screenseqscommand.cpp

   1 /*
   2  *  screenseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 6/3/09.
   6  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 #include "screenseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ScreenSeqsCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta", "start", "end", "maxambig", "maxhomop","optimize","criteria", "minlength", "maxlength",
  17                                                                         "name", "group", "alignreport","processors","outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "ScreenSeqsCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 ScreenSeqsCommand::ScreenSeqsCommand(){
  28         try {
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32                 outputTypes["name"] = tempOutNames;
  33                 outputTypes["group"] = tempOutNames;
  34                 outputTypes["alignreport"] = tempOutNames;
  35                 outputTypes["accnos"] = tempOutNames;
  36         }
  37         catch(exception& e) {
  38                 m->errorOut(e, "ScreenSeqsCommand", "ScreenSeqsCommand");
  39                 exit(1);
  40         }
  41 }
  42 //**********************************************************************************************************************
  43 vector<string> ScreenSeqsCommand::getRequiredParameters(){
  44         try {
  45                 string Array[] =  {"fasta"};
  46                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  47                 return myArray;
  48         }
  49         catch(exception& e) {
  50                 m->errorOut(e, "ScreenSeqsCommand", "getRequiredParameters");
  51                 exit(1);
  52         }
  53 }
  54 //**********************************************************************************************************************
  55 vector<string> ScreenSeqsCommand::getRequiredFiles(){
  56         try {
  57                 vector<string> myArray;
  58                 return myArray;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "ScreenSeqsCommand", "getRequiredFiles");
  62                 exit(1);
  63         }
  64 }
  65 //***************************************************************************************************************
  66
  67 ScreenSeqsCommand::ScreenSeqsCommand(string option)  {
  68         try {
  69                 abort = false;
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; }
  73
  74                 else {
  75                         //valid paramters for this command
  76                         string AlignArray[] =  {"fasta", "start", "end", "maxambig", "maxhomop","optimize","criteria", "minlength", "maxlength",
  77                                                                         "name", "group", "alignreport","processors","outputdir","inputdir"};
  78                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82
  83                         ValidParameters validParameter("screen.seqs");
  84                         map<string,string>::iterator it;
  85
  86                         //check to make sure all parameters are valid for command
  87                         for (it = parameters.begin(); it != parameters.end(); it++) {
  88                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  89                         }
  90
  91                         //initialize outputTypes
  92                         vector<string> tempOutNames;
  93                         outputTypes["fasta"] = tempOutNames;
  94                         outputTypes["name"] = tempOutNames;
  95                         outputTypes["group"] = tempOutNames;
  96                         outputTypes["alignreport"] = tempOutNames;
  97                         outputTypes["accnos"] = tempOutNames;
  98
  99                         //if the user changes the input directory command factory will send this info to us in the output parameter
 100                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 101                         if (inputDir == "not found"){   inputDir = "";          }
 102                         else {
 103                                 string path;
 104                                 it = parameters.find("fasta");
 105                                 //user has given a template file
 106                                 if(it != parameters.end()){
 107                                         path = m->hasPath(it->second);
 108                                         //if the user has not given a path then, add inputdir. else leave path alone.
 109                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 110                                 }
 111
 112                                 it = parameters.find("group");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 118                                 }
 119
 120                                 it = parameters.find("name");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 126                                 }
 127
 128                                 it = parameters.find("alignreport");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 134                                 }
 135                         }
 136
 137                         //check for required parameters
 138                         fastafile = validParameter.validFile(parameters, "fasta", true);
 139                         if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the screen.seqs command."); m->mothurOutEndLine(); abort = true; }
 140                         else if (fastafile == "not open") { abort = true; }
 141
 142                         groupfile = validParameter.validFile(parameters, "group", true);
 143                         if (groupfile == "not open") { abort = true; }
 144                         else if (groupfile == "not found") { groupfile = ""; }
 145
 146                         namefile = validParameter.validFile(parameters, "name", true);
 147                         if (namefile == "not open") { abort = true; }
 148                         else if (namefile == "not found") { namefile = ""; }
 149
 150                         alignreport = validParameter.validFile(parameters, "alignreport", true);
 151                         if (alignreport == "not open") { abort = true; }
 152                         else if (alignreport == "not found") { alignreport = ""; }
 153
 154                         //if the user changes the output directory command factory will send this info to us in the output parameter
 155                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 156                                 outputDir = "";
 157                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 158                         }
 159
 160                         //check for optional parameter and set defaults
 161                         // ...at some point should added some additional type checking...
 162                         string temp;
 163                         temp = validParameter.validFile(parameters, "start", false);            if (temp == "not found") { temp = "-1"; }
 164                         convert(temp, startPos);
 165
 166                         temp = validParameter.validFile(parameters, "end", false);                      if (temp == "not found") { temp = "-1"; }
 167                         convert(temp, endPos);
 168
 169                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
 170                         convert(temp, maxAmbig);
 171
 172                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "-1"; }
 173                         convert(temp, maxHomoP);
 174
 175                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "-1"; }
 176                         convert(temp, minLength);
 177
 178                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "-1"; }
 179                         convert(temp, maxLength);
 180
 181                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 182                         convert(temp, processors);
 183
 184                         temp = validParameter.validFile(parameters, "optimize", false); //optimizing trumps the optimized values original value
 185                         if (temp == "not found"){       temp = "";              }
 186                         else {  m->splitAtDash(temp, optimize);         }
 187
 188                         //check for invalid optimize options
 189                         set<string> validOptimizers;
 190                         validOptimizers.insert("start"); validOptimizers.insert("end"); validOptimizers.insert("maxambig"); validOptimizers.insert("maxhomop"); validOptimizers.insert("minlength"); validOptimizers.insert("maxlength");
 191
 192                         for (int i = 0; i < optimize.size(); i++) {
 193                                 if (validOptimizers.count(optimize[i]) == 0) {
 194                                         m->mothurOut(optimize[i] + " is not a valid optimizer. Valid options are start, end, maxambig, maxhomop, minlength and maxlength."); m->mothurOutEndLine();
 195                                         optimize.erase(optimize.begin()+i);
 196                                         i--;
 197                                 }
 198                         }
 199
 200                         temp = validParameter.validFile(parameters, "criteria", false); if (temp == "not found"){       temp = "90";                            }
 201                         convert(temp, criteria);
 202                 }
 203
 204         }
 205         catch(exception& e) {
 206                 m->errorOut(e, "ScreenSeqsCommand", "ScreenSeqsCommand");
 207                 exit(1);
 208         }
 209 }
 210 //**********************************************************************************************************************
 211
 212 void ScreenSeqsCommand::help(){
 213         try {
 214                 m->mothurOut("The screen.seqs command reads a fastafile and creates .....\n");
 215                 m->mothurOut("The screen.seqs command parameters are fasta, start, end, maxambig, maxhomop, minlength, maxlength, name, group, optimize, criteria and processors.\n");
 216                 m->mothurOut("The fasta parameter is required.\n");
 217                 m->mothurOut("The start parameter .... The default is -1.\n");
 218                 m->mothurOut("The end parameter .... The default is -1.\n");
 219                 m->mothurOut("The maxambig parameter .... The default is -1.\n");
 220                 m->mothurOut("The maxhomop parameter .... The default is -1.\n");
 221                 m->mothurOut("The minlength parameter .... The default is -1.\n");
 222                 m->mothurOut("The maxlength parameter .... The default is -1.\n");
 223                 m->mothurOut("The processors parameter allows you to specify the number of processors to use while running the command. The default is 1.\n");
 224                 m->mothurOut("The optimize and criteria parameters allow you set the start, end, maxabig, maxhomop, minlength and maxlength parameters relative to your set of sequences .\n");
 225                 m->mothurOut("For example optimize=start-end, criteria=90, would set the start and end values to the position 90% of your sequences started and ended.\n");
 226                 m->mothurOut("The name parameter allows you to provide a namesfile, and the group parameter allows you to provide a groupfile.\n");
 227                 m->mothurOut("The screen.seqs command should be in the following format: \n");
 228                 m->mothurOut("screen.seqs(fasta=yourFastaFile, name=youNameFile, group=yourGroupFIle, start=yourStart, end=yourEnd, maxambig=yourMaxambig,  \n");
 229                 m->mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n");
 230                 m->mothurOut("Example screen.seqs(fasta=abrecovery.fasta, name=abrecovery.names, group=abrecovery.groups, start=..., end=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
 231                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 232
 233         }
 234         catch(exception& e) {
 235                 m->errorOut(e, "ScreenSeqsCommand", "help");
 236                 exit(1);
 237         }
 238 }
 239
 240 //***************************************************************************************************************
 241
 242 ScreenSeqsCommand::~ScreenSeqsCommand(){        /*      do nothing      */      }
 243
 244 //***************************************************************************************************************
 245
 246 int ScreenSeqsCommand::execute(){
 247         try{
 248
 249                 if (abort == true) { return 0; }
 250
 251                 //if the user want to optimize we need to no the 90% mark
 252                 vector<unsigned long int> positions;
 253                 if (optimize.size() != 0) {  getSummary(positions); } //get summary is paralellized so we need to divideFile, no need to do this step twice so I moved it here
 254                 else {
 255                         positions = m->divideFile(fastafile, processors);
 256                         for (int i = 0; i < (positions.size()-1); i++) {
 257                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 258                         }
 259                 }
 260
 261                 string goodSeqFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "good" + m->getExtension(fastafile);
 262                 string badAccnosFile =  outputDir + m->getRootName(m->getSimpleName(fastafile)) + "bad.accnos";
 263
 264                 int numFastaSeqs = 0;
 265                 set<string> badSeqNames;
 266                 int start = time(NULL);
 267
 268 #ifdef USE_MPI
 269                         int pid, end, numSeqsPerProcessor;
 270                         int tag = 2001;
 271                         vector<unsigned long int> MPIPos;
 272
 273                         MPI_Status status;
 274                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 275                         MPI_Comm_size(MPI_COMM_WORLD, &processors);
 276
 277                         MPI_File inMPI;
 278                         MPI_File outMPIGood;
 279                         MPI_File outMPIBadAccnos;
 280
 281                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 282                         int inMode=MPI_MODE_RDONLY;
 283
 284                         char outGoodFilename[1024];
 285                         strcpy(outGoodFilename, goodSeqFile.c_str());
 286
 287                         char outBadAccnosFilename[1024];
 288                         strcpy(outBadAccnosFilename, badAccnosFile.c_str());
 289
 290                         char inFileName[1024];
 291                         strcpy(inFileName, fastafile.c_str());
 292
 293                         MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 294                         MPI_File_open(MPI_COMM_WORLD, outGoodFilename, outMode, MPI_INFO_NULL, &outMPIGood);
 295                         MPI_File_open(MPI_COMM_WORLD, outBadAccnosFilename, outMode, MPI_INFO_NULL, &outMPIBadAccnos);
 296
 297                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood); MPI_File_close(&outMPIBadAccnos); return 0; }
 298
 299                         if (pid == 0) { //you are the root process
 300
 301                                 MPIPos = m->setFilePosFasta(fastafile, numFastaSeqs); //fills MPIPos, returns numSeqs
 302
 303                                 //send file positions to all processes
 304                                 for(int i = 1; i < processors; i++) {
 305                                         MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 306                                         MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 307                                 }
 308
 309                                 //figure out how many sequences you have to align
 310                                 numSeqsPerProcessor = numFastaSeqs / processors;
 311                                 int startIndex =  pid * numSeqsPerProcessor;
 312                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 313                 //      cout << pid << '\t' << numSeqsPerProcessor << '\t' <<   startIndex << endl;
 314                                 //align your part
 315                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 316                 //cout << pid << " done" << endl;
 317                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos);  return 0; }
 318
 319                                 for (int i = 1; i < processors; i++) {
 320
 321                                         //get bad lists
 322                                         int badSize;
 323                                         MPI_Recv(&badSize, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 324                                         /*for (int j = 0; j < badSize; j++) {
 325                                                 int length;
 326                                                 MPI_Recv(&length, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);  //recv the length of the name
 327                                                 char* buf2 = new char[length];                                                                          //make space to recieve it
 328                                                 MPI_Recv(buf2, length, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);  //get name
 329
 330                                                 string tempBuf = buf2;
 331                                                 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
 332                                                 delete buf2;
 333
 334                                                 badSeqNames.insert(tempBuf);
 335                                         }*/
 336                                 }
 337                         }else{ //you are a child process
 338                                 MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 339                                 MPIPos.resize(numFastaSeqs+1);
 340                                 MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 341
 342                                 //figure out how many sequences you have to align
 343                                 numSeqsPerProcessor = numFastaSeqs / processors;
 344                                 int startIndex =  pid * numSeqsPerProcessor;
 345                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 346                 //cout << pid << '\t' << numSeqsPerProcessor << '\t' <<         startIndex << endl;
 347                                 //align your part
 348                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 349 //cout << pid << " done" << endl;
 350                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos); return 0; }
 351
 352                                 //send bad list
 353                                 int badSize = badSeqNames.size();
 354                                 MPI_Send(&badSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 355
 356                                 /*
 357                                 set<string>::iterator it;
 358                                 for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 359                                         string name = *it;
 360                                         int length = name.length();
 361                                         char* buf2 = new char[length];
 362                                         memcpy(buf2, name.c_str(), length);
 363
 364                                         MPI_Send(&length, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 365                                         MPI_Send(buf2, length, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 366                                 }*/
 367                         }
 368
 369                         //close files
 370                         MPI_File_close(&inMPI);
 371                         MPI_File_close(&outMPIGood);
 372                         MPI_File_close(&outMPIBadAccnos);
 373                         MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 374
 375 #else
 376
 377         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 378                         if(processors == 1){
 379                                 numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 380
 381                                 if (m->control_pressed) { remove(goodSeqFile.c_str()); return 0; }
 382
 383                         }else{
 384                                 processIDS.resize(0);
 385
 386                                 numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 387
 388                                 rename((goodSeqFile + toString(processIDS[0]) + ".temp").c_str(), goodSeqFile.c_str());
 389                                 rename((badAccnosFile + toString(processIDS[0]) + ".temp").c_str(), badAccnosFile.c_str());
 390
 391                                 //append alignment and report files
 392                                 for(int i=1;i<processors;i++){
 393                                         m->appendFiles((goodSeqFile + toString(processIDS[i]) + ".temp"), goodSeqFile);
 394                                         remove((goodSeqFile + toString(processIDS[i]) + ".temp").c_str());
 395
 396                                         m->appendFiles((badAccnosFile + toString(processIDS[i]) + ".temp"), badAccnosFile);
 397                                         remove((badAccnosFile + toString(processIDS[i]) + ".temp").c_str());
 398                                 }
 399
 400                                 if (m->control_pressed) { remove(goodSeqFile.c_str()); return 0; }
 401
 402                                 //read badSeqs in because root process doesnt know what other "bad" seqs the children found
 403                                 ifstream inBad;
 404                                 int ableToOpen = m->openInputFile(badAccnosFile, inBad, "no error");
 405
 406                                 if (ableToOpen == 0) {
 407                                         badSeqNames.clear();
 408                                         string tempName;
 409                                         while (!inBad.eof()) {
 410                                                 inBad >> tempName; m->gobble(inBad);
 411                                                 badSeqNames.insert(tempName);
 412                                         }
 413                                         inBad.close();
 414                                 }
 415                         }
 416         #else
 417                         numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 418
 419                         if (m->control_pressed) { remove(goodSeqFile.c_str()); return 0; }
 420
 421         #endif
 422
 423 #endif
 424
 425                 #ifdef USE_MPI
 426                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 427
 428                         if (pid == 0) { //only one process should fix files
 429
 430                                 //read accnos file with all names in it, process 0 just has its names
 431                                 MPI_File inMPIAccnos;
 432                                 MPI_Offset size;
 433
 434                                 char inFileName[1024];
 435                                 strcpy(inFileName, badAccnosFile.c_str());
 436
 437                                 MPI_File_open(MPI_COMM_SELF, inFileName, inMode, MPI_INFO_NULL, &inMPIAccnos);  //comm, filename, mode, info, filepointer
 438                                 MPI_File_get_size(inMPIAccnos, &size);
 439
 440                                 char* buffer = new char[size];
 441                                 MPI_File_read(inMPIAccnos, buffer, size, MPI_CHAR, &status);
 442
 443                                 string tempBuf = buffer;
 444                                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
 445                                 istringstream iss (tempBuf,istringstream::in);
 446
 447                                 delete buffer;
 448                                 MPI_File_close(&inMPIAccnos);
 449
 450                                 badSeqNames.clear();
 451                                 string tempName;
 452                                 while (!iss.eof()) {
 453                                         iss >> tempName; m->gobble(iss);
 454                                         badSeqNames.insert(tempName);
 455                                 }
 456                 #endif
 457
 458                 if(namefile != "" && groupfile != "")   {
 459                         screenNameGroupFile(badSeqNames);
 460                         if (m->control_pressed) {  remove(goodSeqFile.c_str()); return 0; }
 461                 }else if(namefile != "")        {
 462                         screenNameGroupFile(badSeqNames);
 463                         if (m->control_pressed) {  remove(goodSeqFile.c_str());  return 0; }
 464                 }else if(groupfile != "")                               {       screenGroupFile(badSeqNames);           }       // this screens just the group
 465
 466                 if (m->control_pressed) { remove(goodSeqFile.c_str());  return 0; }
 467
 468                 if(alignreport != "")                                   {       screenAlignReport(badSeqNames);         }
 469
 470                 if (m->control_pressed) { remove(goodSeqFile.c_str());  return 0; }
 471
 472                 #ifdef USE_MPI
 473                         }
 474                 #endif
 475
 476                 m->mothurOutEndLine();
 477                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 478                 m->mothurOut(goodSeqFile); m->mothurOutEndLine();       outputTypes["fasta"].push_back(goodSeqFile);
 479                 m->mothurOut(badAccnosFile); m->mothurOutEndLine();      outputTypes["accnos"].push_back(badAccnosFile);
 480                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 481                 m->mothurOutEndLine();
 482                 m->mothurOutEndLine();
 483
 484                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to screen " + toString(numFastaSeqs) + " sequences.");
 485                 m->mothurOutEndLine();
 486
 487                 return 0;
 488         }
 489         catch(exception& e) {
 490                 m->errorOut(e, "ScreenSeqsCommand", "execute");
 491                 exit(1);
 492         }
 493 }
 494
 495 //***************************************************************************************************************
 496
 497 int ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
 498         try {
 499                 ifstream inputNames;
 500                 m->openInputFile(namefile, inputNames);
 501                 set<string> badSeqGroups;
 502                 string seqName, seqList, group;
 503                 set<string>::iterator it;
 504
 505                 string goodNameFile = outputDir + m->getRootName(m->getSimpleName(namefile)) + "good" + m->getExtension(namefile);
 506                 outputNames.push_back(goodNameFile);  outputTypes["name"].push_back(goodNameFile);
 507
 508                 ofstream goodNameOut;   m->openOutputFile(goodNameFile, goodNameOut);
 509
 510                 while(!inputNames.eof()){
 511                         if (m->control_pressed) { goodNameOut.close();  inputNames.close(); remove(goodNameFile.c_str());  return 0; }
 512
 513                         inputNames >> seqName >> seqList;
 514                         it = badSeqNames.find(seqName);
 515
 516                         if(it != badSeqNames.end()){
 517                                 badSeqNames.erase(it);
 518
 519                                 if(namefile != ""){
 520                                         int start = 0;
 521                                         for(int i=0;i<seqList.length();i++){
 522                                                 if(seqList[i] == ','){
 523                                                         badSeqGroups.insert(seqList.substr(start,i-start));
 524                                                         start = i+1;
 525                                                 }
 526                                         }
 527                                         badSeqGroups.insert(seqList.substr(start,seqList.length()-start));
 528                                 }
 529                         }
 530                         else{
 531                                 goodNameOut << seqName << '\t' << seqList << endl;
 532                         }
 533                         m->gobble(inputNames);
 534                 }
 535                 inputNames.close();
 536                 goodNameOut.close();
 537
 538                 //we were unable to remove some of the bad sequences
 539                 if (badSeqNames.size() != 0) {
 540                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 541                                 m->mothurOut("Your namefile does not include the sequence " + *it + " please correct.");
 542                                 m->mothurOutEndLine();
 543                         }
 544                 }
 545
 546                 if(groupfile != ""){
 547
 548                         ifstream inputGroups;
 549                         m->openInputFile(groupfile, inputGroups);
 550
 551                         string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
 552                         outputNames.push_back(goodGroupFile);   outputTypes["group"].push_back(goodGroupFile);
 553
 554                         ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 555
 556                         while(!inputGroups.eof()){
 557                                 if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); remove(goodNameFile.c_str());  remove(goodGroupFile.c_str()); return 0; }
 558
 559                                 inputGroups >> seqName >> group;
 560
 561                                 it = badSeqGroups.find(seqName);
 562
 563                                 if(it != badSeqGroups.end()){
 564                                         badSeqGroups.erase(it);
 565                                 }
 566                                 else{
 567                                         goodGroupOut << seqName << '\t' << group << endl;
 568                                 }
 569                                 m->gobble(inputGroups);
 570                         }
 571                         inputGroups.close();
 572                         goodGroupOut.close();
 573
 574                         //we were unable to remove some of the bad sequences
 575                         if (badSeqGroups.size() != 0) {
 576                                 for (it = badSeqGroups.begin(); it != badSeqGroups.end(); it++) {
 577                                         m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 578                                         m->mothurOutEndLine();
 579                                 }
 580                         }
 581                 }
 582
 583                 return 0;
 584
 585         }
 586         catch(exception& e) {
 587                 m->errorOut(e, "ScreenSeqsCommand", "screenNameGroupFile");
 588                 exit(1);
 589         }
 590 }
 591 //***************************************************************************************************************
 592 int ScreenSeqsCommand::getSummary(vector<unsigned long int>& positions){
 593         try {
 594
 595                 vector<int> startPosition;
 596                 vector<int> endPosition;
 597                 vector<int> seqLength;
 598                 vector<int> ambigBases;
 599                 vector<int> longHomoPolymer;
 600
 601                 vector<unsigned long int> positions = m->divideFile(fastafile, processors);
 602
 603                 for (int i = 0; i < (positions.size()-1); i++) {
 604                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 605                 }
 606
 607                 int numSeqs = 0;
 608                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 609                         if(processors == 1){
 610                                 numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 611                         }else{
 612                                 numSeqs = createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile);
 613                         }
 614
 615                         if (m->control_pressed) {  return 0; }
 616                 #else
 617                         numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 618                         if (m->control_pressed) {  return 0; }
 619                 #endif
 620
 621                 sort(startPosition.begin(), startPosition.end());
 622                 sort(endPosition.begin(), endPosition.end());
 623                 sort(seqLength.begin(), seqLength.end());
 624                 sort(ambigBases.begin(), ambigBases.end());
 625                 sort(longHomoPolymer.begin(), longHomoPolymer.end());
 626
 627                 int criteriaPercentile  = int(numSeqs * (criteria / (float) 100));
 628
 629                 for (int i = 0; i < optimize.size(); i++) {
 630                         if (optimize[i] == "start") { startPos = startPosition[criteriaPercentile]; m->mothurOut("Optimizing start to " + toString(startPos) + "."); m->mothurOutEndLine(); }
 631                         else if (optimize[i] == "end") { int endcriteriaPercentile = int(numSeqs * ((100 - criteria) / (float) 100));  endPos = endPosition[endcriteriaPercentile]; m->mothurOut("Optimizing end to " + toString(endPos) + "."); m->mothurOutEndLine();}
 632                         else if (optimize[i] == "maxambig") { maxAmbig = ambigBases[criteriaPercentile]; m->mothurOut("Optimizing maxambig to " + toString(maxAmbig) + "."); m->mothurOutEndLine(); }
 633                         else if (optimize[i] == "maxhomop") { maxHomoP = longHomoPolymer[criteriaPercentile]; m->mothurOut("Optimizing maxhomop to " + toString(maxHomoP) + "."); m->mothurOutEndLine(); }
 634                         else if (optimize[i] == "minlength") { int mincriteriaPercentile = int(numSeqs * ((100 - criteria) / (float) 100)); minLength = seqLength[mincriteriaPercentile]; m->mothurOut("Optimizing minlength to " + toString(minLength) + "."); m->mothurOutEndLine(); }
 635                         else if (optimize[i] == "maxlength") { maxLength = seqLength[criteriaPercentile]; m->mothurOut("Optimizing maxlength to " + toString(maxLength) + "."); m->mothurOutEndLine(); }
 636                 }
 637
 638                 return 0;
 639         }
 640         catch(exception& e) {
 641                 m->errorOut(e, "ScreenSeqsCommand", "getSummary");
 642                 exit(1);
 643         }
 644 }
 645 /**************************************************************************************/
 646 int ScreenSeqsCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, linePair* filePos) {
 647         try {
 648
 649                 ifstream in;
 650                 m->openInputFile(filename, in);
 651
 652                 in.seekg(filePos->start);
 653
 654                 bool done = false;
 655                 int count = 0;
 656
 657                 while (!done) {
 658
 659                         if (m->control_pressed) { in.close(); return 1; }
 660
 661                         Sequence current(in); m->gobble(in);
 662
 663                         if (current.getName() != "") {
 664                                 startPosition.push_back(current.getStartPos());
 665                                 endPosition.push_back(current.getEndPos());
 666                                 seqLength.push_back(current.getNumBases());
 667                                 ambigBases.push_back(current.getAmbigBases());
 668                                 longHomoPolymer.push_back(current.getLongHomoPolymer());
 669                                 count++;
 670                         }
 671
 672                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 673                                 unsigned long int pos = in.tellg();
 674                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 675                         #else
 676                                 if (in.eof()) { break; }
 677                         #endif
 678
 679                 }
 680
 681                 in.close();
 682
 683                 return count;
 684         }
 685         catch(exception& e) {
 686                 m->errorOut(e, "ScreenSeqsCommand", "driverCreateSummary");
 687                 exit(1);
 688         }
 689 }
 690 /**************************************************************************************************/
 691 int ScreenSeqsCommand::createProcessesCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename) {
 692         try {
 693 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 694                 int process = 1;
 695                 int num = 0;
 696                 processIDS.clear();
 697
 698                 //loop through and create all the processes you want
 699                 while (process != processors) {
 700                         int pid = fork();
 701
 702                         if (pid > 0) {
 703                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 704                                 process++;
 705                         }else if (pid == 0){
 706                                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[process]);
 707
 708                                 //pass numSeqs to parent
 709                                 ofstream out;
 710                                 string tempFile = fastafile + toString(getpid()) + ".num.temp";
 711                                 m->openOutputFile(tempFile, out);
 712
 713                                 out << num << endl;
 714                                 for (int k = 0; k < startPosition.size(); k++)          {               out << startPosition[k] << '\t'; }  out << endl;
 715                                 for (int k = 0; k < endPosition.size(); k++)            {               out << endPosition[k] << '\t'; }  out << endl;
 716                                 for (int k = 0; k < seqLength.size(); k++)                      {               out << seqLength[k] << '\t'; }  out << endl;
 717                                 for (int k = 0; k < ambigBases.size(); k++)                     {               out << ambigBases[k] << '\t'; }  out << endl;
 718                                 for (int k = 0; k < longHomoPolymer.size(); k++)        {               out << longHomoPolymer[k] << '\t'; }  out << endl;
 719
 720                                 out.close();
 721
 722                                 exit(0);
 723                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 724                 }
 725
 726                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 727
 728                 //force parent to wait until all the processes are done
 729                 for (int i=0;i<processIDS.size();i++) {
 730                         int temp = processIDS[i];
 731                         wait(&temp);
 732                 }
 733
 734                 //parent reads in and combine Filter info
 735                 for (int i = 0; i < processIDS.size(); i++) {
 736                         string tempFilename = fastafile + toString(processIDS[i]) + ".num.temp";
 737                         ifstream in;
 738                         m->openInputFile(tempFilename, in);
 739
 740                         int temp, tempNum;
 741                         in >> tempNum; m->gobble(in); num += tempNum;
 742                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; startPosition.push_back(temp);              }               m->gobble(in);
 743                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; endPosition.push_back(temp);                }               m->gobble(in);
 744                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; seqLength.push_back(temp);                  }               m->gobble(in);
 745                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; ambigBases.push_back(temp);                 }               m->gobble(in);
 746                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; longHomoPolymer.push_back(temp);    }               m->gobble(in);
 747
 748                         in.close();
 749                         remove(tempFilename.c_str());
 750                 }
 751
 752                 return num;
 753 #endif
 754         }
 755         catch(exception& e) {
 756                 m->errorOut(e, "ScreenSeqsCommand", "createProcessesCreateSummary");
 757                 exit(1);
 758         }
 759 }
 760
 761 //***************************************************************************************************************
 762
 763 int ScreenSeqsCommand::screenGroupFile(set<string> badSeqNames){
 764         try {
 765                 ifstream inputGroups;
 766                 m->openInputFile(groupfile, inputGroups);
 767                 string seqName, group;
 768                 set<string>::iterator it;
 769
 770                 string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
 771                 outputNames.push_back(goodGroupFile);  outputTypes["group"].push_back(goodGroupFile);
 772                 ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 773
 774                 while(!inputGroups.eof()){
 775                         if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); remove(goodGroupFile.c_str()); return 0; }
 776
 777                         inputGroups >> seqName >> group;
 778                         it = badSeqNames.find(seqName);
 779
 780                         if(it != badSeqNames.end()){
 781                                 badSeqNames.erase(it);
 782                         }
 783                         else{
 784                                 goodGroupOut << seqName << '\t' << group << endl;
 785                         }
 786                         m->gobble(inputGroups);
 787                 }
 788
 789                 if (m->control_pressed) { goodGroupOut.close();  inputGroups.close(); remove(goodGroupFile.c_str());  return 0; }
 790
 791                 //we were unable to remove some of the bad sequences
 792                 if (badSeqNames.size() != 0) {
 793                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 794                                 m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 795                                 m->mothurOutEndLine();
 796                         }
 797                 }
 798
 799                 inputGroups.close();
 800                 goodGroupOut.close();
 801
 802                 if (m->control_pressed) { remove(goodGroupFile.c_str());   }
 803
 804                 return 0;
 805
 806         }
 807         catch(exception& e) {
 808                 m->errorOut(e, "ScreenSeqsCommand", "screenGroupFile");
 809                 exit(1);
 810         }
 811 }
 812
 813 //***************************************************************************************************************
 814
 815 int ScreenSeqsCommand::screenAlignReport(set<string> badSeqNames){
 816         try {
 817                 ifstream inputAlignReport;
 818                 m->openInputFile(alignreport, inputAlignReport);
 819                 string seqName, group;
 820                 set<string>::iterator it;
 821
 822                 string goodAlignReportFile = outputDir + m->getRootName(m->getSimpleName(alignreport)) + "good" + m->getExtension(alignreport);
 823                 outputNames.push_back(goodAlignReportFile);  outputTypes["alignreport"].push_back(goodAlignReportFile);
 824                 ofstream goodAlignReportOut;    m->openOutputFile(goodAlignReportFile, goodAlignReportOut);
 825
 826                 while (!inputAlignReport.eof()) {               //      need to copy header
 827                         char c = inputAlignReport.get();
 828                         goodAlignReportOut << c;
 829                         if (c == 10 || c == 13){        break;  }
 830                 }
 831
 832                 while(!inputAlignReport.eof()){
 833                         if (m->control_pressed) { goodAlignReportOut.close(); inputAlignReport.close(); remove(goodAlignReportFile.c_str()); return 0; }
 834
 835                         inputAlignReport >> seqName;
 836                         it = badSeqNames.find(seqName);
 837                         string line;
 838                         while (!inputAlignReport.eof()) {               //      need to copy header
 839                                 char c = inputAlignReport.get();
 840                                 line += c;
 841                                 if (c == 10 || c == 13){        break;  }
 842                         }
 843
 844                         if(it != badSeqNames.end()){
 845                                 badSeqNames.erase(it);
 846                         }
 847                         else{
 848                                 goodAlignReportOut << seqName << '\t' << line;
 849                         }
 850                         m->gobble(inputAlignReport);
 851                 }
 852
 853                 if (m->control_pressed) { goodAlignReportOut.close();  inputAlignReport.close(); remove(goodAlignReportFile.c_str());  return 0; }
 854
 855                 //we were unable to remove some of the bad sequences
 856                 if (badSeqNames.size() != 0) {
 857                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 858                                 m->mothurOut("Your alignreport file does not include the sequence " + *it + " please correct.");
 859                                 m->mothurOutEndLine();
 860                         }
 861                 }
 862
 863                 inputAlignReport.close();
 864                 goodAlignReportOut.close();
 865
 866                 if (m->control_pressed) {  remove(goodAlignReportFile.c_str());  return 0; }
 867
 868                 return 0;
 869
 870         }
 871         catch(exception& e) {
 872                 m->errorOut(e, "ScreenSeqsCommand", "screenAlignReport");
 873                 exit(1);
 874         }
 875
 876 }
 877 //**********************************************************************************************************************
 878
 879 int ScreenSeqsCommand::driver(linePair* filePos, string goodFName, string badAccnosFName, string filename, set<string>& badSeqNames){
 880         try {
 881                 ofstream goodFile;
 882                 m->openOutputFile(goodFName, goodFile);
 883
 884                 ofstream badAccnosFile;
 885                 m->openOutputFile(badAccnosFName, badAccnosFile);
 886
 887                 ifstream inFASTA;
 888                 m->openInputFile(filename, inFASTA);
 889
 890                 inFASTA.seekg(filePos->start);
 891
 892                 bool done = false;
 893                 int count = 0;
 894
 895                 while (!done) {
 896
 897                         if (m->control_pressed) {  return 0; }
 898
 899                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
 900                         if (currSeq.getName() != "") {
 901                                 bool goodSeq = 1;               //      innocent until proven guilty
 902                                 if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
 903                                 if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
 904                                 if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
 905                                 if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
 906                                 if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
 907                                 if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
 908
 909                                 if(goodSeq == 1){
 910                                         currSeq.printSequence(goodFile);
 911                                 }
 912                                 else{
 913                                         badAccnosFile << currSeq.getName() << endl;
 914                                         badSeqNames.insert(currSeq.getName());
 915                                 }
 916                         count++;
 917                         }
 918
 919                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 920                                 unsigned long int pos = inFASTA.tellg();
 921                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 922                         #else
 923                                 if (inFASTA.eof()) { break; }
 924                         #endif
 925
 926                         //report progress
 927                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 928                 }
 929                 //report progress
 930                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 931
 932
 933                 goodFile.close();
 934                 inFASTA.close();
 935                 badAccnosFile.close();
 936
 937                 return count;
 938         }
 939         catch(exception& e) {
 940                 m->errorOut(e, "ScreenSeqsCommand", "driver");
 941                 exit(1);
 942         }
 943 }
 944 //**********************************************************************************************************************
 945 #ifdef USE_MPI
 946 int ScreenSeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& goodFile, MPI_File& badAccnosFile, vector<unsigned long int>& MPIPos, set<string>& badSeqNames){
 947         try {
 948                 string outputString = "";
 949                 MPI_Status statusGood;
 950                 MPI_Status statusBadAccnos;
 951                 MPI_Status status;
 952                 int pid;
 953                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 954
 955                 for(int i=0;i<num;i++){
 956
 957                         if (m->control_pressed) {  return 0; }
 958
 959                         //read next sequence
 960                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 961
 962                         char* buf4 = new char[length];
 963                         memcpy(buf4, outputString.c_str(), length);
 964
 965                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 966
 967                         string tempBuf = buf4;  delete buf4;
 968                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 969                         istringstream iss (tempBuf,istringstream::in);
 970
 971                         Sequence currSeq(iss);
 972
 973                         //process seq
 974                         if (currSeq.getName() != "") {
 975                                 bool goodSeq = 1;               //      innocent until proven guilty
 976                                 if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
 977                                 if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
 978                                 if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
 979                                 if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
 980                                 if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
 981                                 if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
 982
 983                                 if(goodSeq == 1){
 984                                         outputString =  ">" + currSeq.getName() + "\n" + currSeq.getAligned() + "\n";
 985
 986                                         //print good seq
 987                                         length = outputString.length();
 988                                         char* buf2 = new char[length];
 989                                         memcpy(buf2, outputString.c_str(), length);
 990
 991                                         MPI_File_write_shared(goodFile, buf2, length, MPI_CHAR, &statusGood);
 992                                         delete buf2;
 993                                 }
 994                                 else{
 995
 996                                         badSeqNames.insert(currSeq.getName());
 997
 998                                         //write to bad accnos file
 999                                         outputString = currSeq.getName() + "\n";
1000
1001                                         length = outputString.length();
1002                                         char* buf3 = new char[length];
1003                                         memcpy(buf3, outputString.c_str(), length);
1004
1005                                         MPI_File_write_shared(badAccnosFile, buf3, length, MPI_CHAR, &statusBadAccnos);
1006                                         delete buf3;
1007                                 }
1008                         }
1009                 }
1010
1011                 return 1;
1012         }
1013         catch(exception& e) {
1014                 m->errorOut(e, "ScreenSeqsCommand", "driverMPI");
1015                 exit(1);
1016         }
1017 }
1018 #endif
1019 /**************************************************************************************************/
1020
1021 int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, string filename, set<string>& badSeqNames) {
1022         try {
1023 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1024                 int process = 0;
1025                 int num = 0;
1026
1027                 //loop through and create all the processes you want
1028                 while (process != processors) {
1029                         int pid = fork();
1030
1031                         if (pid > 0) {
1032                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1033                                 process++;
1034                         }else if (pid == 0){
1035                                 num = driver(lines[process], goodFileName + toString(getpid()) + ".temp", badAccnos + toString(getpid()) + ".temp", filename, badSeqNames);
1036
1037                                 //pass numSeqs to parent
1038                                 ofstream out;
1039                                 string tempFile = filename + toString(getpid()) + ".num.temp";
1040                                 m->openOutputFile(tempFile, out);
1041                                 out << num << endl;
1042                                 out.close();
1043
1044                                 exit(0);
1045                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
1046                 }
1047
1048                 //force parent to wait until all the processes are done
1049                 for (int i=0;i<processors;i++) {
1050                         int temp = processIDS[i];
1051                         wait(&temp);
1052                 }
1053
1054                 for (int i = 0; i < processIDS.size(); i++) {
1055                         ifstream in;
1056                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
1057                         m->openInputFile(tempFile, in);
1058                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1059                         in.close(); remove(tempFile.c_str());
1060                 }
1061
1062                 return num;
1063 #endif
1064         }
1065         catch(exception& e) {
1066                 m->errorOut(e, "ScreenSeqsCommand", "createProcesses");
1067                 exit(1);
1068         }
1069 }
1070
1071 //***************************************************************************************************************
1072
1073