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