-affine gap penalty) or Gotoh (affine gap penalty) pairwise alignment algorithm.  This is significant because it
  12  *      allows for a global alignment and not the local alignment provided by bLAst.  Furthermore, it has the potential to
  13  *      provide a better alignment because of the banding method employed by blast (I'm not sure about this).
  14  *
  15  */
  16
  17 #include "aligncommand.h"
  18 #include "referencedb.h"
  19
  20 //**********************************************************************************************************************
  21 vector<string> AlignCommand::setParameters(){
  22         try {
  23                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(ptemplate);
  24                 CommandParameter pcandidate("fasta", "InputTypes", "", "", "none", "none", "none","fasta-alignreport-accnos",false,true,true); parameters.push_back(pcandidate);
  25                 CommandParameter psearch("search", "Multiple", "kmer-blast-suffix", "kmer", "", "", "","",false,false,true); parameters.push_back(psearch);
  26                 CommandParameter pksize("ksize", "Number", "", "8", "", "", "","",false,false); parameters.push_back(pksize);
  27                 CommandParameter pmatch("match", "Number", "", "1.0", "", "", "","",false,false); parameters.push_back(pmatch);
  28                 CommandParameter palign("align", "Multiple", "needleman-gotoh-blast-noalign", "needleman", "", "", "","",false,false,true); parameters.push_back(palign);
  29                 CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "","",false,false); parameters.push_back(pmismatch);
  30                 CommandParameter pgapopen("gapopen", "Number", "", "-5.0", "", "", "","",false,false); parameters.push_back(pgapopen);
  31                 CommandParameter pgapextend("gapextend", "Number", "", "-2.0", "", "", "","",false,false); parameters.push_back(pgapextend);
  32                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  33                 CommandParameter pflip("flip", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pflip);
  34                 CommandParameter psave("save", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(psave);
  35                 CommandParameter pthreshold("threshold", "Number", "", "0.50", "", "", "","",false,false); parameters.push_back(pthreshold);
  36                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  37                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  38
  39                 vector<string> myArray;
  40                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  41                 return myArray;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "AlignCommand", "setParameters");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 string AlignCommand::getHelpString(){
  50         try {
  51                 string helpString = "";
  52                 helpString += "The align.seqs command reads a file containing sequences and creates an alignment file and a report file.";
  53                 helpString += "The align.seqs command parameters are reference, fasta, search, ksize, align, match, mismatch, gapopen, gapextend and processors.";
  54                 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.";
  55                 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.";
  56                 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.";
  57                 helpString += "The ksize parameter allows you to specify the kmer size for finding most similar template to candidate.  The default is 8.";
  58                 helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.";
  59                 helpString += "The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.";
  60                 helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -5.0.";
  61                 helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -2.0.";
  62                 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.";
  63                 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.";
  64                 helpString += "If the flip parameter is set to true the reverse complement of the sequence is aligned and the better alignment is reported.";
  65                 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.";
  66                 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.";
  67                 helpString += "The align.seqs command should be in the following format:";
  68                 helpString += "align.seqs(reference=yourTemplateFile, fasta=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty)";
  69                 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)";
  70                 helpString += "Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).";
  71                 return helpString;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "AlignCommand", "getHelpString");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 string AlignCommand::getOutputPattern(string type) {
  80     try {
  81         string pattern = "";
  82
  83         if (type == "fasta") {  pattern = "[filename],align"; } //makes file like: amazon.align
  84         else if (type == "alignreport") {  pattern = "[filename],align.report"; }
  85         else if (type == "accnos") {  pattern = "[filename],flip.accnos"; }
  86         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  87
  88         return pattern;
  89     }
  90     catch(exception& e) {
  91         m->errorOut(e, "AlignCommand", "getOutputPattern");
  92         exit(1);
  93     }
  94 }
  95 //**********************************************************************************************************************
  96 AlignCommand::AlignCommand(){
  97         try {
  98                 abort = true; calledHelp = true;
  99                 setParameters();
 100                 vector<string> tempOutNames;
 101                 outputTypes["fasta"] = tempOutNames;
 102                 outputTypes["alignreport"] = tempOutNames;
 103                 outputTypes["accnos"] = tempOutNames;
 104         }
 105         catch(exception& e) {
 106                 m->errorOut(e, "AlignCommand", "AlignCommand");
 107                 exit(1);
 108         }
 109 }
 110 //**********************************************************************************************************************
 111 AlignCommand::AlignCommand(string option)  {
 112         try {
 113                 abort = false; calledHelp = false;
 114                 ReferenceDB* rdb = ReferenceDB::getInstance();
 115
 116                 //allow user to run help
 117                 if(option == "help") { help(); abort = true; calledHelp = true;}
 118                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 119
 120                 else {
 121                         vector<string> myArray = setParameters();
 122
 123                         OptionParser parser(option);
 124                         map<string, string> parameters = parser.getParameters();
 125
 126                         ValidParameters validParameter("align.seqs");
 127                         map<string, string>::iterator it;
 128
 129                         //check to make sure all parameters are valid for command
 130                         for (it = parameters.begin(); it != parameters.end(); it++) {
 131                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 132                         }
 133
 134                         //initialize outputTypes
 135                         vector<string> tempOutNames;
 136                         outputTypes["fasta"] = tempOutNames;
 137                         outputTypes["alignreport"] = tempOutNames;
 138                         outputTypes["accnos"] = tempOutNames;
 139
 140                         //if the user changes the output directory command factory will send this info to us in the output parameter
 141                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 142
 143
 144                         //if the user changes the input directory command factory will send this info to us in the output parameter
 145                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 146
 147                         if (inputDir == "not found"){   inputDir = "";          }
 148                         else {
 149                                 string path;
 150
 151                                 it = parameters.find("reference");
 152
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 158                                 }
 159                         }
 160
 161                         candidateFileName = validParameter.validFile(parameters, "fasta", false);
 162                         if (candidateFileName == "not found") {
 163                                 //if there is a current fasta file, use it
 164                                 string filename = m->getFastaFile();
 165                                 if (filename != "") { candidateFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 166                                 else {  m->mothurOut("You have no current fastafile and the candidate parameter is required."); m->mothurOutEndLine(); abort = true; }
 167                         }else {
 168                                 m->splitAtDash(candidateFileName, candidateFileNames);
 169
 170                                 //go through files and make sure they are good, if not, then disregard them
 171                                 for (int i = 0; i < candidateFileNames.size(); i++) {
 172                                         //candidateFileNames[i] = m->getFullPathName(candidateFileNames[i]);
 173
 174                                         bool ignore = false;
 175                                         if (candidateFileNames[i] == "current") {
 176                                                 candidateFileNames[i] = m->getFastaFile();
 177                                                 if (candidateFileNames[i] != "") {  m->mothurOut("Using " + candidateFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 178                                                 else {
 179                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 180                                                         //erase from file list
 181                                                         candidateFileNames.erase(candidateFileNames.begin()+i);
 182                                                         i--;
 183                                                 }
 184                                         }
 185
 186                                         if (!ignore) {
 187
 188                                                 if (inputDir != "") {
 189                                                         string path = m->hasPath(candidateFileNames[i]);
 190                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 191                                                         if (path == "") {       candidateFileNames[i] = inputDir + candidateFileNames[i];               }
 192                                                 }
 193
 194                                                 int ableToOpen;
 195                                                 ifstream in;
 196                                                 ableToOpen = m->openInputFile(candidateFileNames[i], in, "noerror");
 197                                                 in.close();
 198
 199                                                 //if you can't open it, try default location
 200                                                 if (ableToOpen == 1) {
 201                                                         if (m->getDefaultPath() != "") { //default path is set
 202                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(candidateFileNames[i]);
 203                                                                 m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 204                                                                 ifstream in2;
 205                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 206                                                                 in2.close();
 207                                                                 candidateFileNames[i] = tryPath;
 208                                                         }
 209                                                 }
 210
 211                                                 //if you can't open it, try output location
 212                                                 if (ableToOpen == 1) {
 213                                                         if (m->getOutputDir() != "") { //default path is set
 214                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(candidateFileNames[i]);
 215                                                                 m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 216                                                                 ifstream in2;
 217                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 218                                                                 in2.close();
 219                                                                 candidateFileNames[i] = tryPath;
 220                                                         }
 221                                                 }
 222
 223
 224
 225                                                 if (ableToOpen == 1) {
 226                                                         m->mothurOut("Unable to open " + candidateFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 227                                                         //erase from file list
 228                                                         candidateFileNames.erase(candidateFileNames.begin()+i);
 229                                                         i--;
 230                                                 }else {
 231                                                         m->setFastaFile(candidateFileNames[i]);
 232                                                 }
 233                                         }
 234                                 }
 235
 236                                 //make sure there is at least one valid file left
 237                                 if (candidateFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 238                         }
 239
 240                         //check for optional parameter and set defaults
 241                         // ...at some point should added some additional type checking...
 242                         string temp;
 243                         temp = validParameter.validFile(parameters, "ksize", false);            if (temp == "not found"){       temp = "8";                             }
 244                         m->mothurConvert(temp, kmerSize);
 245
 246                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 247                         m->mothurConvert(temp, match);
 248
 249                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 250                         m->mothurConvert(temp, misMatch);
 251
 252                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-5.0";                  }
 253                         m->mothurConvert(temp, gapOpen);
 254
 255                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-2.0";                  }
 256                         m->mothurConvert(temp, gapExtend);
 257
 258                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 259                         m->setProcessors(temp);
 260                         m->mothurConvert(temp, processors);
 261
 262                         temp = validParameter.validFile(parameters, "flip", false);                     if (temp == "not found"){       temp = "f";                             }
 263                         flip = m->isTrue(temp);
 264
 265                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 266                         save = m->isTrue(temp);
 267                         rdb->save = save;
 268                         if (save) { //clear out old references
 269                                 rdb->clearMemory();
 270                         }
 271
 272                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 273                         templateFileName = validParameter.validFile(parameters, "reference", true);
 274                         if (templateFileName == "not found") {
 275                                 //check for saved reference sequences
 276                                 if (rdb->referenceSeqs.size() != 0) {
 277                                         templateFileName = "saved";
 278                                 }else {
 279                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the align.seqs command.");
 280                                         m->mothurOutEndLine();
 281                                         abort = true;
 282                                 }
 283                         }else if (templateFileName == "not open") { abort = true; }
 284                         else {  if (save) {     rdb->setSavedReference(templateFileName);       }       }
 285
 286                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found"){       temp = "0.50";                  }
 287                         m->mothurConvert(temp, threshold);
 288
 289                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
 290                         if ((search != "suffix") && (search != "kmer") && (search != "blast")) { m->mothurOut("invalid search option: choices are kmer, suffix or blast."); m->mothurOutEndLine(); abort=true; }
 291
 292                         align = validParameter.validFile(parameters, "align", false);           if (align == "not found"){      align = "needleman";    }
 293                         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; }
 294
 295                 }
 296
 297         }
 298         catch(exception& e) {
 299                 m->errorOut(e, "AlignCommand", "AlignCommand");
 300                 exit(1);
 301         }
 302 }
 303 //**********************************************************************************************************************
 304 AlignCommand::~AlignCommand(){
 305
 306         if (abort == false) {
 307                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 308                 delete templateDB;
 309         }
 310 }
 311 //**********************************************************************************************************************
 312
 313 int AlignCommand::execute(){
 314         try {
 315                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 316
 317                 templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, rand());
 318
 319                 for (int s = 0; s < candidateFileNames.size(); s++) {
 320                         if (m->control_pressed) { outputTypes.clear(); return 0; }
 321
 322                         m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();
 323
 324                         if (outputDir == "") {  outputDir += m->hasPath(candidateFileNames[s]); }
 325             map<string, string> variables; variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s]));
 326                         string alignFileName = getOutputFileName("fasta", variables);
 327                         string reportFileName = getOutputFileName("alignreport", variables);
 328                         string accnosFileName = getOutputFileName("accnos", variables);
 329
 330                         bool hasAccnos = true;
 331
 332                         int numFastaSeqs = 0;
 333                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 334                         int start = time(NULL);
 335
 336 #ifdef USE_MPI
 337                                 int pid, numSeqsPerProcessor;
 338                                 int tag = 2001;
 339                                 vector<unsigned long long> MPIPos;
 340                                 MPIWroteAccnos = false;
 341
 342                                 MPI_Status status;
 343                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 344                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 345
 346                                 MPI_File inMPI;
 347                                 MPI_File outMPIAlign;
 348                                 MPI_File outMPIReport;
 349                                 MPI_File outMPIAccnos;
 350
 351                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 352                                 int inMode=MPI_MODE_RDONLY;
 353
 354                                 char outAlignFilename[1024];
 355                                 strcpy(outAlignFilename, alignFileName.c_str());
 356
 357                                 char outReportFilename[1024];
 358                                 strcpy(outReportFilename, reportFileName.c_str());
 359
 360                                 char outAccnosFilename[1024];
 361                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 362
 363                                 char inFileName[1024];
 364                                 strcpy(inFileName, candidateFileNames[s].c_str());
 365
 366                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 367                                 MPI_File_open(MPI_COMM_WORLD, outAlignFilename, outMode, MPI_INFO_NULL, &outMPIAlign);
 368                                 MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport);
 369                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 370
 371                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIAlign);  MPI_File_close(&outMPIReport);  MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
 372
 373                                 if (pid == 0) { //you are the root process
 374
 375                                         MPIPos = m->setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
 376
 377                                         //send file positions to all processes
 378                                         for(int i = 1; i < processors; i++) {
 379                                                 MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 380                                                 MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 381                                         }
 382
 383                                         //figure out how many sequences you have to align
 384                                         numSeqsPerProcessor = numFastaSeqs / processors;
 385                                         int startIndex =  pid * numSeqsPerProcessor;
 386                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 387
 388                                         //align your part
 389                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);
 390
 391                                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIAlign);  MPI_File_close(&outMPIReport);  MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
 392
 393                                         for (int i = 1; i < processors; i++) {
 394                                                 bool tempResult;
 395                                                 MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 396                                                 if (tempResult != 0) { MPIWroteAccnos = true; }
 397                                         }
 398                                 }else{ //you are a child process
 399                                         MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 400                                         MPIPos.resize(numFastaSeqs+1);
 401                                         MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 402
 403
 404                                         //figure out how many sequences you have to align
 405                                         numSeqsPerProcessor = numFastaSeqs / processors;
 406                                         int startIndex =  pid * numSeqsPerProcessor;
 407                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 408
 409
 410                                         //align your part
 411                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);
 412
 413                                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIAlign);  MPI_File_close(&outMPIReport);  MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
 414
 415                                         MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 416                                 }
 417
 418                                 //close files
 419                                 MPI_File_close(&inMPI);
 420                                 MPI_File_close(&outMPIAlign);
 421                                 MPI_File_close(&outMPIReport);
 422                                 MPI_File_close(&outMPIAccnos);
 423
 424                                 //delete accnos file if blank
 425                                 if (pid == 0) {
 426                                         //delete accnos file if its blank else report to user
 427                                         if (MPIWroteAccnos) {
 428                                                 m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
 429                                                 if (!flip) {
 430                                                         m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well.");
 431                                                 }else{  m->mothurOut(" If the reverse compliment proved to be better it was reported.");  }
 432                                                 m->mothurOutEndLine();
 433                                         }else {
 434                                                 //MPI_Info info;
 435                                                 //MPI_File_delete(outAccnosFilename, info);
 436                                                 hasAccnos = false;
 437                                                 m->mothurRemove(accnosFileName);
 438                                         }
 439                                 }
 440
 441 #else
 442
 443                         vector<unsigned long long> positions;
 444                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 445                         positions = m->divideFile(candidateFileNames[s], processors);
 446                         for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(new linePair(positions[i], positions[(i+1)]));  }
 447                 #else
 448                         if (processors == 1) {
 449                                 lines.push_back(new linePair(0, 1000));
 450                         }else {
 451                                 positions = m->setFilePosFasta(candidateFileNames[s], numFastaSeqs);
 452                                 if (positions.size() < processors) { processors = positions.size(); }
 453
 454                                 //figure out how many sequences you have to process
 455                                 int numSeqsPerProcessor = numFastaSeqs / processors;
 456                                 for (int i = 0; i < processors; i++) {
 457                                         int startIndex =  i * numSeqsPerProcessor;
 458                                         if(i == (processors - 1)){      numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;   }
 459                                         lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
 460                                 }
 461                         }
 462                 #endif
 463
 464                         if(processors == 1){
 465                                 numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
 466                         }else{
 467                                 numFastaSeqs = createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
 468                         }
 469
 470                         if (m->control_pressed) { m->mothurRemove(accnosFileName); m->mothurRemove(alignFileName); m->mothurRemove(reportFileName); outputTypes.clear();  return 0; }
 471
 472                         //delete accnos file if its blank else report to user
 473                         if (m->isBlank(accnosFileName)) {  m->mothurRemove(accnosFileName);  hasAccnos = false; }
 474                         else {
 475                                 m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
 476                                 if (!flip) {
 477                                         m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well.");
 478                                 }else{  m->mothurOut(" If the reverse compliment proved to be better it was reported.");  }
 479                                 m->mothurOutEndLine();
 480                         }
 481
 482 #endif
 483
 484
 485                 #ifdef USE_MPI
 486                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 487
 488                         if (pid == 0) { //only one process should output to screen
 489                 #endif
 490
 491                         outputNames.push_back(alignFileName); outputTypes["fasta"].push_back(alignFileName);
 492                         outputNames.push_back(reportFileName); outputTypes["alignreport"].push_back(reportFileName);
 493                         if (hasAccnos)  {       outputNames.push_back(accnosFileName);  outputTypes["accnos"].push_back(accnosFileName);  }
 494
 495                 #ifdef USE_MPI
 496                         }
 497                 #endif
 498
 499                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences.");
 500                         m->mothurOutEndLine();
 501                         m->mothurOutEndLine();
 502                 }
 503
 504                 //set align file as new current fastafile
 505                 string currentFasta = "";
 506                 itTypes = outputTypes.find("fasta");
 507                 if (itTypes != outputTypes.end()) {
 508                         if ((itTypes->second).size() != 0) { currentFasta = (itTypes->second)[0]; m->setFastaFile(currentFasta); }
 509                 }
 510
 511                 m->mothurOutEndLine();
 512                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 513                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 514                 m->mothurOutEndLine();
 515
 516                 return 0;
 517         }
 518         catch(exception& e) {
 519                 m->errorOut(e, "AlignCommand", "execute");
 520                 exit(1);
 521         }
 522 }
 523
 524 //**********************************************************************************************************************
 525 int AlignCommand::driver(linePair* filePos, string alignFName, string reportFName, string accnosFName, string filename){
 526         try {
 527                 ofstream alignmentFile;
 528                 m->openOutputFile(alignFName, alignmentFile);
 529
 530                 ofstream accnosFile;
 531                 m->openOutputFile(accnosFName, accnosFile);
 532
 533                 NastReport report(reportFName);
 534
 535                 ifstream inFASTA;
 536                 m->openInputFile(filename, inFASTA);
 537
 538                 inFASTA.seekg(filePos->start);
 539
 540                 bool done = false;
 541                 int count = 0;
 542
 543                 //moved this into driver to avoid deep copies in windows paralellized version
 544                 Alignment* alignment;
 545                 int longestBase = templateDB->getLongestBase();
 546         if (m->debug) { m->mothurOut("[DEBUG]: template longest base = "  + toString(templateDB->getLongestBase()) + " \n"); }
 547                 if(align == "gotoh")                    {       alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase);                 }
 548                 else if(align == "needleman")   {       alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);                                }
 549                 else if(align == "blast")               {       alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch);            }
 550                 else if(align == "noalign")             {       alignment = new NoAlign();                                                                                                      }
 551                 else {
 552                         m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
 553                         m->mothurOutEndLine();
 554                         alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
 555                 }
 556
 557                 while (!done) {
 558
 559                         if (m->control_pressed) {  break; }
 560
 561                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 562                         report.setCandidate(candidateSeq);
 563
 564                         int origNumBases = candidateSeq->getNumBases();
 565                         string originalUnaligned = candidateSeq->getUnaligned();
 566                         int numBasesNeeded = origNumBases * threshold;
 567
 568                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 569                                 if (candidateSeq->getUnaligned().length()+1 > alignment->getnRows()) {
 570                     if (m->debug) { m->mothurOut("[DEBUG]: " + candidateSeq->getName() + " " + toString(candidateSeq->getUnaligned().length()) + " " + toString(alignment->getnRows()) + " \n"); }
 571                                         alignment->resize(candidateSeq->getUnaligned().length()+2);
 572                                 }
 573                                 Sequence temp = templateDB->findClosestSequence(candidateSeq);
 574                                 Sequence* templateSeq = new Sequence(temp.getName(), temp.getAligned());
 575
 576                                 float searchScore = templateDB->getSearchScore();
 577
 578                                 Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
 579
 580                                 Sequence* copy;
 581
 582                                 Nast* nast2;
 583                                 bool needToDeleteCopy = false;  //this is needed in case you have you enter the ifs below
 584                                                                                                 //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
 585                                                                                                 //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
 586                                                                                                 //so this bool tells you if you need to delete it
 587
 588                                 //if there is a possibility that this sequence should be reversed
 589                                 if (candidateSeq->getNumBases() < numBasesNeeded) {
 590
 591                                         string wasBetter =  "";
 592                                         //if the user wants you to try the reverse
 593                                         if (flip) {
 594
 595                                                 //get reverse compliment
 596                                                 copy = new Sequence(candidateSeq->getName(), originalUnaligned);
 597                                                 copy->reverseComplement();
 598
 599                         if (m->debug) { m->mothurOut("[DEBUG]: flipping "  + candidateSeq->getName() + " \n"); }
 600
 601                                                 //rerun alignment
 602                                                 Sequence temp2 = templateDB->findClosestSequence(copy);
 603                                                 Sequence* templateSeq2 = new Sequence(temp2.getName(), temp2.getAligned());
 604
 605                         if (m->debug) { m->mothurOut("[DEBUG]: closest template "  + temp2.getName() + " \n"); }
 606
 607                                                 searchScore = templateDB->getSearchScore();
 608
 609                                                 nast2 = new Nast(alignment, copy, templateSeq2);
 610
 611                         if (m->debug) { m->mothurOut("[DEBUG]: completed Nast2 "  + candidateSeq->getName() + " flipped numBases = " + toString(copy->getNumBases()) + " old numbases = " + toString(candidateSeq->getNumBases()) +" \n"); }
 612
 613                                                 //check if any better
 614                                                 if (copy->getNumBases() > candidateSeq->getNumBases()) {
 615                                                         candidateSeq->setAligned(copy->getAligned());  //use reverse compliments alignment since its better
 616                             delete templateSeq;
 617                                                         templateSeq = templateSeq2;
 618                                                         delete nast;
 619                                                         nast = nast2;
 620                                                         needToDeleteCopy = true;
 621                                                         wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
 622                                                 }else{
 623                                                         wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
 624                                                         delete nast2;
 625                             delete templateSeq2;
 626                                                         delete copy;
 627                                                 }
 628                         if (m->debug) { m->mothurOut("[DEBUG]: done.\n"); }
 629                                         }
 630
 631                                         //create accnos file with names
 632                                         accnosFile << candidateSeq->getName() << wasBetter << endl;
 633                                 }
 634
 635                                 report.setTemplate(templateSeq);
 636                                 report.setSearchParameters(search, searchScore);
 637                                 report.setAlignmentParameters(align, alignment);
 638                                 report.setNastParameters(*nast);
 639
 640                                 alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
 641
 642                                 report.print();
 643                                 delete nast;
 644                 delete templateSeq;
 645                                 if (needToDeleteCopy) {   delete copy;   }
 646
 647                                 count++;
 648                         }
 649                         delete candidateSeq;
 650
 651                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 652                                 unsigned long long pos = inFASTA.tellg();
 653                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 654                         #else
 655                                 if (inFASTA.eof()) { break; }
 656                         #endif
 657
 658                         //report progress
 659                         if((count) % 100 == 0){ m->mothurOutJustToScreen(toString(count) + "\n");               }
 660
 661                 }
 662                 //report progress
 663                 if((count) % 100 != 0){ m->mothurOutJustToScreen(toString(count) + "\n");               }
 664
 665                 delete alignment;
 666                 alignmentFile.close();
 667                 inFASTA.close();
 668                 accnosFile.close();
 669
 670                 return count;
 671         }
 672         catch(exception& e) {
 673                 m->errorOut(e, "AlignCommand", "driver");
 674                 exit(1);
 675         }
 676 }
 677 //**********************************************************************************************************************
 678 #ifdef USE_MPI
 679 int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector<unsigned long long>& MPIPos){
 680         try {
 681                 string outputString = "";
 682                 MPI_Status statusReport;
 683                 MPI_Status statusAlign;
 684                 MPI_Status statusAccnos;
 685                 MPI_Status status;
 686                 int pid;
 687                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 688
 689                 NastReport report;
 690
 691                 if (pid == 0) {
 692                         outputString = report.getHeaders();
 693                         int length = outputString.length();
 694
 695                         char* buf = new char[length];
 696                         memcpy(buf, outputString.c_str(), length);
 697
 698                         MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport);
 699
 700             delete buf;
 701                 }
 702
 703                 Alignment* alignment;
 704                 int longestBase = templateDB->getLongestBase();
 705                 if(align == "gotoh")                    {       alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase);                 }
 706                 else if(align == "needleman")   {       alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);                                }
 707                 else if(align == "blast")               {       alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch);            }
 708                 else if(align == "noalign")             {       alignment = new NoAlign();                                                                                                      }
 709                 else {
 710                         m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
 711                         m->mothurOutEndLine();
 712                         alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
 713                 }
 714
 715
 716                 for(int i=0;i<num;i++){
 717
 718                         if (m->control_pressed) { delete alignment; return 0; }
 719
 720                         //read next sequence
 721                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 722
 723                         char* buf4 = new char[length];
 724                         //memcpy(buf4, outputString.c_str(), length);
 725
 726                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 727
 728                         string tempBuf = buf4;
 729
 730                         delete buf4;
 731
 732                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 733
 734                         istringstream iss (tempBuf,istringstream::in);
 735
 736                         Sequence* candidateSeq = new Sequence(iss);
 737                         report.setCandidate(candidateSeq);
 738
 739                         int origNumBases = candidateSeq->getNumBases();
 740                         string originalUnaligned = candidateSeq->getUnaligned();
 741                         int numBasesNeeded = origNumBases * threshold;
 742
 743                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 744                                 if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
 745                                         alignment->resize(candidateSeq->getUnaligned().length()+1);
 746                                 }
 747
 748                                 Sequence temp = templateDB->findClosestSequence(candidateSeq);
 749                                 Sequence* templateSeq = new Sequence(temp.getName(), temp.getAligned());
 750
 751                                 float searchScore = templateDB->getSearchScore();
 752
 753                                 Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
 754                                 Sequence* copy;
 755
 756                                 Nast* nast2;
 757                                 bool needToDeleteCopy = false;  //this is needed in case you have you enter the ifs below
 758                                                                                                 //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
 759                                                                                                 //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
 760                                                                                                 //so this bool tells you if you need to delete it
 761
 762                                 //if there is a possibility that this sequence should be reversed
 763                                 if (candidateSeq->getNumBases() < numBasesNeeded) {
 764
 765                                         string wasBetter = "";
 766                                         //if the user wants you to try the reverse
 767                                         if (flip) {
 768                                                 //get reverse compliment
 769                                                 copy = new Sequence(candidateSeq->getName(), originalUnaligned);
 770                                                 copy->reverseComplement();
 771
 772                                                 //rerun alignment
 773                                                 Sequence temp2 = templateDB->findClosestSequence(copy);
 774                                                 Sequence* templateSeq2 = new Sequence(temp2.getName(), temp2.getAligned());
 775
 776                                                 searchScore = templateDB->getSearchScore();
 777
 778                                                 nast2 = new Nast(alignment, copy, templateSeq2);
 779
 780                                                 //check if any better
 781                                                 if (copy->getNumBases() > candidateSeq->getNumBases()) {
 782                                                         candidateSeq->setAligned(copy->getAligned());  //use reverse compliments alignment since its better
 783                                                         delete templateSeq;
 784                                                         templateSeq = templateSeq2;
 785                                                         delete nast;
 786                                                         nast = nast2;
 787                                                         needToDeleteCopy = true;
 788                                                         wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
 789                                                 }else{
 790                                                         wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";
 791                                                         delete nast2;
 792                             delete templateSeq2;
 793                                                         delete copy;
 794                                                 }
 795                                         }
 796
 797                                         //create accnos file with names
 798                                         outputString = candidateSeq->getName() + wasBetter + "\n";
 799
 800                                         //send results to parent
 801                                         int length = outputString.length();
 802
 803                                         char* buf = new char[length];
 804                                         memcpy(buf, outputString.c_str(), length);
 805
 806                                         MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos);
 807                                         delete buf;
 808                                         MPIWroteAccnos = true;
 809                                 }
 810
 811                                 report.setTemplate(templateSeq);
 812                                 report.setSearchParameters(search, searchScore);
 813                                 report.setAlignmentParameters(align, alignment);
 814                                 report.setNastParameters(*nast);
 815
 816                                 outputString =  ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n";
 817
 818                                 //send results to parent
 819                                 int length = outputString.length();
 820                                 char* buf2 = new char[length];
 821                                 memcpy(buf2, outputString.c_str(), length);
 822
 823                                 MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign);
 824
 825                                 delete buf2;
 826
 827                                 outputString = report.getReport();
 828
 829                                 //send results to parent
 830                                 length = outputString.length();
 831                                 char* buf3 = new char[length];
 832                                 memcpy(buf3, outputString.c_str(), length);
 833
 834                                 MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport);
 835
 836                                 delete buf3;
 837                                 delete nast;
 838                 delete templateSeq;
 839                                 if (needToDeleteCopy) {   delete copy;   }
 840                         }
 841                         delete candidateSeq;
 842
 843                         //report progress
 844                         if((i+1) % 100 == 0){   cout << (toString(i+1)) << endl;                }
 845                 }
 846                 //report progress
 847                 if((num) % 100 != 0){   cout << (toString(num)) << endl;                }
 848
 849                 return 1;
 850         }
 851         catch(exception& e) {
 852                 m->errorOut(e, "AlignCommand", "driverMPI");
 853                 exit(1);
 854         }
 855 }
 856 #endif
 857 /**************************************************************************************************/
 858
 859 int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) {
 860         try {
 861                 int num = 0;
 862                 processIDS.resize(0);
 863         bool recalc = false;
 864
 865 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 866                 int process = 1;
 867
 868                 //loop through and create all the processes you want
 869                 while (process != processors) {
 870                         pid_t pid = fork();
 871
 872                         if (pid > 0) {
 873                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 874                                 process++;
 875                         }else if (pid == 0){
 876                                 num = driver(lines[process], alignFileName + toString(m->mothurGetpid(process)) + ".temp", reportFileName + toString(m->mothurGetpid(process)) + ".temp", accnosFName + m->mothurGetpid(process) + ".temp", filename);
 877
 878                                 //pass numSeqs to parent
 879                                 ofstream out;
 880                                 string tempFile = alignFileName + toString(m->mothurGetpid(process)) + ".num.temp";
 881                                 m->openOutputFile(tempFile, out);
 882                                 out << num << endl;
 883                                 out.close();
 884
 885                                 exit(0);
 886                         }else {
 887                                 m->mothurOut("[ERROR]: unable to spawn the number of processes you requested, reducing number to " + toString(process) + "\n"); processors = process;
 888                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 889                 recalc = true;
 890                                 break;
 891                         }
 892                 }
 893
 894         if (recalc) {
 895             for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 896             vector<unsigned long long> positions;
 897                         positions = m->divideFile(filename, processors);
 898                         for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(new linePair(positions[i], positions[(i+1)]));  }
 899
 900             num = 0;
 901             processIDS.resize(0);
 902             process = 1;
 903
 904             while (process != processors) {
 905                 pid_t pid = fork();
 906
 907                 if (pid > 0) {
 908                     processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 909                     process++;
 910                 }else if (pid == 0){
 911                     num = driver(lines[process], alignFileName + toString(m->mothurGetpid(process)) + ".temp", reportFileName + toString(m->mothurGetpid(process)) + ".temp", accnosFName + m->mothurGetpid(process) + ".temp", filename);
 912
 913                     //pass numSeqs to parent
 914                     ofstream out;
 915                     string tempFile = alignFileName + toString(m->mothurGetpid(process)) + ".num.temp";
 916                     m->openOutputFile(tempFile, out);
 917                     out << num << endl;
 918                     out.close();
 919
 920                     exit(0);
 921                 }else {
 922                     m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 923                     for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 924                     exit(0);
 925                 }
 926             }
 927         }
 928
 929                 //do my part
 930                 num = driver(lines[0], alignFileName, reportFileName, accnosFName, filename);
 931
 932                 //force parent to wait until all the processes are done
 933                 for (int i=0;i<processIDS.size();i++) {
 934                         int temp = processIDS[i];
 935                         wait(&temp);
 936                 }
 937
 938                 vector<string> nonBlankAccnosFiles;
 939                 if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
 940                 else { m->mothurRemove(accnosFName); } //remove so other files can be renamed to it
 941
 942                 for (int i = 0; i < processIDS.size(); i++) {
 943                         ifstream in;
 944                         string tempFile =  alignFileName + toString(processIDS[i]) + ".num.temp";
 945                         m->openInputFile(tempFile, in);
 946                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 947                         in.close(); m->mothurRemove(tempFile);
 948
 949                         m->appendFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
 950                         m->mothurRemove((alignFileName + toString(processIDS[i]) + ".temp"));
 951
 952                         appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);
 953                         m->mothurRemove((reportFileName + toString(processIDS[i]) + ".temp"));
 954
 955                         if (!(m->isBlank(accnosFName + toString(processIDS[i]) + ".temp"))) {
 956                                 nonBlankAccnosFiles.push_back(accnosFName + toString(processIDS[i]) + ".temp");
 957                         }else { m->mothurRemove((accnosFName + toString(processIDS[i]) + ".temp"));  }
 958
 959                 }
 960
 961                 //append accnos files
 962                 if (nonBlankAccnosFiles.size() != 0) {
 963                         rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
 964
 965                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
 966                                 m->appendFiles(nonBlankAccnosFiles[h], accnosFName);
 967                                 m->mothurRemove(nonBlankAccnosFiles[h]);
 968                         }
 969                 }else { //recreate the accnosfile if needed
 970                         ofstream out;
 971                         m->openOutputFile(accnosFName, out);
 972                         out.close();
 973                 }
 974 #else
 975                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 976                 //Windows version shared memory, so be careful when passing variables through the alignData struct.
 977                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 978                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 979
 980                 vector<alignData*> pDataArray;
 981                 DWORD   dwThreadIdArray[processors-1];
 982                 HANDLE  hThreadArray[processors-1];
 983
 984                 //Create processor worker threads.
 985                 for( int i=0; i<processors-1; i++ ){
 986                         //copy templateDb
 987                         //AlignmentDB* tempDB = new AlignmentDB(*templateDB);
 988
 989                         // Allocate memory for thread data.
 990                         string extension = "";
 991                         if (i != 0) { extension = toString(i) + ".temp"; }
 992
 993                         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);
 994                         pDataArray.push_back(tempalign);
 995                         processIDS.push_back(i);
 996
 997                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
 998                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 999                         hThreadArray[i] = CreateThread(NULL, 0, MyAlignThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
1000                 }
1001
1002                 //need to check for line ending error
1003                 ifstream inFASTA;
1004                 m->openInputFile(filename, inFASTA);
1005                 inFASTA.seekg(lines[processors-1]->start-1);
1006                 char c = inFASTA.peek();
1007
1008                 if (c != '>') { //we need to move back
1009                         lines[processors-1]->start--;
1010                 }
1011
1012                 //using the main process as a worker saves time and memory
1013                 //do my part - do last piece because windows is looking for eof
1014                 num = driver(lines[processors-1], (alignFileName + toString(processors-1) + ".temp"), (reportFileName + toString(processors-1) + ".temp"), (accnosFName + toString(processors-1) + ".temp"), filename);
1015
1016                 //Wait until all threads have terminated.
1017                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
1018
1019                 //Close all thread handles and free memory allocations.
1020                 for(int i=0; i < pDataArray.size(); i++){
1021             if (pDataArray[i]->count != pDataArray[i]->end) {
1022                 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;
1023             }
1024                         num += pDataArray[i]->count;
1025                         CloseHandle(hThreadArray[i]);
1026                         delete pDataArray[i];
1027                 }
1028
1029                 vector<string> nonBlankAccnosFiles;
1030                 if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
1031                 else { m->mothurRemove(accnosFName); } //remove so other files can be renamed to it
1032
1033                 for (int i = 1; i < processors; i++) {
1034                         m->appendFiles((alignFileName + toString(i) + ".temp"), alignFileName);
1035                         m->mothurRemove((alignFileName + toString(i) + ".temp"));
1036
1037                         appendReportFiles((reportFileName + toString(i) + ".temp"), reportFileName);
1038                         m->mothurRemove((reportFileName + toString(i) + ".temp"));
1039
1040                         if (!(m->isBlank(accnosFName + toString(i) + ".temp"))) {
1041                                 nonBlankAccnosFiles.push_back(accnosFName + toString(i) + ".temp");
1042                         }else { m->mothurRemove((accnosFName + toString(i) + ".temp"));  }
1043                 }
1044
1045                 //append accnos files
1046                 if (nonBlankAccnosFiles.size() != 0) {
1047                         rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
1048
1049                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
1050                                 m->appendFiles(nonBlankAccnosFiles[h], accnosFName);
1051                                 m->mothurRemove(nonBlankAccnosFiles[h]);
1052                         }
1053                 }else { //recreate the accnosfile if needed
1054                         ofstream out;
1055                         m->openOutputFile(accnosFName, out);
1056                         out.close();
1057                 }
1058 #endif
1059
1060                 return num;
1061         }
1062         catch(exception& e) {
1063                 m->errorOut(e, "AlignCommand", "createProcesses");
1064                 exit(1);
1065         }
1066 }
1067 //**********************************************************************************************************************
1068
1069 void AlignCommand::appendReportFiles(string temp, string filename) {
1070         try{
1071
1072                 ofstream output;
1073                 ifstream input;
1074                 m->openOutputFileAppend(filename, output);
1075                 m->openInputFile(temp, input);
1076
1077                 while (!input.eof())    {       char c = input.get(); if (c == 10 || c == 13){  break;  }       } // get header line
1078
1079         char buffer[4096];
1080         while (!input.eof()) {
1081             input.read(buffer, 4096);
1082             output.write(buffer, input.gcount());
1083         }
1084
1085                 input.close();
1086                 output.close();
1087         }
1088         catch(exception& e) {
1089                 m->errorOut(e, "AlignCommand", "appendReportFiles");
1090                 exit(1);
1091         }
1092 }
1093 //**********************************************************************************************************************