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 "sequence.hpp"
  19
  20 #include "gotohoverlap.hpp"
  21 #include "needlemanoverlap.hpp"
  22 #include "blastalign.hpp"
  23 #include "noalign.hpp"
  24
  25 #include "nast.hpp"
  26 #include "nastreport.hpp"
  27
  28
  29 //**********************************************************************************************************************
  30 vector<string> AlignCommand::getValidParameters(){
  31         try {
  32                 string AlignArray[] =  {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors","flip","threshold","outputdir","inputdir"};
  33                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "AlignCommand", "getValidParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 vector<string> AlignCommand::getRequiredParameters(){
  43         try {
  44                 string AlignArray[] =  {"template","candidate"};
  45                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  46                 return myArray;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "AlignCommand", "getRequiredParameters");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 vector<string> AlignCommand::getRequiredFiles(){
  55         try {
  56                 vector<string> myArray;
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "AlignCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 AlignCommand::AlignCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 vector<string> tempOutNames;
  69                 outputTypes["fasta"] = tempOutNames;
  70                 outputTypes["alignreport"] = tempOutNames;
  71                 outputTypes["accnos"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "AlignCommand", "AlignCommand");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 AlignCommand::AlignCommand(string option)  {
  80         try {
  81                 abort = false; calledHelp = false;
  82
  83                 //allow user to run help
  84                 if(option == "help") { help(); abort = true; calledHelp = true;}
  85
  86                 else {
  87
  88                         //valid paramters for this command
  89                         string AlignArray[] =  {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors","flip","threshold","outputdir","inputdir"};
  90                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  91
  92                         OptionParser parser(option);
  93                         map<string, string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter("align.seqs");
  96                         map<string, string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["fasta"] = tempOutNames;
 106                         outputTypes["alignreport"] = tempOutNames;
 107                         outputTypes["accnos"] = tempOutNames;
 108
 109                         //if the user changes the output directory command factory will send this info to us in the output parameter
 110                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 111
 112
 113                         //if the user changes the input directory command factory will send this info to us in the output parameter
 114                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 115
 116                         if (inputDir == "not found"){   inputDir = "";          }
 117                         else {
 118                                 string path;
 119
 120                                 it = parameters.find("template");
 121
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
 127                                 }
 128                         }
 129
 130                         //check for required parameters
 131                         templateFileName = validParameter.validFile(parameters, "template", true);
 132
 133                         if (templateFileName == "not found") {
 134                                 m->mothurOut("template is a required parameter for the align.seqs command.");
 135                                 m->mothurOutEndLine();
 136                                 abort = true;
 137                         }else if (templateFileName == "not open") { abort = true; }
 138
 139                         candidateFileName = validParameter.validFile(parameters, "candidate", false);
 140                         if (candidateFileName == "not found") { m->mothurOut("candidate is a required parameter for the align.seqs command."); m->mothurOutEndLine(); abort = true;  }
 141                         else {
 142                                 m->splitAtDash(candidateFileName, candidateFileNames);
 143
 144                                 //go through files and make sure they are good, if not, then disregard them
 145                                 for (int i = 0; i < candidateFileNames.size(); i++) {
 146                                         //candidateFileNames[i] = m->getFullPathName(candidateFileNames[i]);
 147
 148                                         if (inputDir != "") {
 149                                                 string path = m->hasPath(candidateFileNames[i]);
 150                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 151                                                 if (path == "") {       candidateFileNames[i] = inputDir + candidateFileNames[i];               }
 152                                         }
 153
 154                                         int ableToOpen;
 155                                         ifstream in;
 156                                         ableToOpen = m->openInputFile(candidateFileNames[i], in, "noerror");
 157                                         in.close();
 158
 159                                         //if you can't open it, try default location
 160                                         if (ableToOpen == 1) {
 161                                                 if (m->getDefaultPath() != "") { //default path is set
 162                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(candidateFileNames[i]);
 163                                                         m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 164                                                         ifstream in2;
 165                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 166                                                         in2.close();
 167                                                         candidateFileNames[i] = tryPath;
 168                                                 }
 169                                         }
 170
 171                                         //if you can't open it, try default location
 172                                         if (ableToOpen == 1) {
 173                                                 if (m->getOutputDir() != "") { //default path is set
 174                                                         string tryPath = m->getOutputDir() + m->getSimpleName(candidateFileNames[i]);
 175                                                         m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 176                                                         ifstream in2;
 177                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 178                                                         in2.close();
 179                                                         candidateFileNames[i] = tryPath;
 180                                                 }
 181                                         }
 182
 183
 184
 185                                         if (ableToOpen == 1) {
 186                                                 m->mothurOut("Unable to open " + candidateFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 187                                                 //erase from file list
 188                                                 candidateFileNames.erase(candidateFileNames.begin()+i);
 189                                                 i--;
 190                                         }
 191
 192                                 }
 193
 194                                 //make sure there is at least one valid file left
 195                                 if (candidateFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 196                         }
 197
 198                         //check for optional parameter and set defaults
 199                         // ...at some point should added some additional type checking...
 200                         string temp;
 201                         temp = validParameter.validFile(parameters, "ksize", false);            if (temp == "not found"){       temp = "8";                             }
 202                         convert(temp, kmerSize);
 203
 204                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 205                         convert(temp, match);
 206
 207                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 208                         convert(temp, misMatch);
 209
 210                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 211                         convert(temp, gapOpen);
 212
 213                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 214                         convert(temp, gapExtend);
 215
 216                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 217                         convert(temp, processors);
 218
 219                         temp = validParameter.validFile(parameters, "flip", false);                     if (temp == "not found"){       temp = "f";                             }
 220                         flip = m->isTrue(temp);
 221
 222                         temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found"){       temp = "0.50";                  }
 223                         convert(temp, threshold);
 224
 225                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
 226
 227                         align = validParameter.validFile(parameters, "align", false);           if (align == "not found"){      align = "needleman";    }
 228                 }
 229
 230         }
 231         catch(exception& e) {
 232                 m->errorOut(e, "AlignCommand", "AlignCommand");
 233                 exit(1);
 234         }
 235 }
 236 //**********************************************************************************************************************
 237
 238 AlignCommand::~AlignCommand(){
 239
 240         if (abort == false) {
 241                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 242                 delete templateDB;
 243                 delete alignment;
 244         }
 245 }
 246
 247 //**********************************************************************************************************************
 248
 249 void AlignCommand::help(){
 250         try {
 251                 m->mothurOut("The align.seqs command reads a file containing sequences and creates an alignment file and a report file.\n");
 252                 m->mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen, gapextend and processors.\n");
 253                 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");
 254                 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");
 255                 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");
 256                 m->mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate.  The default is 8.\n");
 257                 m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n");
 258                 m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n");
 259                 m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n");
 260                 m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n");
 261                 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");
 262                 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");
 263                 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");
 264                 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");
 265                 m->mothurOut("The align.seqs command should be in the following format: \n");
 266                 m->mothurOut("align.seqs(template=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) \n");
 267                 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");
 268                 m->mothurOut("Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).\n\n");
 269         }
 270         catch(exception& e) {
 271                 m->errorOut(e, "AlignCommand", "help");
 272                 exit(1);
 273         }
 274 }
 275
 276
 277 //**********************************************************************************************************************
 278
 279 int AlignCommand::execute(){
 280         try {
 281                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 282
 283                 templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch);
 284                 int longestBase = templateDB->getLongestBase();
 285
 286                 if(align == "gotoh")                    {       alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase);                 }
 287                 else if(align == "needleman")   {       alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);                                }
 288                 else if(align == "blast")               {       alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch);            }
 289                 else if(align == "noalign")             {       alignment = new NoAlign();                                                                                                      }
 290                 else {
 291                         m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
 292                         m->mothurOutEndLine();
 293                         alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
 294                 }
 295
 296                 for (int s = 0; s < candidateFileNames.size(); s++) {
 297                         if (m->control_pressed) { outputTypes.clear(); return 0; }
 298
 299                         m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();
 300
 301                         if (outputDir == "") {  outputDir += m->hasPath(candidateFileNames[s]); }
 302                         string alignFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + "align";
 303                         string reportFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + "align.report";
 304                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + "flip.accnos";
 305                         bool hasAccnos = true;
 306
 307                         int numFastaSeqs = 0;
 308                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 309                         int start = time(NULL);
 310
 311 #ifdef USE_MPI
 312                                 int pid, numSeqsPerProcessor;
 313                                 int tag = 2001;
 314                                 vector<unsigned long int> MPIPos;
 315                                 MPIWroteAccnos = false;
 316
 317                                 MPI_Status status;
 318                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 319                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 320
 321                                 MPI_File inMPI;
 322                                 MPI_File outMPIAlign;
 323                                 MPI_File outMPIReport;
 324                                 MPI_File outMPIAccnos;
 325
 326                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 327                                 int inMode=MPI_MODE_RDONLY;
 328
 329                                 char outAlignFilename[1024];
 330                                 strcpy(outAlignFilename, alignFileName.c_str());
 331
 332                                 char outReportFilename[1024];
 333                                 strcpy(outReportFilename, reportFileName.c_str());
 334
 335                                 char outAccnosFilename[1024];
 336                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 337
 338                                 char inFileName[1024];
 339                                 strcpy(inFileName, candidateFileNames[s].c_str());
 340
 341                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 342                                 MPI_File_open(MPI_COMM_WORLD, outAlignFilename, outMode, MPI_INFO_NULL, &outMPIAlign);
 343                                 MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport);
 344                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 345
 346                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIAlign);  MPI_File_close(&outMPIReport);  MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
 347
 348                                 if (pid == 0) { //you are the root process
 349
 350                                         MPIPos = m->setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
 351
 352                                         //send file positions to all processes
 353                                         for(int i = 1; i < processors; i++) {
 354                                                 MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 355                                                 MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 356                                         }
 357
 358                                         //figure out how many sequences you have to align
 359                                         numSeqsPerProcessor = numFastaSeqs / processors;
 360                                         int startIndex =  pid * numSeqsPerProcessor;
 361                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 362
 363                                         //align your part
 364                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);
 365
 366                                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIAlign);  MPI_File_close(&outMPIReport);  MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
 367
 368                                         for (int i = 1; i < processors; i++) {
 369                                                 bool tempResult;
 370                                                 MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 371                                                 if (tempResult != 0) { MPIWroteAccnos = true; }
 372                                         }
 373                                 }else{ //you are a child process
 374                                         MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 375                                         MPIPos.resize(numFastaSeqs+1);
 376                                         MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 377
 378
 379                                         //figure out how many sequences you have to align
 380                                         numSeqsPerProcessor = numFastaSeqs / processors;
 381                                         int startIndex =  pid * numSeqsPerProcessor;
 382                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 383
 384
 385                                         //align your part
 386                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);
 387
 388                                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIAlign);  MPI_File_close(&outMPIReport);  MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
 389
 390                                         MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 391                                 }
 392
 393                                 //close files
 394                                 MPI_File_close(&inMPI);
 395                                 MPI_File_close(&outMPIAlign);
 396                                 MPI_File_close(&outMPIReport);
 397                                 MPI_File_close(&outMPIAccnos);
 398
 399                                 //delete accnos file if blank
 400                                 if (pid == 0) {
 401                                         //delete accnos file if its blank else report to user
 402                                         if (MPIWroteAccnos) {
 403                                                 m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
 404                                                 if (!flip) {
 405                                                         m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well.");
 406                                                 }else{  m->mothurOut(" If the reverse compliment proved to be better it was reported.");  }
 407                                                 m->mothurOutEndLine();
 408                                         }else {
 409                                                 //MPI_Info info;
 410                                                 //MPI_File_delete(outAccnosFilename, info);
 411                                                 hasAccnos = false;
 412                                                 remove(accnosFileName.c_str());
 413                                         }
 414                                 }
 415
 416 #else
 417
 418                 vector<unsigned long int> positions = m->divideFile(candidateFileNames[s], processors);
 419                 for (int i = 0; i < (positions.size()-1); i++) {
 420                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 421                 }
 422         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 423                         if(processors == 1){
 424                                 numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
 425                         }else{
 426                                 numFastaSeqs = createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
 427                         }
 428         #else
 429                         numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
 430         #endif
 431
 432                         if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); outputTypes.clear();  return 0; }
 433
 434                         //delete accnos file if its blank else report to user
 435                         if (m->isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 436                         else {
 437                                 m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
 438                                 if (!flip) {
 439                                         m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well.");
 440                                 }else{  m->mothurOut(" If the reverse compliment proved to be better it was reported.");  }
 441                                 m->mothurOutEndLine();
 442                         }
 443
 444 #endif
 445
 446
 447                 #ifdef USE_MPI
 448                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 449
 450                         if (pid == 0) { //only one process should output to screen
 451                 #endif
 452
 453                         outputNames.push_back(alignFileName); outputTypes["fasta"].push_back(alignFileName);
 454                         outputNames.push_back(reportFileName); outputTypes["alignreport"].push_back(reportFileName);
 455                         if (hasAccnos)  {       outputNames.push_back(accnosFileName);  outputTypes["accnos"].push_back(accnosFileName);  }
 456
 457                 #ifdef USE_MPI
 458                         }
 459                 #endif
 460
 461                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences.");
 462                         m->mothurOutEndLine();
 463                         m->mothurOutEndLine();
 464                 }
 465
 466
 467                 m->mothurOutEndLine();
 468                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 469                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 470                 m->mothurOutEndLine();
 471
 472                 return 0;
 473         }
 474         catch(exception& e) {
 475                 m->errorOut(e, "AlignCommand", "execute");
 476                 exit(1);
 477         }
 478 }
 479
 480 //**********************************************************************************************************************
 481
 482 int AlignCommand::driver(linePair* filePos, string alignFName, string reportFName, string accnosFName, string filename){
 483         try {
 484                 ofstream alignmentFile;
 485                 m->openOutputFile(alignFName, alignmentFile);
 486
 487                 ofstream accnosFile;
 488                 m->openOutputFile(accnosFName, accnosFile);
 489
 490                 NastReport report(reportFName);
 491
 492                 ifstream inFASTA;
 493                 m->openInputFile(filename, inFASTA);
 494
 495                 inFASTA.seekg(filePos->start);
 496
 497                 bool done = false;
 498                 int count = 0;
 499
 500                 while (!done) {
 501
 502                         if (m->control_pressed) {  return 0; }
 503
 504                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 505                         report.setCandidate(candidateSeq);
 506
 507                         int origNumBases = candidateSeq->getNumBases();
 508                         string originalUnaligned = candidateSeq->getUnaligned();
 509                         int numBasesNeeded = origNumBases * threshold;
 510
 511                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 512                                 if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
 513                                         alignment->resize(candidateSeq->getUnaligned().length()+1);
 514                                 }
 515
 516                                 Sequence temp = templateDB->findClosestSequence(candidateSeq);
 517                                 Sequence* templateSeq = &temp;
 518
 519                                 float searchScore = templateDB->getSearchScore();
 520
 521                                 Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
 522
 523                                 Sequence* copy;
 524
 525                                 Nast* nast2;
 526                                 bool needToDeleteCopy = false;  //this is needed in case you have you enter the ifs below
 527                                                                                                 //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
 528                                                                                                 //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
 529                                                                                                 //so this bool tells you if you need to delete it
 530
 531                                 //if there is a possibility that this sequence should be reversed
 532                                 if (candidateSeq->getNumBases() < numBasesNeeded) {
 533
 534                                         string wasBetter =  "";
 535                                         //if the user wants you to try the reverse
 536                                         if (flip) {
 537
 538                                                 //get reverse compliment
 539                                                 copy = new Sequence(candidateSeq->getName(), originalUnaligned);
 540                                                 copy->reverseComplement();
 541
 542                                                 //rerun alignment
 543                                                 Sequence temp2 = templateDB->findClosestSequence(copy);
 544                                                 Sequence* templateSeq2 = &temp2;
 545
 546                                                 searchScore = templateDB->getSearchScore();
 547
 548                                                 nast2 = new Nast(alignment, copy, templateSeq2);
 549
 550                                                 //check if any better
 551                                                 if (copy->getNumBases() > candidateSeq->getNumBases()) {
 552                                                         candidateSeq->setAligned(copy->getAligned());  //use reverse compliments alignment since its better
 553                                                         templateSeq = templateSeq2;
 554                                                         delete nast;
 555                                                         nast = nast2;
 556                                                         needToDeleteCopy = true;
 557                                                         wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
 558                                                 }else{
 559                                                         wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
 560                                                         delete nast2;
 561                                                         delete copy;
 562                                                 }
 563                                         }
 564
 565                                         //create accnos file with names
 566                                         accnosFile << candidateSeq->getName() << wasBetter << endl;
 567                                 }
 568
 569                                 report.setTemplate(templateSeq);
 570                                 report.setSearchParameters(search, searchScore);
 571                                 report.setAlignmentParameters(align, alignment);
 572                                 report.setNastParameters(*nast);
 573
 574                                 alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
 575
 576                                 report.print();
 577                                 delete nast;
 578                                 if (needToDeleteCopy) {   delete copy;   }
 579
 580                                 count++;
 581                         }
 582                         delete candidateSeq;
 583
 584                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 585                                 unsigned long int pos = inFASTA.tellg();
 586                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 587                         #else
 588                                 if (inFASTA.eof()) { break; }
 589                         #endif
 590
 591                         //report progress
 592                         if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 593
 594                 }
 595                 //report progress
 596                 if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 597
 598                 alignmentFile.close();
 599                 inFASTA.close();
 600                 accnosFile.close();
 601
 602                 return count;
 603         }
 604         catch(exception& e) {
 605                 m->errorOut(e, "AlignCommand", "driver");
 606                 exit(1);
 607         }
 608 }
 609 //**********************************************************************************************************************
 610 #ifdef USE_MPI
 611 int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector<unsigned long int>& MPIPos){
 612         try {
 613                 string outputString = "";
 614                 MPI_Status statusReport;
 615                 MPI_Status statusAlign;
 616                 MPI_Status statusAccnos;
 617                 MPI_Status status;
 618                 int pid;
 619                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 620
 621                 NastReport report;
 622
 623                 if (pid == 0) {
 624                         outputString = report.getHeaders();
 625                         int length = outputString.length();
 626
 627                         char* buf = new char[length];
 628                         memcpy(buf, outputString.c_str(), length);
 629
 630                         MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport);
 631
 632             delete buf;
 633                 }
 634
 635                 for(int i=0;i<num;i++){
 636
 637                         if (m->control_pressed) {  return 0; }
 638
 639                         //read next sequence
 640                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 641
 642                         char* buf4 = new char[length];
 643                         //memcpy(buf4, outputString.c_str(), length);
 644
 645                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 646
 647                         string tempBuf = buf4;
 648
 649                         delete buf4;
 650
 651                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 652
 653                         istringstream iss (tempBuf,istringstream::in);
 654
 655                         Sequence* candidateSeq = new Sequence(iss);
 656                         report.setCandidate(candidateSeq);
 657
 658                         int origNumBases = candidateSeq->getNumBases();
 659                         string originalUnaligned = candidateSeq->getUnaligned();
 660                         int numBasesNeeded = origNumBases * threshold;
 661
 662                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 663                                 if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
 664                                         alignment->resize(candidateSeq->getUnaligned().length()+1);
 665                                 }
 666
 667                                 Sequence temp = templateDB->findClosestSequence(candidateSeq);
 668                                 Sequence* templateSeq = &temp;
 669
 670                                 float searchScore = templateDB->getSearchScore();
 671
 672                                 Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
 673                                 Sequence* copy;
 674
 675                                 Nast* nast2;
 676                                 bool needToDeleteCopy = false;  //this is needed in case you have you enter the ifs below
 677                                                                                                 //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
 678                                                                                                 //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
 679                                                                                                 //so this bool tells you if you need to delete it
 680
 681                                 //if there is a possibility that this sequence should be reversed
 682                                 if (candidateSeq->getNumBases() < numBasesNeeded) {
 683
 684                                         string wasBetter = "";
 685                                         //if the user wants you to try the reverse
 686                                         if (flip) {
 687                                                 //get reverse compliment
 688                                                 copy = new Sequence(candidateSeq->getName(), originalUnaligned);
 689                                                 copy->reverseComplement();
 690
 691                                                 //rerun alignment
 692                                                 Sequence temp2 = templateDB->findClosestSequence(copy);
 693                                                 Sequence* templateSeq2 = &temp2;
 694
 695                                                 searchScore = templateDB->getSearchScore();
 696
 697                                                 nast2 = new Nast(alignment, copy, templateSeq2);
 698
 699                                                 //check if any better
 700                                                 if (copy->getNumBases() > candidateSeq->getNumBases()) {
 701                                                         candidateSeq->setAligned(copy->getAligned());  //use reverse compliments alignment since its better
 702                                                         templateSeq = templateSeq2;
 703                                                         delete nast;
 704                                                         nast = nast2;
 705                                                         needToDeleteCopy = true;
 706                                                         wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
 707                                                 }else{
 708                                                         wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";
 709                                                         delete nast2;
 710                                                         delete copy;
 711                                                 }
 712                                         }
 713
 714                                         //create accnos file with names
 715                                         outputString = candidateSeq->getName() + wasBetter + "\n";
 716
 717                                         //send results to parent
 718                                         int length = outputString.length();
 719
 720                                         char* buf = new char[length];
 721                                         memcpy(buf, outputString.c_str(), length);
 722
 723                                         MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos);
 724                                         delete buf;
 725                                         MPIWroteAccnos = true;
 726                                 }
 727
 728                                 report.setTemplate(templateSeq);
 729                                 report.setSearchParameters(search, searchScore);
 730                                 report.setAlignmentParameters(align, alignment);
 731                                 report.setNastParameters(*nast);
 732
 733                                 outputString =  ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n";
 734
 735                                 //send results to parent
 736                                 int length = outputString.length();
 737                                 char* buf2 = new char[length];
 738                                 memcpy(buf2, outputString.c_str(), length);
 739
 740                                 MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign);
 741
 742                                 delete buf2;
 743
 744                                 outputString = report.getReport();
 745
 746                                 //send results to parent
 747                                 length = outputString.length();
 748                                 char* buf3 = new char[length];
 749                                 memcpy(buf3, outputString.c_str(), length);
 750
 751                                 MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport);
 752
 753                                 delete buf3;
 754                                 delete nast;
 755                                 if (needToDeleteCopy) {   delete copy;   }
 756                         }
 757                         delete candidateSeq;
 758
 759                         //report progress
 760                         if((i+1) % 100 == 0){   cout << (toString(i+1)) << endl;                }
 761                 }
 762                 //report progress
 763                 if((num) % 100 != 0){   cout << (toString(num)) << endl;                }
 764
 765                 return 1;
 766         }
 767         catch(exception& e) {
 768                 m->errorOut(e, "AlignCommand", "driverMPI");
 769                 exit(1);
 770         }
 771 }
 772 #endif
 773 /**************************************************************************************************/
 774
 775 int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) {
 776         try {
 777 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 778                 processIDS.resize(0);
 779                 int process = 1;
 780                 int num = 0;
 781                 //              processIDS.resize(0);
 782
 783                 //loop through and create all the processes you want
 784                 while (process != processors) {
 785                         int pid = fork();
 786
 787                         if (pid > 0) {
 788                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 789                                 process++;
 790                         }else if (pid == 0){
 791                                 num = driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename);
 792
 793                                 //pass numSeqs to parent
 794                                 ofstream out;
 795                                 string tempFile = alignFileName + toString(getpid()) + ".num.temp";
 796                                 m->openOutputFile(tempFile, out);
 797                                 out << num << endl;
 798                                 out.close();
 799
 800                                 exit(0);
 801                         }else {
 802                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 803                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 804                                 exit(0);
 805                         }
 806                 }
 807
 808                 //do my part
 809                 num = driver(lines[0], alignFileName, reportFileName, accnosFName, filename);
 810
 811                 //force parent to wait until all the processes are done
 812                 for (int i=0;i<processors;i++) {
 813                         int temp = processIDS[i];
 814                         wait(&temp);
 815                 }
 816
 817                 vector<string> nonBlankAccnosFiles;
 818                 if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
 819                 else { remove(accnosFName.c_str()); } //remove so other files can be renamed to it
 820
 821                 for (int i = 0; i < processIDS.size(); i++) {
 822                         ifstream in;
 823                         string tempFile =  alignFileName + toString(processIDS[i]) + ".num.temp";
 824                         m->openInputFile(tempFile, in);
 825                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 826                         in.close(); remove(tempFile.c_str());
 827
 828                         appendAlignFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
 829                         remove((alignFileName + toString(processIDS[i]) + ".temp").c_str());
 830
 831                         appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);
 832                         remove((reportFileName + toString(processIDS[i]) + ".temp").c_str());
 833
 834                         if (!(m->isBlank(accnosFName + toString(processIDS[i]) + ".temp"))) {
 835                                 nonBlankAccnosFiles.push_back(accnosFName + toString(processIDS[i]) + ".temp");
 836                         }else { remove((accnosFName + toString(processIDS[i]) + ".temp").c_str());  }
 837
 838                 }
 839
 840                 //append accnos files
 841                 if (nonBlankAccnosFiles.size() != 0) {
 842                         rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
 843
 844                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
 845                                 appendAlignFiles(nonBlankAccnosFiles[h], accnosFName);
 846                                 remove(nonBlankAccnosFiles[h].c_str());
 847                         }
 848                 }else { //recreate the accnosfile if needed
 849                         ofstream out;
 850                         m->openOutputFile(accnosFName, out);
 851                         out.close();
 852                 }
 853
 854                 return num;
 855 #endif
 856         }
 857         catch(exception& e) {
 858                 m->errorOut(e, "AlignCommand", "createProcesses");
 859                 exit(1);
 860         }
 861 }
 862 /**************************************************************************************************/
 863
 864 void AlignCommand::appendAlignFiles(string temp, string filename) {
 865         try{
 866
 867                 ofstream output;
 868                 ifstream input;
 869                 m->openOutputFileAppend(filename, output);
 870                 m->openInputFile(temp, input);
 871
 872                 while(char c = input.get()){
 873                         if(input.eof())         {       break;                  }
 874                         else                            {       output << c;    }
 875                 }
 876
 877                 input.close();
 878                 output.close();
 879         }
 880         catch(exception& e) {
 881                 m->errorOut(e, "AlignCommand", "appendAlignFiles");
 882                 exit(1);
 883         }
 884 }
 885 //**********************************************************************************************************************
 886
 887 void AlignCommand::appendReportFiles(string temp, string filename) {
 888         try{
 889
 890                 ofstream output;
 891                 ifstream input;
 892                 m->openOutputFileAppend(filename, output);
 893                 m->openInputFile(temp, input);
 894
 895                 while (!input.eof())    {       char c = input.get(); if (c == 10 || c == 13){  break;  }       } // get header line
 896
 897                 while(char c = input.get()){
 898                         if(input.eof())         {       break;                  }
 899                         else                            {       output << c;    }
 900                 }
 901
 902                 input.close();
 903                 output.close();
 904         }
 905         catch(exception& e) {
 906                 m->errorOut(e, "AlignCommand", "appendReportFiles");
 907                 exit(1);
 908         }
 909 }
 910 //**********************************************************************************************************************