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