filterseqscommand.cpp

   1 /*
   2  *  filterseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Thomas Ryabin on 5/4/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "filterseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 /**************************************************************************************/
  14
  15 FilterSeqsCommand::FilterSeqsCommand(string option)  {
  16         try {
  17                 abort = false;
  18                 filterFileName = "";
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string Array[] =  {"fasta", "trump", "soft", "hard", "vertical", "outputdir","inputdir", "processors"};
  26                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string,string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string,string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //if the user changes the input directory command factory will send this info to us in the output parameter
  40                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  41                         if (inputDir == "not found"){   inputDir = "";          }
  42                         else {
  43                                 string path;
  44                                 it = parameters.find("fasta");
  45                                 //user has given a template file
  46                                 if(it != parameters.end()){
  47                                         path = hasPath(it->second);
  48                                         //if the user has not given a path then, add inputdir. else leave path alone.
  49                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  50                                 }
  51
  52                                 it = parameters.find("hard");
  53                                 //user has given a template file
  54                                 if(it != parameters.end()){
  55                                         path = hasPath(it->second);
  56                                         //if the user has not given a path then, add inputdir. else leave path alone.
  57                                         if (path == "") {       parameters["hard"] = inputDir + it->second;             }
  58                                 }
  59                         }
  60
  61                         //check for required parameters
  62                         fasta = validParameter.validFile(parameters, "fasta", false);
  63                         if (fasta == "not found") { m->mothurOut("fasta is a required parameter for the filter.seqs command."); m->mothurOutEndLine(); abort = true;  }
  64                         else {
  65                                 splitAtDash(fasta, fastafileNames);
  66
  67                                 //go through files and make sure they are good, if not, then disregard them
  68                                 for (int i = 0; i < fastafileNames.size(); i++) {
  69                                         if (inputDir != "") {
  70                                                 string path = hasPath(fastafileNames[i]);
  71                                                 //if the user has not given a path then, add inputdir. else leave path alone.
  72                                                 if (path == "") {       fastafileNames[i] = inputDir + fastafileNames[i];               }
  73                                         }
  74
  75                                         int ableToOpen;
  76                                         ifstream in;
  77                                         ableToOpen = openInputFile(fastafileNames[i], in);
  78                                         if (ableToOpen == 1) {
  79                                                 m->mothurOut(fastafileNames[i] + " will be disregarded."); m->mothurOutEndLine();
  80                                                 //erase from file list
  81                                                 fastafileNames.erase(fastafileNames.begin()+i);
  82                                                 i--;
  83                                         }else{
  84                                                 string simpleName = getSimpleName(fastafileNames[i]);
  85                                                 filterFileName += simpleName.substr(0, simpleName.find_first_of('.'));
  86                                         }
  87                                         in.close();
  88                                 }
  89
  90                                 //make sure there is at least one valid file left
  91                                 if (fastafileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
  92                         }
  93
  94                         if (!abort) {
  95                                 //if the user changes the output directory command factory will send this info to us in the output parameter
  96                                 outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  97                                         outputDir = "";
  98                                         outputDir += hasPath(fastafileNames[0]); //if user entered a file with a path then preserve it
  99                                 }
 100                         }
 101                         //check for optional parameter and set defaults
 102                         // ...at some point should added some additional type checking...
 103
 104                         string temp;
 105                         temp = validParameter.validFile(parameters, "trump", false);                    if (temp == "not found") { temp = "*"; }
 106                         trump = temp[0];
 107
 108                         temp = validParameter.validFile(parameters, "soft", false);                             if (temp == "not found") { soft = 0; }
 109                         else {  soft = (float)atoi(temp.c_str()) / 100.0;  }
 110
 111                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 112                         convert(temp, processors);
 113
 114                         hard = validParameter.validFile(parameters, "hard", true);                              if (hard == "not found") { hard = ""; }
 115                         else if (hard == "not open") { abort = true; }
 116
 117                         vertical = validParameter.validFile(parameters, "vertical", false);             if (vertical == "not found") { vertical = "T"; }
 118
 119                         numSeqs = 0;
 120
 121                 }
 122
 123         }
 124         catch(exception& e) {
 125                 m->errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
 126                 exit(1);
 127         }
 128 }
 129
 130 //**********************************************************************************************************************
 131
 132 void FilterSeqsCommand::help(){
 133         try {
 134
 135                 m->mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
 136                 m->mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
 137                 m->mothurOut("The fasta parameter is required. You may enter several fasta files to build the filter from and filter, by separating their names with -'s.\n");
 138                 m->mothurOut("For example: fasta=abrecovery.fasta-amazon.fasta \n");
 139                 m->mothurOut("The trump parameter .... The default is ...\n");
 140                 m->mothurOut("The soft parameter .... The default is ....\n");
 141                 m->mothurOut("The hard parameter .... The default is ....\n");
 142                 m->mothurOut("The vertical parameter .... The default is T.\n");
 143                 m->mothurOut("The filter.seqs command should be in the following format: \n");
 144                 m->mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
 145                 m->mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
 146                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 147
 148         }
 149         catch(exception& e) {
 150                 m->errorOut(e, "FilterSeqsCommand", "help");
 151                 exit(1);
 152         }
 153 }
 154
 155 /**************************************************************************************/
 156
 157 int FilterSeqsCommand::execute() {
 158         try {
 159
 160                 if (abort == true) { return 0; }
 161
 162                 ifstream inFASTA;
 163                 openInputFile(fastafileNames[0], inFASTA);
 164
 165                 Sequence testSeq(inFASTA);
 166                 alignmentLength = testSeq.getAlignLength();
 167                 inFASTA.close();
 168
 169                 ////////////create filter/////////////////
 170                 m->mothurOut("Creating Filter... "); m->mothurOutEndLine();
 171
 172                 filter = createFilter();
 173
 174                 m->mothurOutEndLine();  m->mothurOutEndLine();
 175
 176                 if (m->control_pressed) { return 0; }
 177
 178                 #ifdef USE_MPI
 179                         int pid;
 180                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 181
 182                         if (pid == 0) { //only one process should output the filter
 183                 #endif
 184
 185                 ofstream outFilter;
 186
 187                 string filterFile = outputDir + filterFileName + ".filter";
 188                 openOutputFile(filterFile, outFilter);
 189                 outFilter << filter << endl;
 190                 outFilter.close();
 191                 outputNames.push_back(filterFile);
 192
 193                 #ifdef USE_MPI
 194                         }
 195                 #endif
 196
 197                 ////////////run filter/////////////////
 198
 199                 m->mothurOut("Running Filter... "); m->mothurOutEndLine();
 200
 201                 filterSequences();
 202
 203                 m->mothurOutEndLine();  m->mothurOutEndLine();
 204
 205                 int filteredLength = 0;
 206                 for(int i=0;i<alignmentLength;i++){
 207                         if(filter[i] == '1'){   filteredLength++;       }
 208                 }
 209
 210                 if (m->control_pressed) {  for(int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }  return 0; }
 211
 212
 213                 m->mothurOutEndLine();
 214                 m->mothurOut("Length of filtered alignment: " + toString(filteredLength)); m->mothurOutEndLine();
 215                 m->mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); m->mothurOutEndLine();
 216                 m->mothurOut("Length of the original alignment: " + toString(alignmentLength)); m->mothurOutEndLine();
 217                 m->mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); m->mothurOutEndLine();
 218
 219
 220                 m->mothurOutEndLine();
 221                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 222                 for(int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();      }
 223                 m->mothurOutEndLine();
 224
 225                 return 0;
 226
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "FilterSeqsCommand", "execute");
 230                 exit(1);
 231         }
 232 }
 233 /**************************************************************************************/
 234 int FilterSeqsCommand::filterSequences() {
 235         try {
 236
 237                 numSeqs = 0;
 238
 239                 for (int s = 0; s < fastafileNames.size(); s++) {
 240
 241                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 242
 243                                 string filteredFasta = outputDir + getRootName(getSimpleName(fastafileNames[s])) + "filter.fasta";
 244 #ifdef USE_MPI
 245                                 int pid, start, end, numSeqsPerProcessor, num;
 246                                 int tag = 2001;
 247                                 vector<long>MPIPos;
 248
 249                                 MPI_Status status;
 250                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 251                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 252
 253                                 MPI_File outMPI;
 254                                 MPI_File tempMPI;
 255                                 MPI_File inMPI;
 256                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 257                                 int inMode=MPI_MODE_RDONLY;
 258
 259                                 //char* outFilename = new char[filteredFasta.length()];
 260                                 //memcpy(outFilename, filteredFasta.c_str(), filteredFasta.length());
 261
 262                                 char outFilename[1024];
 263                                 strcpy(outFilename, filteredFasta.c_str());
 264
 265                                 //char* inFileName = new char[fastafileNames[s].length()];
 266                                 //memcpy(inFileName, fastafileNames[s].c_str(), fastafileNames[s].length());
 267
 268                                 char inFileName[1024];
 269                                 strcpy(inFileName, fastafileNames[s].c_str());
 270
 271                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 272                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 273
 274                                 //delete inFileName;
 275                                 //delete outFilename;
 276
 277                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 278
 279                                 if (pid == 0) { //you are the root process
 280
 281                                         MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
 282                                         numSeqs += num;
 283
 284                                         //send file positions to all processes
 285                                         MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD);  //send numSeqs
 286                                         MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
 287
 288                                         //figure out how many sequences you have to do
 289                                         numSeqsPerProcessor = num / processors;
 290                                         int startIndex =  pid * numSeqsPerProcessor;
 291                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 292
 293
 294                                         //do your part
 295                                         driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 296
 297                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 298
 299                                         //wait on chidren
 300                                         for(int i = 1; i < processors; i++) {
 301                                                 char buf[4];
 302                                                 MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 303                                         }
 304
 305                                 }else { //you are a child process
 306                                         MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
 307                                         numSeqs += num;
 308                                         MPIPos.resize(num+1);
 309                                         MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
 310
 311                                         //figure out how many sequences you have to align
 312                                         numSeqsPerProcessor = num / processors;
 313                                         int startIndex =  pid * numSeqsPerProcessor;
 314                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 315
 316
 317                                         //align your part
 318                                         driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 319
 320                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 321
 322                                         char buf[4];
 323                                         strcpy(buf, "done");
 324
 325                                         //tell parent you are done.
 326                                         MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 327                                 }
 328
 329                                 MPI_File_close(&outMPI);
 330                                 MPI_File_close(&inMPI);
 331
 332 #else
 333                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 334                                 if(processors == 1){
 335                                         ifstream inFASTA;
 336                                         openInputFile(fastafileNames[s], inFASTA);
 337                                         int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 338                                         inFASTA.close();
 339
 340                                         lines.push_back(new linePair(0, numFastaSeqs));
 341
 342                                         numSeqs += numFastaSeqs;
 343
 344                                         driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
 345                                 }else{
 346                                         setLines(fastafileNames[s]);
 347                                         createProcessesRunFilter(filter, fastafileNames[s]);
 348
 349                                         rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
 350
 351                                         //append fasta files
 352                                         for(int i=1;i<processors;i++){
 353                                                 appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
 354                                                 remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
 355                                         }
 356                                 }
 357
 358                                 if (m->control_pressed) {  return 1; }
 359                 #else
 360                                 ifstream inFASTA;
 361                                 openInputFile(fastafileNames[s], inFASTA);
 362                                 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 363                                 inFASTA.close();
 364
 365                                 lines.push_back(new linePair(0, numFastaSeqs));
 366
 367                                 numSeqs += numFastaSeqs;
 368
 369                                 driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
 370
 371                                 if (m->control_pressed) {  return 1; }
 372                 #endif
 373 #endif
 374                         outputNames.push_back(filteredFasta);
 375                 }
 376
 377                 return 0;
 378         }
 379         catch(exception& e) {
 380                 m->errorOut(e, "FilterSeqsCommand", "filterSequences");
 381                 exit(1);
 382         }
 383 }
 384 #ifdef USE_MPI
 385 /**************************************************************************************/
 386 int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<long>& MPIPos) {
 387         try {
 388                 string outputString = "";
 389                 int count = 0;
 390                 MPI_Status status;
 391
 392                 for(int i=0;i<num;i++){
 393
 394                         if (m->control_pressed) { return 0; }
 395
 396                         //read next sequence
 397                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 398                         char* buf4 = new char[length];
 399                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 400
 401                         string tempBuf = buf4;
 402                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 403                         istringstream iss (tempBuf,istringstream::in);
 404                         delete buf4;
 405
 406                         Sequence seq(iss);  gobble(iss);
 407
 408                         if (seq.getName() != "") {
 409                                 string align = seq.getAligned();
 410                                 string filterSeq = "";
 411
 412                                 for(int j=0;j<alignmentLength;j++){
 413                                         if(filter[j] == '1'){
 414                                                 filterSeq += align[j];
 415                                         }
 416                                 }
 417
 418                                 count++;
 419                                 outputString += ">" + seq.getName() + "\n" + filterSeq + "\n";
 420
 421                                 if(count % 10 == 0){ //output to file
 422                                         //send results to parent
 423                                         int length = outputString.length();
 424                                         char* buf = new char[length];
 425                                         memcpy(buf, outputString.c_str(), length);
 426
 427                                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 428                                         outputString = "";
 429                                         delete buf;
 430                                 }
 431
 432                         }
 433
 434                         if((i+1) % 100 == 0){   cout << (i+1) << endl;   m->mothurOutJustToLog(toString(i+1) + "\n");   }
 435                 }
 436
 437                 if(outputString != ""){ //output to file
 438                         //send results to parent
 439                         int length = outputString.length();
 440                         char* buf = new char[length];
 441                         memcpy(buf, outputString.c_str(), length);
 442
 443                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 444                         outputString = "";
 445                         delete buf;
 446                 }
 447
 448                 if((num) % 100 != 0){   cout << (num) << endl;   m->mothurOutJustToLog(toString(num) + "\n");   }
 449
 450                 return 0;
 451         }
 452         catch(exception& e) {
 453                 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
 454                 exit(1);
 455         }
 456 }
 457 #endif
 458 /**************************************************************************************/
 459 int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* line) {
 460         try {
 461                 ofstream out;
 462                 openOutputFile(outputFilename, out);
 463
 464                 ifstream in;
 465                 openInputFile(inputFilename, in);
 466
 467                 in.seekg(line->start);
 468
 469                 for(int i=0;i<line->num;i++){
 470
 471                                 if (m->control_pressed) { in.close(); out.close(); return 0; }
 472
 473                                 Sequence seq(in);
 474                                 if (seq.getName() != "") {
 475                                         string align = seq.getAligned();
 476                                         string filterSeq = "";
 477
 478                                         for(int j=0;j<alignmentLength;j++){
 479                                                 if(filter[j] == '1'){
 480                                                         filterSeq += align[j];
 481                                                 }
 482                                         }
 483
 484                                         out << '>' << seq.getName() << endl << filterSeq << endl;
 485                                 }
 486                                 gobble(in);
 487                 }
 488                 out.close();
 489                 in.close();
 490
 491                 return 0;
 492         }
 493         catch(exception& e) {
 494                 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
 495                 exit(1);
 496         }
 497 }
 498 /**************************************************************************************************/
 499
 500 int FilterSeqsCommand::createProcessesRunFilter(string F, string filename) {
 501         try {
 502 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 503                 int process = 0;
 504                 int exitCommand = 1;
 505                 processIDS.clear();
 506
 507                 //loop through and create all the processes you want
 508                 while (process != processors) {
 509                         int pid = fork();
 510
 511                         if (pid > 0) {
 512                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 513                                 process++;
 514                         }else if (pid == 0){
 515                                 string filteredFasta = filename + toString(getpid()) + ".temp";
 516                                 driverRunFilter(F, filteredFasta, filename, lines[process]);
 517                                 exit(0);
 518                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 519                 }
 520
 521                 //force parent to wait until all the processes are done
 522                 for (int i=0;i<processors;i++) {
 523                         int temp = processIDS[i];
 524                         wait(&temp);
 525                 }
 526
 527                 return exitCommand;
 528 #endif
 529         }
 530         catch(exception& e) {
 531                 m->errorOut(e, "FilterSeqsCommand", "createProcessesRunFilter");
 532                 exit(1);
 533         }
 534 }
 535 /**************************************************************************************/
 536 string FilterSeqsCommand::createFilter() {
 537         try {
 538                 string filterString = "";
 539                 Filters F;
 540
 541                 if (soft != 0)                  {  F.setSoft(soft);             }
 542                 if (trump != '*')               {  F.setTrump(trump);   }
 543
 544                 F.setLength(alignmentLength);
 545
 546                 if(soft != 0 || isTrue(vertical)){
 547                         F.initialize();
 548                 }
 549
 550                 if(hard.compare("") != 0)       {       F.doHard(hard);         }
 551                 else                                            {       F.setFilter(string(alignmentLength, '1'));      }
 552
 553                 numSeqs = 0;
 554                 if(trump != '*' || isTrue(vertical) || soft != 0){
 555                         for (int s = 0; s < fastafileNames.size(); s++) {
 556
 557                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 558
 559 #ifdef USE_MPI
 560                                 int pid, numSeqsPerProcessor, num;
 561                                 int tag = 2001;
 562                                 vector<long> MPIPos;
 563
 564                                 MPI_Status status;
 565                                 MPI_File inMPI;
 566                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 567                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 568
 569                                 //char* tempFileName = new char(fastafileNames[s].length());
 570                                 //tempFileName = &(fastafileNames[s][0]);
 571
 572                                 char tempFileName[1024];
 573                                 strcpy(tempFileName, fastafileNames[s].c_str());
 574
 575                                 MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 576
 577                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 578
 579                                 if (pid == 0) { //you are the root process
 580                                                 MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
 581                                                 numSeqs += num;
 582
 583                                                 //send file positions to all processes
 584                                                 MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD);  //send numSeqs
 585                                                 MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
 586
 587                                                 //figure out how many sequences you have to do
 588                                                 numSeqsPerProcessor = num / processors;
 589                                                 int startIndex =  pid * numSeqsPerProcessor;
 590                                                 if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 591
 592
 593                                                 //do your part
 594                                                 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
 595
 596                                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 597
 598                                 }else { //i am the child process
 599
 600                                         MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
 601                                         MPIPos.resize(num+1);
 602                                         numSeqs += num;
 603                                         MPI_Bcast(&MPIPos[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
 604
 605                                         //figure out how many sequences you have to align
 606                                         numSeqsPerProcessor = num / processors;
 607                                         int startIndex =  pid * numSeqsPerProcessor;
 608                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 609
 610
 611                                         //do your part
 612                                         MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI,  MPIPos);
 613
 614                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 615                                 }
 616
 617                                 MPI_File_close(&inMPI);
 618
 619 #else
 620                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 621                                 if(processors == 1){
 622                                         ifstream inFASTA;
 623                                         openInputFile(fastafileNames[s], inFASTA);
 624                                         int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 625                                         inFASTA.close();
 626
 627                                         numSeqs += numFastaSeqs;
 628
 629                                         lines.push_back(new linePair(0, numFastaSeqs));
 630
 631                                         driverCreateFilter(F, fastafileNames[s], lines[0]);
 632                                 }else{
 633                                         setLines(fastafileNames[s]);
 634                                         createProcessesCreateFilter(F, fastafileNames[s]);
 635                                 }
 636
 637                                 if (m->control_pressed) {  return filterString; }
 638                 #else
 639                                 ifstream inFASTA;
 640                                 openInputFile(fastafileNames[s], inFASTA);
 641                                 int numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 642                                 inFASTA.close();
 643
 644                                 numSeqs += numFastaSeqs;
 645
 646                                 lines.push_back(new linePair(0, numFastaSeqs));
 647
 648                                 driverCreateFilter(F, fastafileNames[s], lines[0]);
 649                                 if (m->control_pressed) {  return filterString; }
 650                 #endif
 651 #endif
 652
 653                         }
 654                 }
 655
 656
 657 #ifdef USE_MPI
 658                 int pid;
 659                 int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
 660                 MPI_Status status;
 661
 662                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 663
 664                 if(trump != '*' || isTrue(vertical) || soft != 0){
 665
 666                         if (pid == 0) { //only one process should output the filter
 667
 668                                 vector<int> temp; temp.resize(alignmentLength+1);
 669
 670                                 //get the frequencies from the child processes
 671                                 for(int i = 0; i < ((processors-1)*5); i++) {
 672                                         MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, MPI_ANY_SOURCE, 2001, MPI_COMM_WORLD, &status);
 673                                         int receiveTag = temp[temp.size()-1];  //child process added a int to the end to indicate what letter count this is for
 674
 675                                         if (receiveTag == Atag) { //you are recieveing the A frequencies
 676                                                 for (int k = 0; k < alignmentLength; k++) {             F.a[k] += temp[k];      }
 677                                         }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
 678                                                 for (int k = 0; k < alignmentLength; k++) {             F.t[k] += temp[k];      }
 679                                         }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
 680                                                 for (int k = 0; k < alignmentLength; k++) {             F.c[k] += temp[k];      }
 681                                         }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
 682                                                 for (int k = 0; k < alignmentLength; k++) {             F.g[k] += temp[k];      }
 683                                         }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
 684                                                 for (int k = 0; k < alignmentLength; k++) {             F.gap[k] += temp[k];    }
 685                                         }
 686                                 }
 687                         }else{
 688
 689                                 //send my fequency counts
 690                                 F.a.push_back(Atag);
 691                                 int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 692                                 F.t.push_back(Ttag);
 693                                 ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 694                                 F.c.push_back(Ctag);
 695                                 ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 696                                 F.g.push_back(Gtag);
 697                                 ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 698                                 F.gap.push_back(Gaptag);
 699                                 ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 700                         }
 701
 702                 }
 703
 704                 if (pid == 0) { //only one process should output the filter
 705 #endif
 706                 F.setNumSeqs(numSeqs);
 707
 708                 if(isTrue(vertical) == 1)       {       F.doVertical(); }
 709                 if(soft != 0)                           {       F.doSoft();             }
 710
 711                 filterString = F.getFilter();
 712
 713 #ifdef USE_MPI
 714                 //send filter string to kids
 715                 MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
 716         }else{
 717                 //recieve filterString
 718                 char* tempBuf = new char[alignmentLength];
 719                 MPI_Bcast(tempBuf, alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
 720
 721                 filterString = tempBuf;
 722                 if (filterString.length() > alignmentLength) { filterString = filterString.substr(0, alignmentLength);  }
 723                 delete tempBuf;
 724         }
 725
 726         MPI_Barrier(MPI_COMM_WORLD);
 727 #endif
 728
 729
 730                 return filterString;
 731         }
 732         catch(exception& e) {
 733                 m->errorOut(e, "FilterSeqsCommand", "createFilter");
 734                 exit(1);
 735         }
 736 }
 737 /**************************************************************************************/
 738 int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* line) {
 739         try {
 740
 741                 ifstream in;
 742                 openInputFile(filename, in);
 743
 744                 in.seekg(line->start);
 745
 746                 for(int i=0;i<line->num;i++){
 747
 748                         if (m->control_pressed) { in.close(); return 1; }
 749
 750                         Sequence seq(in);
 751                         if (seq.getName() != "") {
 752                                         if (seq.getAligned().length() != alignmentLength) { m->mothurOut("Sequences are not all the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true;  }
 753
 754                                         if(trump != '*'){       F.doTrump(seq); }
 755                                         if(isTrue(vertical) || soft != 0){      F.getFreqs(seq);        }
 756                                         cout.flush();
 757                         }
 758
 759                         //report progress
 760                         if((i+1) % 100 == 0){   m->mothurOut(toString(i+1)); m->mothurOutEndLine();             }
 761                 }
 762
 763                 //report progress
 764                 if((line->num) % 100 != 0){     m->mothurOut(toString(line->num)); m->mothurOutEndLine();               }
 765
 766                 in.close();
 767
 768                 return 0;
 769         }
 770         catch(exception& e) {
 771                 m->errorOut(e, "FilterSeqsCommand", "driverCreateFilter");
 772                 exit(1);
 773         }
 774 }
 775 #ifdef USE_MPI
 776 /**************************************************************************************/
 777 int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<long>& MPIPos) {
 778         try {
 779
 780                 MPI_Status status;
 781                 int pid;
 782                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 783
 784                 for(int i=0;i<num;i++){
 785
 786                         if (m->control_pressed) { return 0; }
 787
 788                         //read next sequence
 789                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 790
 791                         char* buf4 = new char[length];
 792                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 793
 794                         string tempBuf = buf4;
 795                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 796                         istringstream iss (tempBuf,istringstream::in);
 797                         delete buf4;
 798
 799                         Sequence seq(iss);
 800
 801                         if (seq.getAligned().length() != alignmentLength) {  cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1);  }
 802
 803                         if(trump != '*'){       F.doTrump(seq); }
 804                         if(isTrue(vertical) || soft != 0){      F.getFreqs(seq);        }
 805                         cout.flush();
 806
 807                         //report progress
 808                         if((i+1) % 100 == 0){   cout << (i+1) << endl;   m->mothurOutJustToLog(toString(i+1) + "\n");   }
 809                 }
 810
 811                 //report progress
 812                 if((num) % 100 != 0){   cout << num << endl; m->mothurOutJustToLog(toString(num) + "\n");       }
 813
 814                 return 0;
 815         }
 816         catch(exception& e) {
 817                 m->errorOut(e, "FilterSeqsCommand", "MPICreateFilter");
 818                 exit(1);
 819         }
 820 }
 821 #endif
 822 /**************************************************************************************************/
 823
 824 int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
 825         try {
 826 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 827                 int process = 0;
 828                 int exitCommand = 1;
 829                 processIDS.clear();
 830
 831                 //loop through and create all the processes you want
 832                 while (process != processors) {
 833                         int pid = vfork();
 834
 835                         if (pid > 0) {
 836                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 837                                 process++;
 838                         }else if (pid == 0){
 839                                 driverCreateFilter(F, filename, lines[process]);
 840                                 exit(0);
 841                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 842                 }
 843
 844                 //force parent to wait until all the processes are done
 845                 for (int i=0;i<processors;i++) {
 846                         int temp = processIDS[i];
 847                         wait(&temp);
 848                 }
 849
 850                 return exitCommand;
 851 #endif
 852         }
 853         catch(exception& e) {
 854                 m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
 855                 exit(1);
 856         }
 857 }
 858 /**************************************************************************************************/
 859
 860 int FilterSeqsCommand::setLines(string filename) {
 861         try {
 862
 863                 vector<long int> positions;
 864                 bufferSizes.clear();
 865
 866                 ifstream inFASTA;
 867                 openInputFile(filename, inFASTA);
 868
 869                 string input;
 870                 while(!inFASTA.eof()){
 871                         input = getline(inFASTA);
 872
 873                         if (input.length() != 0) {
 874                                 if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);     }
 875                         }
 876                 }
 877                 inFASTA.close();
 878
 879                 int numFastaSeqs = positions.size();
 880
 881                 FILE * pFile;
 882                 long size;
 883
 884                 //get num bytes in file
 885                 pFile = fopen (filename.c_str(),"rb");
 886                 if (pFile==NULL) perror ("Error opening file");
 887                 else{
 888                         fseek (pFile, 0, SEEK_END);
 889                         size=ftell (pFile);
 890                         fclose (pFile);
 891                 }
 892
 893                 numSeqs += numFastaSeqs;
 894
 895                 int numSeqsPerProcessor = numFastaSeqs / processors;
 896
 897                 for (int i = 0; i < processors; i++) {
 898
 899                         long int startPos = positions[ i * numSeqsPerProcessor ];
 900                         if(i == processors - 1){
 901                                 numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 902                                 bufferSizes.push_back(size - startPos);
 903                         }else{
 904                                 long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
 905                                 bufferSizes.push_back(myEnd-startPos);
 906                         }
 907                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 908                 }
 909
 910                 return 0;
 911         }
 912         catch(exception& e) {
 913                 m->errorOut(e, "FilterSeqsCommand", "setLines");
 914                 exit(1);
 915         }
 916 }
 917 /**************************************************************************************/