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 "counttable.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ScreenSeqsCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","fasta",false,true,true); parameters.push_back(pfasta);
  17         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "none","name",false,false,true); parameters.push_back(pname);
  18         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "none", "none","count",false,false,true); parameters.push_back(pcount);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none","group",false,false,true); parameters.push_back(pgroup);
  20                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "none","qfile",false,false); parameters.push_back(pqfile);
  21                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "none", "none","alignreport",false,false); parameters.push_back(palignreport);
  22                 CommandParameter ptax("taxonomy", "InputTypes", "", "", "none", "none", "none","taxonomy",false,false); parameters.push_back(ptax);
  23                 CommandParameter pstart("start", "Number", "", "-1", "", "", "","",false,false,true); parameters.push_back(pstart);
  24                 CommandParameter pend("end", "Number", "", "-1", "", "", "","",false,false,true); parameters.push_back(pend);
  25                 CommandParameter pmaxambig("maxambig", "Number", "", "-1", "", "", "","",false,false); parameters.push_back(pmaxambig);
  26                 CommandParameter pmaxhomop("maxhomop", "Number", "", "-1", "", "", "","",false,false); parameters.push_back(pmaxhomop);
  27                 CommandParameter pminlength("minlength", "Number", "", "-1", "", "", "","",false,false); parameters.push_back(pminlength);
  28                 CommandParameter pmaxlength("maxlength", "Number", "", "-1", "", "", "","",false,false); parameters.push_back(pmaxlength);
  29                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  30                 CommandParameter pcriteria("criteria", "Number", "", "90", "", "", "","",false,false); parameters.push_back(pcriteria);
  31                 CommandParameter poptimize("optimize", "Multiple", "none-start-end-maxambig-maxhomop-minlength-maxlength", "none", "", "", "","",true,false); parameters.push_back(poptimize);
  32                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  33                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  34
  35                 vector<string> myArray;
  36                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  37                 return myArray;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "ScreenSeqsCommand", "setParameters");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 string ScreenSeqsCommand::getHelpString(){
  46         try {
  47                 string helpString = "";
  48                 helpString += "The screen.seqs command reads a fastafile and screens sequences.\n";
  49                 helpString += "The screen.seqs command parameters are fasta, start, end, maxambig, maxhomop, minlength, maxlength, name, group, count, qfile, alignreport, taxonomy, optimize, criteria and processors.\n";
  50                 helpString += "The fasta parameter is required.\n";
  51                 helpString += "The alignreport and taxonomy parameters allow you to remove bad seqs from taxonomy and alignreport files.\n";
  52                 helpString += "The start parameter is used to set a position the \"good\" sequences must start by. The default is -1.\n";
  53                 helpString += "The end parameter is used to set a position the \"good\" sequences must end after. The default is -1.\n";
  54                 helpString += "The maxambig parameter allows you to set the maximum number of ambigious bases allowed. The default is -1.\n";
  55                 helpString += "The maxhomop parameter allows you to set a maximum homopolymer length. \n";
  56                 helpString += "The minlength parameter allows you to set and minimum sequence length. \n";
  57                 helpString += "The maxlength parameter allows you to set and maximum sequence length. \n";
  58                 helpString += "The processors parameter allows you to specify the number of processors to use while running the command. The default is 1.\n";
  59                 helpString += "The optimize and criteria parameters allow you set the start, end, maxabig, maxhomop, minlength and maxlength parameters relative to your set of sequences .\n";
  60                 helpString += "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";
  61                 helpString += "The name parameter allows you to provide a namesfile, and the group parameter allows you to provide a groupfile.\n";
  62                 helpString += "The screen.seqs command should be in the following format: \n";
  63                 helpString += "screen.seqs(fasta=yourFastaFile, name=youNameFile, group=yourGroupFIle, start=yourStart, end=yourEnd, maxambig=yourMaxambig,  \n";
  64                 helpString += "maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  \n";
  65                 helpString += "Example screen.seqs(fasta=abrecovery.fasta, name=abrecovery.names, group=abrecovery.groups, start=..., end=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n";
  66                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  67                 return helpString;
  68         }
  69         catch(exception& e) {
  70                 m->errorOut(e, "ScreenSeqsCommand", "getHelpString");
  71                 exit(1);
  72         }
  73 }
  74 //**********************************************************************************************************************
  75 string ScreenSeqsCommand::getOutputPattern(string type) {
  76     try {
  77         string pattern = "";
  78
  79         if (type == "fasta")            {   pattern = "[filename],good,[extension]";    }
  80         else if (type == "taxonomy")    {   pattern = "[filename],good,[extension]";    }
  81         else if (type == "name")        {   pattern = "[filename],good,[extension]";    }
  82         else if (type == "group")       {   pattern = "[filename],good,[extension]";    }
  83         else if (type == "count")       {   pattern = "[filename],good,[extension]";    }
  84         else if (type == "accnos")      {   pattern = "[filename],bad.accnos";          }
  85         else if (type == "qfile")       {   pattern = "[filename],good,[extension]";    }
  86         else if (type == "alignreport")      {   pattern = "[filename],good.align.report";    }
  87         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  88
  89         return pattern;
  90     }
  91     catch(exception& e) {
  92         m->errorOut(e, "ScreenSeqsCommand", "getOutputPattern");
  93         exit(1);
  94     }
  95 }
  96 //**********************************************************************************************************************
  97 ScreenSeqsCommand::ScreenSeqsCommand(){
  98         try {
  99                 abort = true; calledHelp = true;
 100                 setParameters();
 101                 vector<string> tempOutNames;
 102                 outputTypes["fasta"] = tempOutNames;
 103                 outputTypes["name"] = tempOutNames;
 104                 outputTypes["group"] = tempOutNames;
 105                 outputTypes["alignreport"] = tempOutNames;
 106                 outputTypes["accnos"] = tempOutNames;
 107                 outputTypes["qfile"] = tempOutNames;
 108                 outputTypes["taxonomy"] = tempOutNames;
 109         outputTypes["count"] = tempOutNames;
 110         }
 111         catch(exception& e) {
 112                 m->errorOut(e, "ScreenSeqsCommand", "ScreenSeqsCommand");
 113                 exit(1);
 114         }
 115 }
 116 //***************************************************************************************************************
 117
 118 ScreenSeqsCommand::ScreenSeqsCommand(string option)  {
 119         try {
 120                 abort = false; calledHelp = false;
 121
 122                 //allow user to run help
 123                 if(option == "help") { help(); abort = true; calledHelp = true; }
 124                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 125
 126                 else {
 127                         vector<string> myArray = setParameters();
 128
 129                         OptionParser parser(option);
 130                         map<string,string> parameters = parser.getParameters();
 131
 132                         ValidParameters validParameter("screen.seqs");
 133                         map<string,string>::iterator it;
 134
 135                         //check to make sure all parameters are valid for command
 136                         for (it = parameters.begin(); it != parameters.end(); it++) {
 137                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 138                         }
 139
 140                         //initialize outputTypes
 141                         vector<string> tempOutNames;
 142                         outputTypes["fasta"] = tempOutNames;
 143                         outputTypes["name"] = tempOutNames;
 144                         outputTypes["group"] = tempOutNames;
 145                         outputTypes["alignreport"] = tempOutNames;
 146                         outputTypes["accnos"] = tempOutNames;
 147                         outputTypes["qfile"] = tempOutNames;
 148                         outputTypes["taxonomy"] = tempOutNames;
 149             outputTypes["count"] = tempOutNames;
 150
 151                         //if the user changes the input directory command factory will send this info to us in the output parameter
 152                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 153                         if (inputDir == "not found"){   inputDir = "";          }
 154                         else {
 155                                 string path;
 156                                 it = parameters.find("fasta");
 157                                 //user has given a template file
 158                                 if(it != parameters.end()){
 159                                         path = m->hasPath(it->second);
 160                                         //if the user has not given a path then, add inputdir. else leave path alone.
 161                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 162                                 }
 163
 164                                 it = parameters.find("group");
 165                                 //user has given a template file
 166                                 if(it != parameters.end()){
 167                                         path = m->hasPath(it->second);
 168                                         //if the user has not given a path then, add inputdir. else leave path alone.
 169                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 170                                 }
 171
 172                                 it = parameters.find("name");
 173                                 //user has given a template file
 174                                 if(it != parameters.end()){
 175                                         path = m->hasPath(it->second);
 176                                         //if the user has not given a path then, add inputdir. else leave path alone.
 177                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 178                                 }
 179
 180                                 it = parameters.find("alignreport");
 181                                 //user has given a template file
 182                                 if(it != parameters.end()){
 183                                         path = m->hasPath(it->second);
 184                                         //if the user has not given a path then, add inputdir. else leave path alone.
 185                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 186                                 }
 187
 188                                 it = parameters.find("qfile");
 189                                 //user has given a template file
 190                                 if(it != parameters.end()){
 191                                         path = m->hasPath(it->second);
 192                                         //if the user has not given a path then, add inputdir. else leave path alone.
 193                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 194                                 }
 195
 196                                 it = parameters.find("taxonomy");
 197                                 //user has given a template file
 198                                 if(it != parameters.end()){
 199                                         path = m->hasPath(it->second);
 200                                         //if the user has not given a path then, add inputdir. else leave path alone.
 201                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 202                                 }
 203
 204                 it = parameters.find("count");
 205                                 //user has given a template file
 206                                 if(it != parameters.end()){
 207                                         path = m->hasPath(it->second);
 208                                         //if the user has not given a path then, add inputdir. else leave path alone.
 209                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 210                                 }
 211                         }
 212
 213                         //check for required parameters
 214                         fastafile = validParameter.validFile(parameters, "fasta", true);
 215                         if (fastafile == "not found") {
 216                                 fastafile = m->getFastaFile();
 217                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 218                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 219                         }
 220                         else if (fastafile == "not open") { abort = true; }
 221                         else { m->setFastaFile(fastafile); }
 222
 223                         groupfile = validParameter.validFile(parameters, "group", true);
 224                         if (groupfile == "not open") { abort = true; }
 225                         else if (groupfile == "not found") { groupfile = ""; }
 226                         else { m->setGroupFile(groupfile); }
 227
 228                         qualfile = validParameter.validFile(parameters, "qfile", true);
 229                         if (qualfile == "not open") { abort = true; }
 230                         else if (qualfile == "not found") { qualfile = ""; }
 231                         else { m->setQualFile(qualfile); }
 232
 233                         namefile = validParameter.validFile(parameters, "name", true);
 234                         if (namefile == "not open") { namefile = ""; abort = true; }
 235                         else if (namefile == "not found") { namefile = ""; }
 236                         else { m->setNameFile(namefile); }
 237
 238             countfile = validParameter.validFile(parameters, "count", true);
 239                         if (countfile == "not open") { countfile = ""; abort = true; }
 240                         else if (countfile == "not found") { countfile = "";  }
 241                         else { m->setCountTableFile(countfile); }
 242
 243             if ((namefile != "") && (countfile != "")) {
 244                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 245             }
 246
 247             if ((groupfile != "") && (countfile != "")) {
 248                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 249             }
 250
 251                         alignreport = validParameter.validFile(parameters, "alignreport", true);
 252                         if (alignreport == "not open") { abort = true; }
 253                         else if (alignreport == "not found") { alignreport = ""; }
 254
 255                         taxonomy = validParameter.validFile(parameters, "taxonomy", true);
 256                         if (taxonomy == "not open") { abort = true; }
 257                         else if (taxonomy == "not found") { taxonomy = ""; }
 258
 259                         //if the user changes the output directory command factory will send this info to us in the output parameter
 260                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 261                                 outputDir = "";
 262                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 263                         }
 264
 265                         //check for optional parameter and set defaults
 266                         // ...at some point should added some additional type checking...
 267                         string temp;
 268                         temp = validParameter.validFile(parameters, "start", false);            if (temp == "not found") { temp = "-1"; }
 269                         m->mothurConvert(temp, startPos);
 270
 271                         temp = validParameter.validFile(parameters, "end", false);                      if (temp == "not found") { temp = "-1"; }
 272                         m->mothurConvert(temp, endPos);
 273
 274                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
 275                         m->mothurConvert(temp, maxAmbig);
 276
 277                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "-1"; }
 278                         m->mothurConvert(temp, maxHomoP);
 279
 280                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "-1"; }
 281                         m->mothurConvert(temp, minLength);
 282
 283                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "-1"; }
 284                         m->mothurConvert(temp, maxLength);
 285
 286                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 287                         m->setProcessors(temp);
 288                         m->mothurConvert(temp, processors);
 289
 290                         temp = validParameter.validFile(parameters, "optimize", false); //optimizing trumps the optimized values original value
 291                         if (temp == "not found"){       temp = "none";          }
 292                         m->splitAtDash(temp, optimize);
 293
 294                         //check for invalid optimize options
 295                         set<string> validOptimizers;
 296                         validOptimizers.insert("none"); validOptimizers.insert("start"); validOptimizers.insert("end"); validOptimizers.insert("maxambig"); validOptimizers.insert("maxhomop"); validOptimizers.insert("minlength"); validOptimizers.insert("maxlength");
 297                         for (int i = 0; i < optimize.size(); i++) {
 298                                 if (validOptimizers.count(optimize[i]) == 0) {
 299                                         m->mothurOut(optimize[i] + " is not a valid optimizer. Valid options are start, end, maxambig, maxhomop, minlength and maxlength."); m->mothurOutEndLine();
 300                                         optimize.erase(optimize.begin()+i);
 301                                         i--;
 302                                 }
 303                         }
 304
 305                         if (optimize.size() == 1) { if (optimize[0] == "none") { optimize.clear(); } }
 306
 307                         temp = validParameter.validFile(parameters, "criteria", false); if (temp == "not found"){       temp = "90";                            }
 308                         m->mothurConvert(temp, criteria);
 309
 310                         if (countfile == "") {
 311                 if (namefile == "") {
 312                     vector<string> files; files.push_back(fastafile);
 313                     parser.getNameFile(files);
 314                 }
 315             }
 316                 }
 317
 318         }
 319         catch(exception& e) {
 320                 m->errorOut(e, "ScreenSeqsCommand", "ScreenSeqsCommand");
 321                 exit(1);
 322         }
 323 }
 324 //***************************************************************************************************************
 325
 326 int ScreenSeqsCommand::execute(){
 327         try{
 328
 329                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 330
 331                 //if the user want to optimize we need to know the 90% mark
 332                 vector<unsigned long long> positions;
 333                 if (optimize.size() != 0) {  //get summary is paralellized so we need to divideFile, no need to do this step twice so I moved it here
 334                         //use the namefile to optimize correctly
 335                         if (namefile != "") { nameMap = m->readNames(namefile); }
 336             else if (countfile != "") {
 337                 CountTable ct;
 338                 ct.readTable(countfile);
 339                 nameMap = ct.getNameMap();
 340             }
 341                         getSummary(positions);
 342                 }
 343                 else {
 344                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 345                 positions = m->divideFile(fastafile, processors);
 346                 for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(linePair(positions[i], positions[(i+1)])); }
 347                         #else
 348                 if(processors == 1){ lines.push_back(linePair(0, 1000));  }
 349                 else {
 350                     int numFastaSeqs = 0;
 351                     positions = m->setFilePosFasta(fastafile, numFastaSeqs);
 352                     if (positions.size() < processors) { processors = positions.size(); }
 353
 354                     //figure out how many sequences you have to process
 355                     int numSeqsPerProcessor = numFastaSeqs / processors;
 356                     for (int i = 0; i < processors; i++) {
 357                         int startIndex =  i * numSeqsPerProcessor;
 358                         if(i == (processors - 1)){      numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;   }
 359                         lines.push_back(linePair(positions[startIndex], numSeqsPerProcessor));
 360                     }
 361                 }
 362                         #endif
 363                 }
 364
 365         map<string, string> variables;
 366         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 367         string badAccnosFile =  getOutputFileName("accnos",variables);
 368         variables["[extension]"] = m->getExtension(fastafile);
 369                 string goodSeqFile = getOutputFileName("fasta", variables);
 370
 371
 372                 int numFastaSeqs = 0;
 373                 set<string> badSeqNames;
 374                 int start = time(NULL);
 375
 376 #ifdef USE_MPI
 377                         int pid, numSeqsPerProcessor;
 378                         int tag = 2001;
 379                         vector<unsigned long long> MPIPos;
 380
 381                         MPI_Status status;
 382                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 383                         MPI_Comm_size(MPI_COMM_WORLD, &processors);
 384
 385                         MPI_File inMPI;
 386                         MPI_File outMPIGood;
 387                         MPI_File outMPIBadAccnos;
 388
 389                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 390                         int inMode=MPI_MODE_RDONLY;
 391
 392                         char outGoodFilename[1024];
 393                         strcpy(outGoodFilename, goodSeqFile.c_str());
 394
 395                         char outBadAccnosFilename[1024];
 396                         strcpy(outBadAccnosFilename, badAccnosFile.c_str());
 397
 398                         char inFileName[1024];
 399                         strcpy(inFileName, fastafile.c_str());
 400
 401                         MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 402                         MPI_File_open(MPI_COMM_WORLD, outGoodFilename, outMode, MPI_INFO_NULL, &outMPIGood);
 403                         MPI_File_open(MPI_COMM_WORLD, outBadAccnosFilename, outMode, MPI_INFO_NULL, &outMPIBadAccnos);
 404
 405                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood); MPI_File_close(&outMPIBadAccnos); return 0; }
 406
 407                         if (pid == 0) { //you are the root process
 408
 409                                 MPIPos = m->setFilePosFasta(fastafile, numFastaSeqs); //fills MPIPos, returns numSeqs
 410
 411                                 //send file positions to all processes
 412                                 for(int i = 1; i < processors; i++) {
 413                                         MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 414                                         MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 415                                 }
 416
 417                                 //figure out how many sequences you have to align
 418                                 numSeqsPerProcessor = numFastaSeqs / processors;
 419                                 int startIndex =  pid * numSeqsPerProcessor;
 420                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 421
 422                                 //align your part
 423                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 424
 425                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos);  return 0; }
 426
 427                                 for (int i = 1; i < processors; i++) {
 428                                         //get bad lists
 429                                         int badSize;
 430                                         MPI_Recv(&badSize, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 431                                 }
 432                         }else{ //you are a child process
 433                                 MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 434                                 MPIPos.resize(numFastaSeqs+1);
 435                                 MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 436
 437                                 //figure out how many sequences you have to align
 438                                 numSeqsPerProcessor = numFastaSeqs / processors;
 439                                 int startIndex =  pid * numSeqsPerProcessor;
 440                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 441
 442                                 //align your part
 443                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 444
 445                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos); return 0; }
 446
 447                                 //send bad list
 448                                 int badSize = badSeqNames.size();
 449                                 MPI_Send(&badSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 450                         }
 451
 452                         //close files
 453                         MPI_File_close(&inMPI);
 454                         MPI_File_close(&outMPIGood);
 455                         MPI_File_close(&outMPIBadAccnos);
 456                         MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 457
 458 #else
 459         if(processors == 1){ numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);       }
 460         else{ numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames); }
 461
 462         if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 463 #endif
 464
 465                 #ifdef USE_MPI
 466                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 467
 468                         if (pid == 0) { //only one process should fix files
 469
 470                                 //read accnos file with all names in it, process 0 just has its names
 471                                 MPI_File inMPIAccnos;
 472                                 MPI_Offset size;
 473
 474                                 char inFileName[1024];
 475                                 strcpy(inFileName, badAccnosFile.c_str());
 476
 477                                 MPI_File_open(MPI_COMM_SELF, inFileName, inMode, MPI_INFO_NULL, &inMPIAccnos);  //comm, filename, mode, info, filepointer
 478                                 MPI_File_get_size(inMPIAccnos, &size);
 479
 480                                 char* buffer = new char[size];
 481                                 MPI_File_read(inMPIAccnos, buffer, size, MPI_CHAR, &status);
 482
 483                                 string tempBuf = buffer;
 484                                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
 485                                 istringstream iss (tempBuf,istringstream::in);
 486
 487                                 delete buffer;
 488                                 MPI_File_close(&inMPIAccnos);
 489
 490                                 badSeqNames.clear();
 491                                 string tempName;
 492                                 while (!iss.eof()) {
 493                                         iss >> tempName; m->gobble(iss);
 494                                         badSeqNames.insert(tempName);
 495                                 }
 496                 #endif
 497
 498                 if(namefile != "" && groupfile != "")   {
 499                         screenNameGroupFile(badSeqNames);
 500                         if (m->control_pressed) {  m->mothurRemove(goodSeqFile); return 0; }
 501                 }else if(namefile != "")        {
 502                         screenNameGroupFile(badSeqNames);
 503                         if (m->control_pressed) {  m->mothurRemove(goodSeqFile);  return 0; }
 504                 }else if(groupfile != "")                               {       screenGroupFile(badSeqNames);           }       // this screens just the group
 505                 else if (countfile != "") {     screenCountFile(badSeqNames);           }
 506
 507
 508                 if (m->control_pressed) { m->mothurRemove(goodSeqFile);  return 0; }
 509
 510                 if(alignreport != "")                                   {       screenAlignReport(badSeqNames);         }
 511                 if(qualfile != "")                                              {       screenQual(badSeqNames);                        }
 512                 if(taxonomy != "")                                              {       screenTaxonomy(badSeqNames);            }
 513
 514                 if (m->control_pressed) { m->mothurRemove(goodSeqFile);  return 0; }
 515
 516                 #ifdef USE_MPI
 517                         }
 518                 #endif
 519
 520                 m->mothurOutEndLine();
 521                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 522                 m->mothurOut(goodSeqFile); m->mothurOutEndLine();       outputTypes["fasta"].push_back(goodSeqFile);
 523                 m->mothurOut(badAccnosFile); m->mothurOutEndLine();      outputTypes["accnos"].push_back(badAccnosFile);
 524                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 525                 m->mothurOutEndLine();
 526                 m->mothurOutEndLine();
 527
 528                 //set fasta file as new current fastafile
 529                 string current = "";
 530                 itTypes = outputTypes.find("fasta");
 531                 if (itTypes != outputTypes.end()) {
 532                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 533                 }
 534
 535                 itTypes = outputTypes.find("name");
 536                 if (itTypes != outputTypes.end()) {
 537                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 538                 }
 539
 540                 itTypes = outputTypes.find("group");
 541                 if (itTypes != outputTypes.end()) {
 542                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 543                 }
 544
 545                 itTypes = outputTypes.find("qfile");
 546                 if (itTypes != outputTypes.end()) {
 547                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 548                 }
 549
 550                 itTypes = outputTypes.find("taxonomy");
 551                 if (itTypes != outputTypes.end()) {
 552                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 553                 }
 554
 555         itTypes = outputTypes.find("count");
 556                 if (itTypes != outputTypes.end()) {
 557                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 558                 }
 559
 560                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to screen " + toString(numFastaSeqs) + " sequences.");
 561                 m->mothurOutEndLine();
 562
 563                 return 0;
 564         }
 565         catch(exception& e) {
 566                 m->errorOut(e, "ScreenSeqsCommand", "execute");
 567                 exit(1);
 568         }
 569 }
 570
 571 //***************************************************************************************************************
 572
 573 int ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
 574         try {
 575                 ifstream inputNames;
 576                 m->openInputFile(namefile, inputNames);
 577                 set<string> badSeqGroups;
 578                 string seqName, seqList, group;
 579                 set<string>::iterator it;
 580         map<string, string> variables;
 581                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(namefile));
 582         variables["[extension]"] = m->getExtension(namefile);
 583                 string goodNameFile = getOutputFileName("name", variables);
 584                 outputNames.push_back(goodNameFile);  outputTypes["name"].push_back(goodNameFile);
 585
 586                 ofstream goodNameOut;   m->openOutputFile(goodNameFile, goodNameOut);
 587
 588                 while(!inputNames.eof()){
 589                         if (m->control_pressed) { goodNameOut.close();  inputNames.close(); m->mothurRemove(goodNameFile);  return 0; }
 590
 591                         inputNames >> seqName; m->gobble(inputNames); inputNames >> seqList;
 592                         it = badSeqNames.find(seqName);
 593
 594                         if(it != badSeqNames.end()){
 595                                 badSeqNames.erase(it);
 596
 597                                 if(namefile != ""){
 598                                         int start = 0;
 599                                         for(int i=0;i<seqList.length();i++){
 600                                                 if(seqList[i] == ','){
 601                                                         badSeqGroups.insert(seqList.substr(start,i-start));
 602                                                         start = i+1;
 603                                                 }
 604                                         }
 605                                         badSeqGroups.insert(seqList.substr(start,seqList.length()-start));
 606                                 }
 607                         }
 608                         else{
 609                                 goodNameOut << seqName << '\t' << seqList << endl;
 610                         }
 611                         m->gobble(inputNames);
 612                 }
 613                 inputNames.close();
 614                 goodNameOut.close();
 615
 616                 //we were unable to remove some of the bad sequences
 617                 if (badSeqNames.size() != 0) {
 618                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 619                                 m->mothurOut("Your namefile does not include the sequence " + *it + " please correct.");
 620                                 m->mothurOutEndLine();
 621                         }
 622                 }
 623
 624                 if(groupfile != ""){
 625
 626                         ifstream inputGroups;
 627                         m->openInputFile(groupfile, inputGroups);
 628             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(groupfile));
 629             variables["[extension]"] = m->getExtension(groupfile);
 630             string goodGroupFile = getOutputFileName("group", variables);
 631
 632                         outputNames.push_back(goodGroupFile);   outputTypes["group"].push_back(goodGroupFile);
 633
 634                         ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 635
 636                         while(!inputGroups.eof()){
 637                                 if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); m->mothurRemove(goodNameFile);  m->mothurRemove(goodGroupFile); return 0; }
 638
 639                                 inputGroups >> seqName; m->gobble(inputGroups); inputGroups >> group;
 640
 641                                 it = badSeqGroups.find(seqName);
 642
 643                                 if(it != badSeqGroups.end()){
 644                                         badSeqGroups.erase(it);
 645                                 }
 646                                 else{
 647                                         goodGroupOut << seqName << '\t' << group << endl;
 648                                 }
 649                                 m->gobble(inputGroups);
 650                         }
 651                         inputGroups.close();
 652                         goodGroupOut.close();
 653
 654                         //we were unable to remove some of the bad sequences
 655                         if (badSeqGroups.size() != 0) {
 656                                 for (it = badSeqGroups.begin(); it != badSeqGroups.end(); it++) {
 657                                         m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 658                                         m->mothurOutEndLine();
 659                                 }
 660                         }
 661                 }
 662
 663
 664                 return 0;
 665
 666         }
 667         catch(exception& e) {
 668                 m->errorOut(e, "ScreenSeqsCommand", "screenNameGroupFile");
 669                 exit(1);
 670         }
 671 }
 672 //***************************************************************************************************************
 673 int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
 674         try {
 675
 676                 vector<int> startPosition;
 677                 vector<int> endPosition;
 678                 vector<int> seqLength;
 679                 vector<int> ambigBases;
 680                 vector<int> longHomoPolymer;
 681
 682         vector<unsigned long long> positions;
 683 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 684                 positions = m->divideFile(fastafile, processors);
 685                 for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(linePair(positions[i], positions[(i+1)])); }
 686 #else
 687                 if(processors == 1){ lines.push_back(linePair(0, 1000));  }
 688         else {
 689             int numFastaSeqs = 0;
 690             positions = m->setFilePosFasta(fastafile, numFastaSeqs);
 691             if (positions.size() < processors) { processors = positions.size(); }
 692
 693             //figure out how many sequences you have to process
 694             int numSeqsPerProcessor = numFastaSeqs / processors;
 695             for (int i = 0; i < processors; i++) {
 696                 int startIndex =  i * numSeqsPerProcessor;
 697                 if(i == (processors - 1)){      numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;   }
 698                 lines.push_back(linePair(positions[startIndex], numSeqsPerProcessor));
 699             }
 700         }
 701 #endif
 702
 703 #ifdef USE_MPI
 704                 int pid;
 705                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 706
 707                 if (pid == 0) {
 708                         driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 709 #else
 710                 int numSeqs = 0;
 711                 //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 712                         if(processors == 1){
 713                                 numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 714                         }else{
 715                                 numSeqs = createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile);
 716                         }
 717
 718                         if (m->control_pressed) {  return 0; }
 719                 //#else
 720                 //      numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 721                 //      if (m->control_pressed) {  return 0; }
 722                 //#endif
 723 #endif
 724                 sort(startPosition.begin(), startPosition.end());
 725                 sort(endPosition.begin(), endPosition.end());
 726                 sort(seqLength.begin(), seqLength.end());
 727                 sort(ambigBases.begin(), ambigBases.end());
 728                 sort(longHomoPolymer.begin(), longHomoPolymer.end());
 729
 730                 //numSeqs is the number of unique seqs, startPosition.size() is the total number of seqs, we want to optimize using all seqs
 731                 int criteriaPercentile  = int(startPosition.size() * (criteria / (float) 100));
 732
 733                 for (int i = 0; i < optimize.size(); i++) {
 734                         if (optimize[i] == "start") { startPos = startPosition[criteriaPercentile]; m->mothurOut("Optimizing start to " + toString(startPos) + "."); m->mothurOutEndLine(); }
 735                         else if (optimize[i] == "end") { int endcriteriaPercentile = int(endPosition.size() * ((100 - criteria) / (float) 100));  endPos = endPosition[endcriteriaPercentile]; m->mothurOut("Optimizing end to " + toString(endPos) + "."); m->mothurOutEndLine();}
 736                         else if (optimize[i] == "maxambig") { maxAmbig = ambigBases[criteriaPercentile]; m->mothurOut("Optimizing maxambig to " + toString(maxAmbig) + "."); m->mothurOutEndLine(); }
 737                         else if (optimize[i] == "maxhomop") { maxHomoP = longHomoPolymer[criteriaPercentile]; m->mothurOut("Optimizing maxhomop to " + toString(maxHomoP) + "."); m->mothurOutEndLine(); }
 738                         else if (optimize[i] == "minlength") { int mincriteriaPercentile = int(seqLength.size() * ((100 - criteria) / (float) 100)); minLength = seqLength[mincriteriaPercentile]; m->mothurOut("Optimizing minlength to " + toString(minLength) + "."); m->mothurOutEndLine(); }
 739                         else if (optimize[i] == "maxlength") { maxLength = seqLength[criteriaPercentile]; m->mothurOut("Optimizing maxlength to " + toString(maxLength) + "."); m->mothurOutEndLine(); }
 740                 }
 741
 742 #ifdef USE_MPI
 743                 }
 744
 745                 MPI_Status status;
 746                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 747                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 748
 749                 if (pid == 0) {
 750                         //send file positions to all processes
 751                         for(int i = 1; i < processors; i++) {
 752                                 MPI_Send(&startPos, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 753                                 MPI_Send(&endPos, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 754                                 MPI_Send(&maxAmbig, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 755                                 MPI_Send(&maxHomoP, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 756                                 MPI_Send(&minLength, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 757                                 MPI_Send(&maxLength, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 758                         }
 759                 }else {
 760                         MPI_Recv(&startPos, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 761                         MPI_Recv(&endPos, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 762                         MPI_Recv(&maxAmbig, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 763                         MPI_Recv(&maxHomoP, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 764                         MPI_Recv(&minLength, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 765                         MPI_Recv(&maxLength, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 766                 }
 767                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 768 #endif
 769                 return 0;
 770         }
 771         catch(exception& e) {
 772                 m->errorOut(e, "ScreenSeqsCommand", "getSummary");
 773                 exit(1);
 774         }
 775 }
 776 /**************************************************************************************/
 777 int ScreenSeqsCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, linePair filePos) {
 778         try {
 779
 780                 ifstream in;
 781                 m->openInputFile(filename, in);
 782
 783                 in.seekg(filePos.start);
 784
 785                 bool done = false;
 786                 int count = 0;
 787
 788                 while (!done) {
 789
 790                         if (m->control_pressed) { in.close(); return 1; }
 791
 792                         Sequence current(in); m->gobble(in);
 793
 794                         if (current.getName() != "") {
 795                                 int num = 1;
 796                                 if (namefile != "") {
 797                                         //make sure this sequence is in the namefile, else error
 798                                         map<string, int>::iterator it = nameMap.find(current.getName());
 799
 800                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 801                                         else { num = it->second; }
 802                                 }
 803
 804                                 //for each sequence this sequence represents
 805                                 for (int i = 0; i < num; i++) {
 806                                         startPosition.push_back(current.getStartPos());
 807                                         endPosition.push_back(current.getEndPos());
 808                                         seqLength.push_back(current.getNumBases());
 809                                         ambigBases.push_back(current.getAmbigBases());
 810                                         longHomoPolymer.push_back(current.getLongHomoPolymer());
 811                                 }
 812
 813                                 count++;
 814                         }
 815                         //if((count) % 100 == 0){       m->mothurOut("Optimizing sequence: " + toString(count)); m->mothurOutEndLine();         }
 816                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 817                                 unsigned long long pos = in.tellg();
 818                                 if ((pos == -1) || (pos >= filePos.end)) { break; }
 819                         #else
 820                                 if (in.eof()) { break; }
 821                         #endif
 822
 823                 }
 824
 825                 in.close();
 826
 827                 return count;
 828         }
 829         catch(exception& e) {
 830                 m->errorOut(e, "ScreenSeqsCommand", "driverCreateSummary");
 831                 exit(1);
 832         }
 833 }
 834 /**************************************************************************************************/
 835 int ScreenSeqsCommand::createProcessesCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename) {
 836         try {
 837
 838         int process = 1;
 839                 int num = 0;
 840                 vector<int> processIDS;
 841
 842 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 843
 844                 //loop through and create all the processes you want
 845                 while (process != processors) {
 846                         int pid = fork();
 847
 848                         if (pid > 0) {
 849                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 850                                 process++;
 851                         }else if (pid == 0){
 852                                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[process]);
 853
 854                                 //pass numSeqs to parent
 855                                 ofstream out;
 856                                 string tempFile = fastafile + toString(getpid()) + ".num.temp";
 857                                 m->openOutputFile(tempFile, out);
 858
 859                                 out << num << endl;
 860                                 out << startPosition.size() << endl;
 861                                 for (int k = 0; k < startPosition.size(); k++)          {               out << startPosition[k] << '\t'; }  out << endl;
 862                                 for (int k = 0; k < endPosition.size(); k++)            {               out << endPosition[k] << '\t'; }  out << endl;
 863                                 for (int k = 0; k < seqLength.size(); k++)                      {               out << seqLength[k] << '\t'; }  out << endl;
 864                                 for (int k = 0; k < ambigBases.size(); k++)                     {               out << ambigBases[k] << '\t'; }  out << endl;
 865                                 for (int k = 0; k < longHomoPolymer.size(); k++)        {               out << longHomoPolymer[k] << '\t'; }  out << endl;
 866
 867                                 out.close();
 868
 869                                 exit(0);
 870                         }else {
 871                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 872                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 873                                 exit(0);
 874                         }
 875                 }
 876
 877                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 878
 879                 //force parent to wait until all the processes are done
 880                 for (int i=0;i<processIDS.size();i++) {
 881                         int temp = processIDS[i];
 882                         wait(&temp);
 883                 }
 884
 885                 //parent reads in and combine Filter info
 886                 for (int i = 0; i < processIDS.size(); i++) {
 887                         string tempFilename = fastafile + toString(processIDS[i]) + ".num.temp";
 888                         ifstream in;
 889                         m->openInputFile(tempFilename, in);
 890
 891                         int temp, tempNum;
 892                         in >> tempNum; m->gobble(in); num += tempNum;
 893                         in >> tempNum; m->gobble(in);
 894                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; startPosition.push_back(temp);              }               m->gobble(in);
 895                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; endPosition.push_back(temp);                }               m->gobble(in);
 896                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; seqLength.push_back(temp);                  }               m->gobble(in);
 897                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; ambigBases.push_back(temp);                 }               m->gobble(in);
 898                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; longHomoPolymer.push_back(temp);    }               m->gobble(in);
 899
 900                         in.close();
 901                         m->mothurRemove(tempFilename);
 902                 }
 903
 904
 905 #else
 906         //////////////////////////////////////////////////////////////////////////////////////////////////////
 907                 //Windows version shared memory, so be careful when passing variables through the seqSumData struct.
 908                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 909                 //Taking advantage of shared memory to allow both threads to add info to vectors.
 910                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 911
 912                 vector<sumData*> pDataArray;
 913                 DWORD   dwThreadIdArray[processors-1];
 914                 HANDLE  hThreadArray[processors-1];
 915
 916                 //Create processor worker threads.
 917                 for( int i=0; i<processors-1; i++ ){
 918
 919                         // Allocate memory for thread data.
 920                         sumData* tempSum = new sumData(filename, m, lines[i].start, lines[i].end, namefile, nameMap);
 921                         pDataArray.push_back(tempSum);
 922
 923                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
 924                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 925                         hThreadArray[i] = CreateThread(NULL, 0, MySumThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 926                 }
 927
 928         //do your part
 929                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[processors-1]);
 930
 931                 //Wait until all threads have terminated.
 932                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 933
 934                 //Close all thread handles and free memory allocations.
 935                 for(int i=0; i < pDataArray.size(); i++){
 936                         num += pDataArray[i]->count;
 937             if (pDataArray[i]->count != pDataArray[i]->end) {
 938                 m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
 939             }
 940             for (int k = 0; k < pDataArray[i]->startPosition.size(); k++) {     startPosition.push_back(pDataArray[i]->startPosition[k]);       }
 941                         for (int k = 0; k < pDataArray[i]->endPosition.size(); k++) {   endPosition.push_back(pDataArray[i]->endPosition[k]);       }
 942             for (int k = 0; k < pDataArray[i]->seqLength.size(); k++) { seqLength.push_back(pDataArray[i]->seqLength[k]);       }
 943             for (int k = 0; k < pDataArray[i]->ambigBases.size(); k++) {        ambigBases.push_back(pDataArray[i]->ambigBases[k]);       }
 944             for (int k = 0; k < pDataArray[i]->longHomoPolymer.size(); k++) {   longHomoPolymer.push_back(pDataArray[i]->longHomoPolymer[k]);       }
 945                         CloseHandle(hThreadArray[i]);
 946                         delete pDataArray[i];
 947                 }
 948
 949 #endif
 950         return num;
 951         }
 952         catch(exception& e) {
 953                 m->errorOut(e, "ScreenSeqsCommand", "createProcessesCreateSummary");
 954                 exit(1);
 955         }
 956 }
 957
 958 //***************************************************************************************************************
 959
 960 int ScreenSeqsCommand::screenGroupFile(set<string> badSeqNames){
 961         try {
 962                 ifstream inputGroups;
 963                 m->openInputFile(groupfile, inputGroups);
 964                 string seqName, group;
 965                 set<string>::iterator it;
 966                 map<string, string> variables;
 967                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(groupfile));
 968         variables["[extension]"] = m->getExtension(groupfile);
 969         string goodGroupFile = getOutputFileName("group", variables);
 970         outputNames.push_back(goodGroupFile);  outputTypes["group"].push_back(goodGroupFile);
 971                 ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 972
 973                 while(!inputGroups.eof()){
 974                         if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); m->mothurRemove(goodGroupFile); return 0; }
 975
 976                         inputGroups >> seqName; m->gobble(inputGroups); inputGroups >> group;
 977                         it = badSeqNames.find(seqName);
 978
 979                         if(it != badSeqNames.end()){
 980                                 badSeqNames.erase(it);
 981                         }
 982                         else{
 983                                 goodGroupOut << seqName << '\t' << group << endl;
 984                         }
 985                         m->gobble(inputGroups);
 986                 }
 987
 988                 if (m->control_pressed) { goodGroupOut.close();  inputGroups.close(); m->mothurRemove(goodGroupFile);  return 0; }
 989
 990                 //we were unable to remove some of the bad sequences
 991                 if (badSeqNames.size() != 0) {
 992                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 993                                 m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 994                                 m->mothurOutEndLine();
 995                         }
 996                 }
 997
 998                 inputGroups.close();
 999                 goodGroupOut.close();
1000
1001                 if (m->control_pressed) { m->mothurRemove(goodGroupFile);   }
1002
1003                 return 0;
1004
1005         }
1006         catch(exception& e) {
1007                 m->errorOut(e, "ScreenSeqsCommand", "screenGroupFile");
1008                 exit(1);
1009         }
1010 }
1011 //***************************************************************************************************************
1012 int ScreenSeqsCommand::screenCountFile(set<string> badSeqNames){
1013         try {
1014                 ifstream in;
1015                 m->openInputFile(countfile, in);
1016                 set<string>::iterator it;
1017                 map<string, string> variables;
1018                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(countfile));
1019         variables["[extension]"] = m->getExtension(countfile);
1020         string goodCountFile = getOutputFileName("count", variables);
1021
1022         outputNames.push_back(goodCountFile);  outputTypes["count"].push_back(goodCountFile);
1023                 ofstream goodCountOut;  m->openOutputFile(goodCountFile, goodCountOut);
1024
1025         string headers = m->getline(in); m->gobble(in);
1026         goodCountOut << headers << endl;
1027
1028         string name, rest; int thisTotal;
1029         while (!in.eof()) {
1030
1031                         if (m->control_pressed) { goodCountOut.close(); in.close(); m->mothurRemove(goodCountFile); return 0; }
1032
1033                         in >> name; m->gobble(in);
1034             in >> thisTotal; m->gobble(in);
1035             rest = m->getline(in); m->gobble(in);
1036
1037                         it = badSeqNames.find(name);
1038
1039                         if(it != badSeqNames.end()){
1040                                 badSeqNames.erase(it);
1041                         }
1042                         else{
1043                                 goodCountOut << name << '\t' << thisTotal << '\t' << rest << endl;
1044                         }
1045                 }
1046
1047                 if (m->control_pressed) { goodCountOut.close();  in.close(); m->mothurRemove(goodCountFile);  return 0; }
1048
1049                 //we were unable to remove some of the bad sequences
1050                 if (badSeqNames.size() != 0) {
1051                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
1052                                 m->mothurOut("Your count file does not include the sequence " + *it + " please correct.");
1053                                 m->mothurOutEndLine();
1054                         }
1055                 }
1056
1057                 in.close();
1058                 goodCountOut.close();
1059
1060         //check for groups that have been eliminated
1061         CountTable ct;
1062         if (ct.testGroups(goodCountFile)) {
1063             ct.readTable(goodCountFile);
1064             ct.printTable(goodCountFile);
1065         }
1066
1067                 if (m->control_pressed) { m->mothurRemove(goodCountFile);   }
1068
1069                 return 0;
1070
1071         }
1072         catch(exception& e) {
1073                 m->errorOut(e, "ScreenSeqsCommand", "screenCountFile");
1074                 exit(1);
1075         }
1076 }
1077 //***************************************************************************************************************
1078
1079 int ScreenSeqsCommand::screenAlignReport(set<string> badSeqNames){
1080         try {
1081                 ifstream inputAlignReport;
1082                 m->openInputFile(alignreport, inputAlignReport);
1083                 string seqName, group;
1084                 set<string>::iterator it;
1085
1086         map<string, string> variables;
1087                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(alignreport));
1088         string goodAlignReportFile = getOutputFileName("alignreport", variables);
1089
1090                 outputNames.push_back(goodAlignReportFile);  outputTypes["alignreport"].push_back(goodAlignReportFile);
1091                 ofstream goodAlignReportOut;    m->openOutputFile(goodAlignReportFile, goodAlignReportOut);
1092
1093                 while (!inputAlignReport.eof()) {               //      need to copy header
1094                         char c = inputAlignReport.get();
1095                         goodAlignReportOut << c;
1096                         if (c == 10 || c == 13){        break;  }
1097                 }
1098
1099                 while(!inputAlignReport.eof()){
1100                         if (m->control_pressed) { goodAlignReportOut.close(); inputAlignReport.close(); m->mothurRemove(goodAlignReportFile); return 0; }
1101
1102                         inputAlignReport >> seqName;
1103                         it = badSeqNames.find(seqName);
1104                         string line;
1105                         while (!inputAlignReport.eof()) {               //      need to copy header
1106                                 char c = inputAlignReport.get();
1107                                 line += c;
1108                                 if (c == 10 || c == 13){        break;  }
1109                         }
1110
1111                         if(it != badSeqNames.end()){
1112                                 badSeqNames.erase(it);
1113                         }
1114                         else{
1115                                 goodAlignReportOut << seqName << '\t' << line;
1116                         }
1117                         m->gobble(inputAlignReport);
1118                 }
1119
1120                 if (m->control_pressed) { goodAlignReportOut.close();  inputAlignReport.close(); m->mothurRemove(goodAlignReportFile);  return 0; }
1121
1122                 //we were unable to remove some of the bad sequences
1123                 if (badSeqNames.size() != 0) {
1124                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
1125                                 m->mothurOut("Your alignreport file does not include the sequence " + *it + " please correct.");
1126                                 m->mothurOutEndLine();
1127                         }
1128                 }
1129
1130                 inputAlignReport.close();
1131                 goodAlignReportOut.close();
1132
1133                 if (m->control_pressed) {  m->mothurRemove(goodAlignReportFile);  return 0; }
1134
1135                 return 0;
1136
1137         }
1138         catch(exception& e) {
1139                 m->errorOut(e, "ScreenSeqsCommand", "screenAlignReport");
1140                 exit(1);
1141         }
1142
1143 }
1144 //***************************************************************************************************************
1145
1146 int ScreenSeqsCommand::screenTaxonomy(set<string> badSeqNames){
1147         try {
1148                 ifstream input;
1149                 m->openInputFile(taxonomy, input);
1150                 string seqName, tax;
1151                 set<string>::iterator it;
1152         map<string, string> variables;
1153                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(taxonomy));
1154         variables["[extension]"] = m->getExtension(taxonomy);
1155         string goodTaxFile = getOutputFileName("taxonomy", variables);
1156
1157                 outputNames.push_back(goodTaxFile);  outputTypes["taxonomy"].push_back(goodTaxFile);
1158                 ofstream goodTaxOut;    m->openOutputFile(goodTaxFile, goodTaxOut);
1159
1160                 while(!input.eof()){
1161                         if (m->control_pressed) { goodTaxOut.close(); input.close(); m->mothurRemove(goodTaxFile); return 0; }
1162
1163                         input >> seqName; m->gobble(input); input >> tax;
1164                         it = badSeqNames.find(seqName);
1165
1166                         if(it != badSeqNames.end()){ badSeqNames.erase(it); }
1167                         else{
1168                                 goodTaxOut << seqName << '\t' << tax << endl;
1169                         }
1170                         m->gobble(input);
1171                 }
1172
1173                 if (m->control_pressed) { goodTaxOut.close(); input.close(); m->mothurRemove(goodTaxFile); return 0; }
1174
1175                 //we were unable to remove some of the bad sequences
1176                 if (badSeqNames.size() != 0) {
1177                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
1178                                 m->mothurOut("Your taxonomy file does not include the sequence " + *it + " please correct.");
1179                                 m->mothurOutEndLine();
1180                         }
1181                 }
1182
1183                 input.close();
1184                 goodTaxOut.close();
1185
1186                 if (m->control_pressed) {  m->mothurRemove(goodTaxFile);  return 0; }
1187
1188                 return 0;
1189
1190         }
1191         catch(exception& e) {
1192                 m->errorOut(e, "ScreenSeqsCommand", "screenTaxonomy");
1193                 exit(1);
1194         }
1195
1196 }
1197 //***************************************************************************************************************
1198
1199 int ScreenSeqsCommand::screenQual(set<string> badSeqNames){
1200         try {
1201                 ifstream in;
1202                 m->openInputFile(qualfile, in);
1203                 set<string>::iterator it;
1204                 map<string, string> variables;
1205                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(qualfile));
1206         variables["[extension]"] = m->getExtension(qualfile);
1207         string goodQualFile = getOutputFileName("qfile", variables);
1208
1209                 outputNames.push_back(goodQualFile);  outputTypes["qfile"].push_back(goodQualFile);
1210                 ofstream goodQual;      m->openOutputFile(goodQualFile, goodQual);
1211
1212                 while(!in.eof()){
1213
1214                         if (m->control_pressed) { goodQual.close(); in.close(); m->mothurRemove(goodQualFile); return 0; }
1215
1216                         string saveName = "";
1217                         string name = "";
1218                         string scores = "";
1219
1220                         in >> name;
1221
1222                         if (name.length() != 0) {
1223                                 saveName = name.substr(1);
1224                                 while (!in.eof())       {
1225                                         char c = in.get();
1226                                         if (c == 10 || c == 13){        break;  }
1227                                         else { name += c; }
1228                                 }
1229                                 m->gobble(in);
1230                         }
1231
1232                         while(in){
1233                                 char letter= in.get();
1234                                 if(letter == '>'){      in.putback(letter);     break;  }
1235                                 else{ scores += letter; }
1236                         }
1237
1238                         m->gobble(in);
1239
1240                         it = badSeqNames.find(saveName);
1241
1242                         if(it != badSeqNames.end()){
1243                                 badSeqNames.erase(it);
1244                         }else{
1245                                 goodQual << name << endl << scores;
1246                         }
1247
1248                         m->gobble(in);
1249                 }
1250
1251                 in.close();
1252                 goodQual.close();
1253
1254                 //we were unable to remove some of the bad sequences
1255                 if (badSeqNames.size() != 0) {
1256                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
1257                                 m->mothurOut("Your qual file does not include the sequence " + *it + " please correct.");
1258                                 m->mothurOutEndLine();
1259                         }
1260                 }
1261
1262                 if (m->control_pressed) {  m->mothurRemove(goodQualFile);  return 0; }
1263
1264                 return 0;
1265
1266         }
1267         catch(exception& e) {
1268                 m->errorOut(e, "ScreenSeqsCommand", "screenQual");
1269                 exit(1);
1270         }
1271
1272 }
1273 //**********************************************************************************************************************
1274
1275 int ScreenSeqsCommand::driver(linePair filePos, string goodFName, string badAccnosFName, string filename, set<string>& badSeqNames){
1276         try {
1277                 ofstream goodFile;
1278                 m->openOutputFile(goodFName, goodFile);
1279
1280                 ofstream badAccnosFile;
1281                 m->openOutputFile(badAccnosFName, badAccnosFile);
1282
1283                 ifstream inFASTA;
1284                 m->openInputFile(filename, inFASTA);
1285
1286                 inFASTA.seekg(filePos.start);
1287
1288                 bool done = false;
1289                 int count = 0;
1290
1291                 while (!done) {
1292
1293                         if (m->control_pressed) {  return 0; }
1294
1295                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
1296                         if (currSeq.getName() != "") {
1297                                 bool goodSeq = 1;               //      innocent until proven guilty
1298                                 if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
1299                                 if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
1300                                 if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
1301                                 if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
1302                                 if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
1303                                 if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
1304
1305                                 if(goodSeq == 1){
1306                                         currSeq.printSequence(goodFile);
1307                                 }
1308                                 else{
1309                                         badAccnosFile << currSeq.getName() << endl;
1310                                         badSeqNames.insert(currSeq.getName());
1311                                 }
1312                         count++;
1313                         }
1314
1315                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1316                                 unsigned long long pos = inFASTA.tellg();
1317                                 if ((pos == -1) || (pos >= filePos.end)) { break; }
1318                         #else
1319                                 if (inFASTA.eof()) { break; }
1320                         #endif
1321
1322                         //report progress
1323                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1324                 }
1325                 //report progress
1326                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1327
1328
1329                 goodFile.close();
1330                 inFASTA.close();
1331                 badAccnosFile.close();
1332
1333                 return count;
1334         }
1335         catch(exception& e) {
1336                 m->errorOut(e, "ScreenSeqsCommand", "driver");
1337                 exit(1);
1338         }
1339 }
1340 //**********************************************************************************************************************
1341 #ifdef USE_MPI
1342 int ScreenSeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& goodFile, MPI_File& badAccnosFile, vector<unsigned long long>& MPIPos, set<string>& badSeqNames){
1343         try {
1344                 string outputString = "";
1345                 MPI_Status statusGood;
1346                 MPI_Status statusBadAccnos;
1347                 MPI_Status status;
1348                 int pid;
1349                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
1350
1351                 for(int i=0;i<num;i++){
1352
1353                         if (m->control_pressed) {  return 0; }
1354
1355                         //read next sequence
1356                         int length = MPIPos[start+i+1] - MPIPos[start+i];
1357
1358                         char* buf4 = new char[length];
1359
1360                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
1361
1362                         string tempBuf = buf4;  delete buf4;
1363                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
1364                         istringstream iss (tempBuf,istringstream::in);
1365
1366                         Sequence currSeq(iss);
1367
1368                         //process seq
1369                         if (currSeq.getName() != "") {
1370                                 bool goodSeq = 1;               //      innocent until proven guilty
1371                                 if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
1372                                 if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
1373                                 if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
1374                                 if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
1375                                 if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
1376                                 if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
1377
1378                                 if(goodSeq == 1){
1379                                         outputString =  ">" + currSeq.getName() + "\n" + currSeq.getAligned() + "\n";
1380
1381                                         //print good seq
1382                                         length = outputString.length();
1383                                         char* buf2 = new char[length];
1384                                         memcpy(buf2, outputString.c_str(), length);
1385
1386                                         MPI_File_write_shared(goodFile, buf2, length, MPI_CHAR, &statusGood);
1387                                         delete buf2;
1388                                 }
1389                                 else{
1390
1391                                         badSeqNames.insert(currSeq.getName());
1392
1393                                         //write to bad accnos file
1394                                         outputString = currSeq.getName() + "\n";
1395
1396                                         length = outputString.length();
1397                                         char* buf3 = new char[length];
1398                                         memcpy(buf3, outputString.c_str(), length);
1399
1400                                         MPI_File_write_shared(badAccnosFile, buf3, length, MPI_CHAR, &statusBadAccnos);
1401                                         delete buf3;
1402                                 }
1403                         }
1404
1405                         //report progress
1406                         if((i) % 100 == 0){     m->mothurOut("Processing sequence: " + toString(i)); m->mothurOutEndLine();             }
1407                 }
1408
1409                 return 1;
1410         }
1411         catch(exception& e) {
1412                 m->errorOut(e, "ScreenSeqsCommand", "driverMPI");
1413                 exit(1);
1414         }
1415 }
1416 #endif
1417 /**************************************************************************************************/
1418
1419 int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, string filename, set<string>& badSeqNames) {
1420         try {
1421
1422         vector<int> processIDS;
1423         int process = 1;
1424                 int num = 0;
1425
1426 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1427
1428                 //loop through and create all the processes you want
1429                 while (process != processors) {
1430                         int pid = fork();
1431
1432                         if (pid > 0) {
1433                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1434                                 process++;
1435                         }else if (pid == 0){
1436                                 num = driver(lines[process], goodFileName + toString(getpid()) + ".temp", badAccnos + toString(getpid()) + ".temp", filename, badSeqNames);
1437
1438                                 //pass numSeqs to parent
1439                                 ofstream out;
1440                                 string tempFile = filename + toString(getpid()) + ".num.temp";
1441                                 m->openOutputFile(tempFile, out);
1442                                 out << num << endl;
1443                                 out.close();
1444
1445                                 exit(0);
1446                         }else {
1447                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1448                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1449                                 exit(0);
1450                         }
1451                 }
1452
1453         num = driver(lines[0], goodFileName, badAccnos, filename, badSeqNames);
1454
1455                 //force parent to wait until all the processes are done
1456                 for (int i=0;i<processIDS.size();i++) {
1457                         int temp = processIDS[i];
1458                         wait(&temp);
1459                 }
1460
1461                 for (int i = 0; i < processIDS.size(); i++) {
1462                         ifstream in;
1463                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
1464                         m->openInputFile(tempFile, in);
1465                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1466                         in.close(); m->mothurRemove(tempFile);
1467
1468             m->appendFiles((goodFileName + toString(processIDS[i]) + ".temp"), goodFileName);
1469             m->mothurRemove((goodFileName + toString(processIDS[i]) + ".temp"));
1470
1471             m->appendFiles((badAccnos + toString(processIDS[i]) + ".temp"), badAccnos);
1472             m->mothurRemove((badAccnos + toString(processIDS[i]) + ".temp"));
1473                 }
1474
1475         //read badSeqs in because root process doesnt know what other "bad" seqs the children found
1476         ifstream inBad;
1477         int ableToOpen = m->openInputFile(badAccnos, inBad, "no error");
1478
1479         if (ableToOpen == 0) {
1480             badSeqNames.clear();
1481             string tempName;
1482             while (!inBad.eof()) {
1483                 inBad >> tempName; m->gobble(inBad);
1484                 badSeqNames.insert(tempName);
1485             }
1486             inBad.close();
1487         }
1488 #else
1489
1490         //////////////////////////////////////////////////////////////////////////////////////////////////////
1491                 //Windows version shared memory, so be careful when passing variables through the sumScreenData struct.
1492                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1493                 //Taking advantage of shared memory to allow both threads to add info to badSeqNames.
1494                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1495
1496                 vector<sumScreenData*> pDataArray;
1497                 DWORD   dwThreadIdArray[processors-1];
1498                 HANDLE  hThreadArray[processors-1];
1499
1500                 //Create processor worker threads.
1501                 for( int i=0; i<processors-1; i++ ){
1502
1503             string extension = "";
1504             if (i!=0) {extension += toString(i) + ".temp"; processIDS.push_back(i); }
1505
1506                         // Allocate memory for thread data.
1507                         sumScreenData* tempSum = new sumScreenData(startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, filename, m, lines[i].start, lines[i].end,goodFileName+extension, badAccnos+extension);
1508                         pDataArray.push_back(tempSum);
1509
1510                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1511                         hThreadArray[i] = CreateThread(NULL, 0, MySumScreenThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
1512                 }
1513
1514         //do your part
1515         num = driver(lines[processors-1], (goodFileName+toString(processors-1)+".temp"), (badAccnos+toString(processors-1)+".temp"), filename, badSeqNames);
1516         processIDS.push_back(processors-1);
1517
1518                 //Wait until all threads have terminated.
1519                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
1520
1521                 //Close all thread handles and free memory allocations.
1522                 for(int i=0; i < pDataArray.size(); i++){
1523                         num += pDataArray[i]->count;
1524             if (pDataArray[i]->count != pDataArray[i]->end) {
1525                 m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
1526             }
1527             for (set<string>::iterator it = pDataArray[i]->badSeqNames.begin(); it != pDataArray[i]->badSeqNames.end(); it++) { badSeqNames.insert(*it);       }
1528                         CloseHandle(hThreadArray[i]);
1529                         delete pDataArray[i];
1530                 }
1531
1532         for (int i = 0; i < processIDS.size(); i++) {
1533             m->appendFiles((goodFileName + toString(processIDS[i]) + ".temp"), goodFileName);
1534             m->mothurRemove((goodFileName + toString(processIDS[i]) + ".temp"));
1535
1536             m->appendFiles((badAccnos + toString(processIDS[i]) + ".temp"), badAccnos);
1537             m->mothurRemove((badAccnos + toString(processIDS[i]) + ".temp"));
1538                 }
1539
1540 #endif
1541
1542         return num;
1543
1544         }
1545         catch(exception& e) {
1546                 m->errorOut(e, "ScreenSeqsCommand", "createProcesses");
1547                 exit(1);
1548         }
1549 }
1550
1551 //***************************************************************************************************************
1552
1553