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 vector<string> FilterSeqsCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta", "trump", "soft", "hard", "vertical", "outputdir","inputdir", "processors"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "FilterSeqsCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 FilterSeqsCommand::FilterSeqsCommand(){
  27         try {
  28                 abort = true;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32                 outputTypes["filter"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> FilterSeqsCommand::getRequiredParameters(){
  41         try {
  42                 string Array[] =  {"fasta"};
  43                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "FilterSeqsCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> FilterSeqsCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "FilterSeqsCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 /**************************************************************************************/
  63 FilterSeqsCommand::FilterSeqsCommand(string option)  {
  64         try {
  65                 abort = false;
  66                 filterFileName = "";
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string Array[] =  {"fasta", "trump", "soft", "hard", "vertical", "outputdir","inputdir", "processors"};
  74                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string,string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter("filter.seqs");
  80                         map<string,string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["fasta"] = tempOutNames;
  90                         outputTypes["filter"] = tempOutNames;
  91
  92                         //if the user changes the input directory command factory will send this info to us in the output parameter
  93                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  94                         if (inputDir == "not found"){   inputDir = "";          }
  95                         else {
  96                                 string path;
  97                                 it = parameters.find("fasta");
  98                                 //user has given a template file
  99                                 if(it != parameters.end()){
 100                                         path = m->hasPath(it->second);
 101                                         //if the user has not given a path then, add inputdir. else leave path alone.
 102                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 103                                 }
 104
 105                                 it = parameters.find("hard");
 106                                 //user has given a template file
 107                                 if(it != parameters.end()){
 108                                         path = m->hasPath(it->second);
 109                                         //if the user has not given a path then, add inputdir. else leave path alone.
 110                                         if (path == "") {       parameters["hard"] = inputDir + it->second;             }
 111                                 }
 112                         }
 113
 114                         //check for required parameters
 115                         fasta = validParameter.validFile(parameters, "fasta", false);
 116                         if (fasta == "not found") { m->mothurOut("fasta is a required parameter for the filter.seqs command."); m->mothurOutEndLine(); abort = true;  }
 117                         else {
 118                                 m->splitAtDash(fasta, fastafileNames);
 119
 120                                 //go through files and make sure they are good, if not, then disregard them
 121                                 for (int i = 0; i < fastafileNames.size(); i++) {
 122                                         if (inputDir != "") {
 123                                                 string path = m->hasPath(fastafileNames[i]);
 124                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 125                                                 if (path == "") {       fastafileNames[i] = inputDir + fastafileNames[i];               }
 126                                         }
 127
 128                                         ifstream in;
 129                                         int ableToOpen = m->openInputFile(fastafileNames[i], in, "noerror");
 130
 131                                         //if you can't open it, try default location
 132                                         if (ableToOpen == 1) {
 133                                                 if (m->getDefaultPath() != "") { //default path is set
 134                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastafileNames[i]);
 135                                                         m->mothurOut("Unable to open " + fastafileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 136                                                         ifstream in2;
 137                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 138                                                         in2.close();
 139                                                         fastafileNames[i] = tryPath;
 140                                                 }
 141                                         }
 142
 143                                         //if you can't open it, try default location
 144                                         if (ableToOpen == 1) {
 145                                                 if (m->getOutputDir() != "") { //default path is set
 146                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastafileNames[i]);
 147                                                         m->mothurOut("Unable to open " + fastafileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 148                                                         ifstream in2;
 149                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 150                                                         in2.close();
 151                                                         fastafileNames[i] = tryPath;
 152                                                 }
 153                                         }
 154
 155                                         in.close();
 156
 157                                         if (ableToOpen == 1) {
 158                                                 m->mothurOut("Unable to open " + fastafileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 159                                                 //erase from file list
 160                                                 fastafileNames.erase(fastafileNames.begin()+i);
 161                                                 i--;
 162                                         }else{
 163                                                 string simpleName = m->getSimpleName(fastafileNames[i]);
 164                                                 filterFileName += simpleName.substr(0, simpleName.find_first_of('.'));
 165                                         }
 166                                         in.close();
 167                                 }
 168
 169                                 //make sure there is at least one valid file left
 170                                 if (fastafileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 171                         }
 172
 173                         if (!abort) {
 174                                 //if the user changes the output directory command factory will send this info to us in the output parameter
 175                                 outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 176                                         outputDir = "";
 177                                         outputDir += m->hasPath(fastafileNames[0]); //if user entered a file with a path then preserve it
 178                                 }
 179                         }
 180                         //check for optional parameter and set defaults
 181                         // ...at some point should added some additional type checking...
 182
 183                         string temp;
 184                         hard = validParameter.validFile(parameters, "hard", true);                              if (hard == "not found") { hard = ""; }
 185                         else if (hard == "not open") { hard = ""; abort = true; }
 186
 187                         temp = validParameter.validFile(parameters, "trump", false);                    if (temp == "not found") { temp = "*"; }
 188                         trump = temp[0];
 189
 190                         temp = validParameter.validFile(parameters, "soft", false);                             if (temp == "not found") { soft = 0; }
 191                         else {  soft = (float)atoi(temp.c_str()) / 100.0;  }
 192
 193                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 194                         convert(temp, processors);
 195
 196                         vertical = validParameter.validFile(parameters, "vertical", false);
 197                         if (vertical == "not found") {
 198                                 if ((hard == "") && (trump == '*') && (soft == 0)) { vertical = "T"; } //you have not given a hard file or set the trump char.
 199                                 else { vertical = "F";  }
 200                         }
 201
 202                         numSeqs = 0;
 203                 }
 204
 205         }
 206         catch(exception& e) {
 207                 m->errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
 208                 exit(1);
 209         }
 210 }
 211
 212 //**********************************************************************************************************************
 213
 214 void FilterSeqsCommand::help(){
 215         try {
 216
 217                 m->mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
 218                 m->mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard, processors and vertical. \n");
 219                 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");
 220                 m->mothurOut("For example: fasta=abrecovery.fasta-amazon.fasta \n");
 221                 m->mothurOut("The trump parameter .... The default is ...\n");
 222                 m->mothurOut("The soft parameter .... The default is ....\n");
 223                 m->mothurOut("The hard parameter allows you to enter a file containing the filter you want to use.\n");
 224                 m->mothurOut("The vertical parameter removes columns where all sequences contain a gap character. The default is T.\n");
 225                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 226                 m->mothurOut("The filter.seqs command should be in the following format: \n");
 227                 m->mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump) \n");
 228                 m->mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=.).\n");
 229                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 230
 231         }
 232         catch(exception& e) {
 233                 m->errorOut(e, "FilterSeqsCommand", "help");
 234                 exit(1);
 235         }
 236 }
 237
 238 /**************************************************************************************/
 239
 240 int FilterSeqsCommand::execute() {
 241         try {
 242
 243                 if (abort == true) { return 0; }
 244
 245                 ifstream inFASTA;
 246                 m->openInputFile(fastafileNames[0], inFASTA);
 247
 248                 Sequence testSeq(inFASTA);
 249                 alignmentLength = testSeq.getAlignLength();
 250                 inFASTA.close();
 251
 252                 ////////////create filter/////////////////
 253                 m->mothurOut("Creating Filter... "); m->mothurOutEndLine();
 254
 255                 filter = createFilter();
 256
 257                 m->mothurOutEndLine();  m->mothurOutEndLine();
 258
 259                 if (m->control_pressed) { outputTypes.clear(); return 0; }
 260
 261                 #ifdef USE_MPI
 262                         int pid;
 263                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 264
 265                         if (pid == 0) { //only one process should output the filter
 266                 #endif
 267
 268                 ofstream outFilter;
 269
 270                 //prevent giantic file name
 271                 string filterFile;
 272                 if (fastafileNames.size() > 3) { filterFile = outputDir + "merge.filter"; }
 273                 else {  filterFile = outputDir + filterFileName + ".filter";  }
 274
 275                 m->openOutputFile(filterFile, outFilter);
 276                 outFilter << filter << endl;
 277                 outFilter.close();
 278                 outputNames.push_back(filterFile); outputTypes["filter"].push_back(filterFile);
 279
 280                 #ifdef USE_MPI
 281                         }
 282                 #endif
 283
 284                 ////////////run filter/////////////////
 285
 286                 m->mothurOut("Running Filter... "); m->mothurOutEndLine();
 287
 288                 filterSequences();
 289
 290                 m->mothurOutEndLine();  m->mothurOutEndLine();
 291
 292                 int filteredLength = 0;
 293                 for(int i=0;i<alignmentLength;i++){
 294                         if(filter[i] == '1'){   filteredLength++;       }
 295                 }
 296
 297                 if (m->control_pressed) {  outputTypes.clear(); for(int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }  return 0; }
 298
 299
 300                 m->mothurOutEndLine();
 301                 m->mothurOut("Length of filtered alignment: " + toString(filteredLength)); m->mothurOutEndLine();
 302                 m->mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); m->mothurOutEndLine();
 303                 m->mothurOut("Length of the original alignment: " + toString(alignmentLength)); m->mothurOutEndLine();
 304                 m->mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); m->mothurOutEndLine();
 305
 306
 307                 m->mothurOutEndLine();
 308                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 309                 for(int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();      }
 310                 m->mothurOutEndLine();
 311
 312                 return 0;
 313
 314         }
 315         catch(exception& e) {
 316                 m->errorOut(e, "FilterSeqsCommand", "execute");
 317                 exit(1);
 318         }
 319 }
 320 /**************************************************************************************/
 321 int FilterSeqsCommand::filterSequences() {
 322         try {
 323
 324                 numSeqs = 0;
 325
 326                 for (int s = 0; s < fastafileNames.size(); s++) {
 327
 328                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 329
 330                                 string filteredFasta = outputDir + m->getRootName(m->getSimpleName(fastafileNames[s])) + "filter.fasta";
 331 #ifdef USE_MPI
 332                                 int pid, start, end, numSeqsPerProcessor, num;
 333                                 int tag = 2001;
 334                                 vector<unsigned long int>MPIPos;
 335
 336                                 MPI_Status status;
 337                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 338                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 339
 340                                 MPI_File outMPI;
 341                                 MPI_File tempMPI;
 342                                 MPI_File inMPI;
 343                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 344                                 int inMode=MPI_MODE_RDONLY;
 345
 346                                 char outFilename[1024];
 347                                 strcpy(outFilename, filteredFasta.c_str());
 348
 349                                 char inFileName[1024];
 350                                 strcpy(inFileName, fastafileNames[s].c_str());
 351
 352                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 353                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 354
 355                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 356
 357                                 if (pid == 0) { //you are the root process
 358
 359                                         MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
 360                                         numSeqs += num;
 361
 362                                         //send file positions to all processes
 363                                         for(int i = 1; i < processors; i++) {
 364                                                 MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 365                                                 MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 366                                         }
 367
 368                                         //figure out how many sequences you have to do
 369                                         numSeqsPerProcessor = num / processors;
 370                                         int startIndex =  pid * numSeqsPerProcessor;
 371                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 372
 373
 374                                         //do your part
 375                                         driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 376
 377                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 378
 379                                         //wait on chidren
 380                                         for(int i = 1; i < processors; i++) {
 381                                                 char buf[5];
 382                                                 MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 383                                         }
 384
 385                                 }else { //you are a child process
 386                                         MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 387                                         MPIPos.resize(num+1);
 388                                         numSeqs += num;
 389                                         MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 390
 391                                         //figure out how many sequences you have to align
 392                                         numSeqsPerProcessor = num / processors;
 393                                         int startIndex =  pid * numSeqsPerProcessor;
 394                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 395
 396
 397                                         //align your part
 398                                         driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 399
 400                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 401
 402                                         char buf[5];
 403                                         strcpy(buf, "done");
 404
 405                                         //tell parent you are done.
 406                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 407                                 }
 408
 409                                 MPI_File_close(&outMPI);
 410                                 MPI_File_close(&inMPI);
 411                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 412
 413 #else
 414                         vector<unsigned long int> positions = m->divideFile(fastafileNames[s], processors);
 415
 416                         for (int i = 0; i < (positions.size()-1); i++) {
 417                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 418                         }
 419                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 420                                 if(processors == 1){
 421                                         int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
 422                                         numSeqs += numFastaSeqs;
 423                                 }else{
 424                                         int numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s]);
 425                                         numSeqs += numFastaSeqs;
 426
 427                                         rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
 428
 429                                         //append fasta files
 430                                         for(int i=1;i<processors;i++){
 431                                                 m->appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
 432                                                 remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
 433                                         }
 434                                 }
 435
 436                                 if (m->control_pressed) {  return 1; }
 437                 #else
 438                                 int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
 439                                 numSeqs += numFastaSeqs;
 440
 441                                 if (m->control_pressed) {  return 1; }
 442                 #endif
 443 #endif
 444                         outputNames.push_back(filteredFasta); outputTypes["fasta"].push_back(filteredFasta);
 445                 }
 446
 447                 return 0;
 448         }
 449         catch(exception& e) {
 450                 m->errorOut(e, "FilterSeqsCommand", "filterSequences");
 451                 exit(1);
 452         }
 453 }
 454 #ifdef USE_MPI
 455 /**************************************************************************************/
 456 int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
 457         try {
 458                 string outputString = "";
 459                 int count = 0;
 460                 MPI_Status status;
 461
 462                 for(int i=0;i<num;i++){
 463
 464                         if (m->control_pressed) { return 0; }
 465
 466                         //read next sequence
 467                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 468                         char* buf4 = new char[length];
 469                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 470
 471                         string tempBuf = buf4;
 472                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 473                         istringstream iss (tempBuf,istringstream::in);
 474                         delete buf4;
 475
 476                         Sequence seq(iss);  m->gobble(iss);
 477
 478                         if (seq.getName() != "") {
 479                                 string align = seq.getAligned();
 480                                 string filterSeq = "";
 481
 482                                 for(int j=0;j<alignmentLength;j++){
 483                                         if(filter[j] == '1'){
 484                                                 filterSeq += align[j];
 485                                         }
 486                                 }
 487
 488                                 count++;
 489                                 outputString += ">" + seq.getName() + "\n" + filterSeq + "\n";
 490
 491                                 if(count % 10 == 0){ //output to file
 492                                         //send results to parent
 493                                         int length = outputString.length();
 494                                         char* buf = new char[length];
 495                                         memcpy(buf, outputString.c_str(), length);
 496
 497                                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 498                                         outputString = "";
 499                                         delete buf;
 500                                 }
 501
 502                         }
 503
 504                         if((i+1) % 100 == 0){   cout << (i+1) << endl;   m->mothurOutJustToLog(toString(i+1) + "\n");   }
 505                 }
 506
 507                 if(outputString != ""){ //output to file
 508                         //send results to parent
 509                         int length = outputString.length();
 510                         char* buf = new char[length];
 511                         memcpy(buf, outputString.c_str(), length);
 512
 513                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 514                         outputString = "";
 515                         delete buf;
 516                 }
 517
 518                 if((num) % 100 != 0){   cout << (num) << endl;   m->mothurOutJustToLog(toString(num) + "\n");   }
 519
 520                 return 0;
 521         }
 522         catch(exception& e) {
 523                 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
 524                 exit(1);
 525         }
 526 }
 527 #endif
 528 /**************************************************************************************/
 529 int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* filePos) {
 530         try {
 531                 ofstream out;
 532                 m->openOutputFile(outputFilename, out);
 533
 534                 ifstream in;
 535                 m->openInputFile(inputFilename, in);
 536
 537                 in.seekg(filePos->start);
 538
 539                 bool done = false;
 540                 int count = 0;
 541
 542                 while (!done) {
 543
 544                                 if (m->control_pressed) { in.close(); out.close(); return 0; }
 545
 546                                 Sequence seq(in); m->gobble(in);
 547                                 if (seq.getName() != "") {
 548                                         string align = seq.getAligned();
 549                                         string filterSeq = "";
 550
 551                                         for(int j=0;j<alignmentLength;j++){
 552                                                 if(filter[j] == '1'){
 553                                                         filterSeq += align[j];
 554                                                 }
 555                                         }
 556
 557                                         out << '>' << seq.getName() << endl << filterSeq << endl;
 558                                 count++;
 559                         }
 560
 561                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 562                                 unsigned long int pos = in.tellg();
 563                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 564                         #else
 565                                 if (in.eof()) { break; }
 566                         #endif
 567
 568                         //report progress
 569                         if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 570                 }
 571                 //report progress
 572                 if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 573
 574
 575                 out.close();
 576                 in.close();
 577
 578                 return count;
 579         }
 580         catch(exception& e) {
 581                 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
 582                 exit(1);
 583         }
 584 }
 585 /**************************************************************************************************/
 586
 587 int FilterSeqsCommand::createProcessesRunFilter(string F, string filename) {
 588         try {
 589 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 590                 int process = 0;
 591                 int num = 0;
 592                 processIDS.clear();
 593
 594                 //loop through and create all the processes you want
 595                 while (process != processors) {
 596                         int pid = fork();
 597
 598                         if (pid > 0) {
 599                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 600                                 process++;
 601                         }else if (pid == 0){
 602                                 string filteredFasta = filename + toString(getpid()) + ".temp";
 603                                 num = driverRunFilter(F, filteredFasta, filename, lines[process]);
 604
 605                                 //pass numSeqs to parent
 606                                 ofstream out;
 607                                 string tempFile = filename +  toString(getpid()) + ".num.temp";
 608                                 m->openOutputFile(tempFile, out);
 609                                 out << num << endl;
 610                                 out.close();
 611
 612                                 exit(0);
 613                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 614                 }
 615
 616                 //force parent to wait until all the processes are done
 617                 for (int i=0;i<processors;i++) {
 618                         int temp = processIDS[i];
 619                         wait(&temp);
 620                 }
 621
 622                 for (int i = 0; i < processIDS.size(); i++) {
 623                         ifstream in;
 624                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
 625                         m->openInputFile(tempFile, in);
 626                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 627                         in.close(); remove(tempFile.c_str());
 628                 }
 629
 630
 631                 return num;
 632 #endif
 633         }
 634         catch(exception& e) {
 635                 m->errorOut(e, "FilterSeqsCommand", "createProcessesRunFilter");
 636                 exit(1);
 637         }
 638 }
 639 /**************************************************************************************/
 640 string FilterSeqsCommand::createFilter() {
 641         try {
 642                 string filterString = "";
 643                 Filters F;
 644
 645                 if (soft != 0)                  {  F.setSoft(soft);             }
 646                 if (trump != '*')               {  F.setTrump(trump);   }
 647
 648                 F.setLength(alignmentLength);
 649
 650                 if(trump != '*' || m->isTrue(vertical) || soft != 0){
 651                         F.initialize();
 652                 }
 653
 654                 if(hard.compare("") != 0)       {       F.doHard(hard);         }
 655                 else                                            {       F.setFilter(string(alignmentLength, '1'));      }
 656
 657                 numSeqs = 0;
 658                 if(trump != '*' || m->isTrue(vertical) || soft != 0){
 659                         for (int s = 0; s < fastafileNames.size(); s++) {
 660
 661                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 662
 663 #ifdef USE_MPI
 664                                 int pid, numSeqsPerProcessor, num;
 665                                 int tag = 2001;
 666                                 vector<unsigned long int> MPIPos;
 667
 668                                 MPI_Status status;
 669                                 MPI_File inMPI;
 670                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 671                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 672
 673                                 //char* tempFileName = new char(fastafileNames[s].length());
 674                                 //tempFileName = &(fastafileNames[s][0]);
 675
 676                                 char tempFileName[1024];
 677                                 strcpy(tempFileName, fastafileNames[s].c_str());
 678
 679                                 MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 680
 681                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 682
 683                                 if (pid == 0) { //you are the root process
 684                                                 MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
 685                                                 numSeqs += num;
 686
 687                                                 //send file positions to all processes
 688                                                 for(int i = 1; i < processors; i++) {
 689                                                         MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 690                                                         MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 691                                                 }
 692
 693                                                 //figure out how many sequences you have to do
 694                                                 numSeqsPerProcessor = num / processors;
 695                                                 int startIndex =  pid * numSeqsPerProcessor;
 696                                                 if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 697
 698
 699                                                 //do your part
 700                                                 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
 701
 702                                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 703
 704                                 }else { //i am the child process
 705                                         MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 706                                         MPIPos.resize(num+1);
 707                                         numSeqs += num;
 708                                         MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 709
 710                                         //figure out how many sequences you have to align
 711                                         numSeqsPerProcessor = num / processors;
 712                                         int startIndex =  pid * numSeqsPerProcessor;
 713                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 714
 715
 716                                         //do your part
 717                                         MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI,  MPIPos);
 718
 719                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 720                                 }
 721
 722                                 MPI_File_close(&inMPI);
 723                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 724
 725 #else
 726                 vector<unsigned long int> positions = m->divideFile(fastafileNames[s], processors);
 727
 728                 for (int i = 0; i < (positions.size()-1); i++) {
 729                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 730                 }
 731                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 732                                 if(processors == 1){
 733                                         int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
 734                                         numSeqs += numFastaSeqs;
 735                                 }else{
 736                                         int numFastaSeqs = createProcessesCreateFilter(F, fastafileNames[s]);
 737                                         numSeqs += numFastaSeqs;
 738                                 }
 739
 740                                 if (m->control_pressed) {  return filterString; }
 741                 #else
 742                                 int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
 743                                 numSeqs += numFastaSeqs;
 744                                 if (m->control_pressed) {  return filterString; }
 745                 #endif
 746 #endif
 747
 748                         }
 749                 }
 750
 751
 752 #ifdef USE_MPI
 753                 int pid;
 754                 int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
 755                 MPI_Status status;
 756
 757                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 758
 759                 if(trump != '*' || m->isTrue(vertical) || soft != 0){
 760
 761                         if (pid == 0) { //only one process should output the filter
 762
 763                                 vector<int> temp; temp.resize(alignmentLength+1);
 764
 765                                 //get the frequencies from the child processes
 766                                 for(int i = 1; i < processors; i++) {
 767
 768                                         for (int j = 0; j < 5; j++) {
 769
 770                                                 MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, i, 2001, MPI_COMM_WORLD, &status);
 771                                                 int receiveTag = temp[temp.size()-1];  //child process added a int to the end to indicate what letter count this is for
 772
 773                                                 if (receiveTag == Atag) { //you are recieveing the A frequencies
 774                                                         for (int k = 0; k < alignmentLength; k++) {             F.a[k] += temp[k];      }
 775                                                 }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
 776                                                         for (int k = 0; k < alignmentLength; k++) {             F.t[k] += temp[k];      }
 777                                                 }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
 778                                                         for (int k = 0; k < alignmentLength; k++) {             F.c[k] += temp[k];      }
 779                                                 }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
 780                                                         for (int k = 0; k < alignmentLength; k++) {             F.g[k] += temp[k];      }
 781                                                 }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
 782                                                         for (int k = 0; k < alignmentLength; k++) {             F.gap[k] += temp[k];    }
 783                                                 }
 784                                         }
 785                                 }
 786                         }else{
 787
 788                                 //send my fequency counts
 789                                 F.a.push_back(Atag);
 790                                 int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 791                                 F.t.push_back(Ttag);
 792                                 ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 793                                 F.c.push_back(Ctag);
 794                                 ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 795                                 F.g.push_back(Gtag);
 796                                 ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 797                                 F.gap.push_back(Gaptag);
 798                                 ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 799                         }
 800
 801                 }
 802
 803                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 804
 805                 if (pid == 0) { //only one process should output the filter
 806 #endif
 807
 808                 F.setNumSeqs(numSeqs);
 809                 if(m->isTrue(vertical) == 1)    {       F.doVertical(); }
 810                 if(soft != 0)                           {       F.doSoft();             }
 811                 filterString = F.getFilter();
 812
 813 #ifdef USE_MPI
 814                 //send filter string to kids
 815                 //for(int i = 1; i < processors; i++) {
 816                 //      MPI_Send(&filterString[0], alignmentLength, MPI_CHAR, i, 2001, MPI_COMM_WORLD);
 817                 //}
 818                 MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
 819         }else{
 820                 //recieve filterString
 821                 char* tempBuf = new char[alignmentLength];
 822                 //MPI_Recv(&tempBuf[0], alignmentLength, MPI_CHAR, 0, 2001, MPI_COMM_WORLD, &status);
 823                 MPI_Bcast(tempBuf, alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
 824
 825                 filterString = tempBuf;
 826                 if (filterString.length() > alignmentLength) { filterString = filterString.substr(0, alignmentLength);  }
 827                 delete tempBuf;
 828         }
 829
 830         MPI_Barrier(MPI_COMM_WORLD);
 831 #endif
 832
 833                 return filterString;
 834         }
 835         catch(exception& e) {
 836                 m->errorOut(e, "FilterSeqsCommand", "createFilter");
 837                 exit(1);
 838         }
 839 }
 840 /**************************************************************************************/
 841 int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* filePos) {
 842         try {
 843
 844                 ifstream in;
 845                 m->openInputFile(filename, in);
 846
 847                 in.seekg(filePos->start);
 848
 849                 bool done = false;
 850                 int count = 0;
 851
 852                 while (!done) {
 853
 854                         if (m->control_pressed) { in.close(); return 1; }
 855
 856                         Sequence seq(in); m->gobble(in);
 857                         if (seq.getName() != "") {
 858                                         if (seq.getAligned().length() != alignmentLength) { m->mothurOut("Sequences are not all the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true;  }
 859
 860                                         if(trump != '*')                        {       F.doTrump(seq);         }
 861                                         if(m->isTrue(vertical) || soft != 0)    {       F.getFreqs(seq);        }
 862                                         cout.flush();
 863                                         count++;
 864                         }
 865
 866                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 867                                 unsigned long int pos = in.tellg();
 868                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 869                         #else
 870                                 if (in.eof()) { break; }
 871                         #endif
 872
 873                         //report progress
 874                         if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 875                 }
 876                 //report progress
 877                 if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 878                 in.close();
 879
 880                 return count;
 881         }
 882         catch(exception& e) {
 883                 m->errorOut(e, "FilterSeqsCommand", "driverCreateFilter");
 884                 exit(1);
 885         }
 886 }
 887 #ifdef USE_MPI
 888 /**************************************************************************************/
 889 int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<unsigned long int>& MPIPos) {
 890         try {
 891
 892                 MPI_Status status;
 893                 int pid;
 894                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 895
 896                 for(int i=0;i<num;i++){
 897
 898                         if (m->control_pressed) { return 0; }
 899
 900                         //read next sequence
 901                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 902
 903                         char* buf4 = new char[length];
 904                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 905
 906                         string tempBuf = buf4;
 907                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 908                         istringstream iss (tempBuf,istringstream::in);
 909                         delete buf4;
 910
 911                         Sequence seq(iss);
 912
 913                         if (seq.getAligned().length() != alignmentLength) {  cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1);  }
 914
 915                         if(trump != '*'){       F.doTrump(seq); }
 916                         if(m->isTrue(vertical) || soft != 0){   F.getFreqs(seq);        }
 917                         cout.flush();
 918
 919                         //report progress
 920                         if((i+1) % 100 == 0){   cout << (i+1) << endl;   m->mothurOutJustToLog(toString(i+1) + "\n");   }
 921                 }
 922
 923                 //report progress
 924                 if((num) % 100 != 0){   cout << num << endl; m->mothurOutJustToLog(toString(num) + "\n");       }
 925
 926                 return 0;
 927         }
 928         catch(exception& e) {
 929                 m->errorOut(e, "FilterSeqsCommand", "MPICreateFilter");
 930                 exit(1);
 931         }
 932 }
 933 #endif
 934 /**************************************************************************************************/
 935
 936 int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
 937         try {
 938 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 939                 int process = 1;
 940                 int num = 0;
 941                 processIDS.clear();
 942
 943                 //loop through and create all the processes you want
 944                 while (process != processors) {
 945                         int pid = fork();
 946
 947                         if (pid > 0) {
 948                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 949                                 process++;
 950                         }else if (pid == 0){
 951                                 //reset child's filter counts to 0;
 952                                 F.a.clear(); F.a.resize(alignmentLength, 0);
 953                                 F.t.clear(); F.t.resize(alignmentLength, 0);
 954                                 F.g.clear(); F.g.resize(alignmentLength, 0);
 955                                 F.c.clear(); F.c.resize(alignmentLength, 0);
 956                                 F.gap.clear(); F.gap.resize(alignmentLength, 0);
 957
 958                                 num = driverCreateFilter(F, filename, lines[process]);
 959
 960                                 //write out filter counts to file
 961                                 filename += toString(getpid()) + "filterValues.temp";
 962                                 ofstream out;
 963                                 m->openOutputFile(filename, out);
 964
 965                                 out << num << endl;
 966                                 out << F.getFilter() << endl;
 967                                 for (int k = 0; k < alignmentLength; k++) {             out << F.a[k] << '\t'; }  out << endl;
 968                                 for (int k = 0; k < alignmentLength; k++) {             out << F.t[k] << '\t'; }  out << endl;
 969                                 for (int k = 0; k < alignmentLength; k++) {             out << F.g[k] << '\t'; }  out << endl;
 970                                 for (int k = 0; k < alignmentLength; k++) {             out << F.c[k] << '\t'; }  out << endl;
 971                                 for (int k = 0; k < alignmentLength; k++) {             out << F.gap[k] << '\t'; }  out << endl;
 972
 973                                 //cout << F.getFilter() << endl;
 974                                 out.close();
 975
 976                                 exit(0);
 977                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 978                 }
 979
 980                 //parent do your part
 981                 num = driverCreateFilter(F, filename, lines[0]);
 982
 983                 //force parent to wait until all the processes are done
 984                 for (int i=0;i<(processors-1);i++) {
 985                         int temp = processIDS[i];
 986                         wait(&temp);
 987                 }
 988
 989                 //parent reads in and combines Filter info
 990                 for (int i = 0; i < processIDS.size(); i++) {
 991                         string tempFilename = filename + toString(processIDS[i]) + "filterValues.temp";
 992                         ifstream in;
 993                         m->openInputFile(tempFilename, in);
 994
 995                         int temp, tempNum;
 996                         string tempFilterString;
 997
 998                         in >> tempNum; m->gobble(in); num += tempNum;
 999
1000                         in >> tempFilterString;
1001                         F.mergeFilter(tempFilterString);
1002
1003                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.a[k] += temp; }           m->gobble(in);
1004                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.t[k] += temp; }           m->gobble(in);
1005                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.g[k] += temp; }           m->gobble(in);
1006                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.c[k] += temp; }           m->gobble(in);
1007                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.gap[k] += temp; } m->gobble(in);
1008
1009                         in.close();
1010                         remove(tempFilename.c_str());
1011                 }
1012
1013                 return num;
1014 #endif
1015         }
1016         catch(exception& e) {
1017                 m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
1018                 exit(1);
1019         }
1020 }
1021 /**************************************************************************************/