screenseqscommand.cpp

   1 /*
   2  *  screenseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 6/3/09.
   6  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 #include "screenseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ScreenSeqsCommand::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
 265         }
 266         catch(exception& e) {
 267                 m->errorOut(e, "ScreenSeqsCommand", "ScreenSeqsCommand");
 268                 exit(1);
 269         }
 270 }
 271 //***************************************************************************************************************
 272
 273 int ScreenSeqsCommand::execute(){
 274         try{
 275
 276                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 277
 278                 //if the user want to optimize we need to know the 90% mark
 279                 vector<unsigned long long> positions;
 280                 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
 281                         //use the namefile to optimize correctly
 282                         if (namefile != "") { nameMap = m->readNames(namefile); }
 283                         getSummary(positions);
 284                 }
 285                 else {
 286                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 287                         positions = m->divideFile(fastafile, processors);
 288                         for (int i = 0; i < (positions.size()-1); i++) {
 289                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 290                         }
 291                         #else
 292                                 positions.push_back(0); positions.push_back(1000);
 293                                 lines.push_back(new linePair(0, 1000));
 294                         #endif
 295                 }
 296
 297                 string goodSeqFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "good" + m->getExtension(fastafile);
 298                 string badAccnosFile =  outputDir + m->getRootName(m->getSimpleName(fastafile)) + "bad.accnos";
 299
 300                 int numFastaSeqs = 0;
 301                 set<string> badSeqNames;
 302                 int start = time(NULL);
 303
 304 #ifdef USE_MPI
 305                         int pid, numSeqsPerProcessor;
 306                         int tag = 2001;
 307                         vector<unsigned long long> MPIPos;
 308
 309                         MPI_Status status;
 310                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 311                         MPI_Comm_size(MPI_COMM_WORLD, &processors);
 312
 313                         MPI_File inMPI;
 314                         MPI_File outMPIGood;
 315                         MPI_File outMPIBadAccnos;
 316
 317                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 318                         int inMode=MPI_MODE_RDONLY;
 319
 320                         char outGoodFilename[1024];
 321                         strcpy(outGoodFilename, goodSeqFile.c_str());
 322
 323                         char outBadAccnosFilename[1024];
 324                         strcpy(outBadAccnosFilename, badAccnosFile.c_str());
 325
 326                         char inFileName[1024];
 327                         strcpy(inFileName, fastafile.c_str());
 328
 329                         MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 330                         MPI_File_open(MPI_COMM_WORLD, outGoodFilename, outMode, MPI_INFO_NULL, &outMPIGood);
 331                         MPI_File_open(MPI_COMM_WORLD, outBadAccnosFilename, outMode, MPI_INFO_NULL, &outMPIBadAccnos);
 332
 333                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood); MPI_File_close(&outMPIBadAccnos); return 0; }
 334
 335                         if (pid == 0) { //you are the root process
 336
 337                                 MPIPos = m->setFilePosFasta(fastafile, numFastaSeqs); //fills MPIPos, returns numSeqs
 338
 339                                 //send file positions to all processes
 340                                 for(int i = 1; i < processors; i++) {
 341                                         MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 342                                         MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 343                                 }
 344
 345                                 //figure out how many sequences you have to align
 346                                 numSeqsPerProcessor = numFastaSeqs / processors;
 347                                 int startIndex =  pid * numSeqsPerProcessor;
 348                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 349                 //      cout << pid << '\t' << numSeqsPerProcessor << '\t' <<   startIndex << endl;
 350                                 //align your part
 351                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 352                 //cout << pid << " done" << endl;
 353                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos);  return 0; }
 354
 355                                 for (int i = 1; i < processors; i++) {
 356
 357                                         //get bad lists
 358                                         int badSize;
 359                                         MPI_Recv(&badSize, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 360                                         /*for (int j = 0; j < badSize; j++) {
 361                                                 int length;
 362                                                 MPI_Recv(&length, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);  //recv the length of the name
 363                                                 char* buf2 = new char[length];                                                                          //make space to recieve it
 364                                                 MPI_Recv(buf2, length, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);  //get name
 365
 366                                                 string tempBuf = buf2;
 367                                                 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
 368                                                 delete buf2;
 369
 370                                                 badSeqNames.insert(tempBuf);
 371                                         }*/
 372                                 }
 373                         }else{ //you are a child process
 374                                 MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 375                                 MPIPos.resize(numFastaSeqs+1);
 376                                 MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 377
 378                                 //figure out how many sequences you have to align
 379                                 numSeqsPerProcessor = numFastaSeqs / processors;
 380                                 int startIndex =  pid * numSeqsPerProcessor;
 381                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 382                 //cout << pid << '\t' << numSeqsPerProcessor << '\t' <<         startIndex << endl;
 383                                 //align your part
 384                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 385 //cout << pid << " done" << endl;
 386                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos); return 0; }
 387
 388                                 //send bad list
 389                                 int badSize = badSeqNames.size();
 390                                 MPI_Send(&badSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 391
 392                                 /*
 393                                 set<string>::iterator it;
 394                                 for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 395                                         string name = *it;
 396                                         int length = name.length();
 397                                         char* buf2 = new char[length];
 398                                         memcpy(buf2, name.c_str(), length);
 399
 400                                         MPI_Send(&length, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 401                                         MPI_Send(buf2, length, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 402                                 }*/
 403                         }
 404
 405                         //close files
 406                         MPI_File_close(&inMPI);
 407                         MPI_File_close(&outMPIGood);
 408                         MPI_File_close(&outMPIBadAccnos);
 409                         MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 410
 411 #else
 412
 413         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 414                         if(processors == 1){
 415                                 numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 416
 417                                 if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 418
 419                         }else{
 420                                 processIDS.resize(0);
 421
 422                                 numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 423
 424                                 rename((goodSeqFile + toString(processIDS[0]) + ".temp").c_str(), goodSeqFile.c_str());
 425                                 rename((badAccnosFile + toString(processIDS[0]) + ".temp").c_str(), badAccnosFile.c_str());
 426
 427                                 //append alignment and report files
 428                                 for(int i=1;i<processors;i++){
 429                                         m->appendFiles((goodSeqFile + toString(processIDS[i]) + ".temp"), goodSeqFile);
 430                                         m->mothurRemove((goodSeqFile + toString(processIDS[i]) + ".temp"));
 431
 432                                         m->appendFiles((badAccnosFile + toString(processIDS[i]) + ".temp"), badAccnosFile);
 433                                         m->mothurRemove((badAccnosFile + toString(processIDS[i]) + ".temp"));
 434                                 }
 435
 436                                 if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 437
 438                                 //read badSeqs in because root process doesnt know what other "bad" seqs the children found
 439                                 ifstream inBad;
 440                                 int ableToOpen = m->openInputFile(badAccnosFile, inBad, "no error");
 441
 442                                 if (ableToOpen == 0) {
 443                                         badSeqNames.clear();
 444                                         string tempName;
 445                                         while (!inBad.eof()) {
 446                                                 inBad >> tempName; m->gobble(inBad);
 447                                                 badSeqNames.insert(tempName);
 448                                         }
 449                                         inBad.close();
 450                                 }
 451                         }
 452         #else
 453                         numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 454
 455                         if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 456
 457         #endif
 458
 459 #endif
 460
 461                 #ifdef USE_MPI
 462                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 463
 464                         if (pid == 0) { //only one process should fix files
 465
 466                                 //read accnos file with all names in it, process 0 just has its names
 467                                 MPI_File inMPIAccnos;
 468                                 MPI_Offset size;
 469
 470                                 char inFileName[1024];
 471                                 strcpy(inFileName, badAccnosFile.c_str());
 472
 473                                 MPI_File_open(MPI_COMM_SELF, inFileName, inMode, MPI_INFO_NULL, &inMPIAccnos);  //comm, filename, mode, info, filepointer
 474                                 MPI_File_get_size(inMPIAccnos, &size);
 475
 476                                 char* buffer = new char[size];
 477                                 MPI_File_read(inMPIAccnos, buffer, size, MPI_CHAR, &status);
 478
 479                                 string tempBuf = buffer;
 480                                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
 481                                 istringstream iss (tempBuf,istringstream::in);
 482
 483                                 delete buffer;
 484                                 MPI_File_close(&inMPIAccnos);
 485
 486                                 badSeqNames.clear();
 487                                 string tempName;
 488                                 while (!iss.eof()) {
 489                                         iss >> tempName; m->gobble(iss);
 490                                         badSeqNames.insert(tempName);
 491                                 }
 492                 #endif
 493
 494                 if(namefile != "" && groupfile != "")   {
 495                         screenNameGroupFile(badSeqNames);
 496                         if (m->control_pressed) {  m->mothurRemove(goodSeqFile); return 0; }
 497                 }else if(namefile != "")        {
 498                         screenNameGroupFile(badSeqNames);
 499                         if (m->control_pressed) {  m->mothurRemove(goodSeqFile);  return 0; }
 500                 }else if(groupfile != "")                               {       screenGroupFile(badSeqNames);           }       // this screens just the group
 501
 502                 if (m->control_pressed) { m->mothurRemove(goodSeqFile);  return 0; }
 503
 504                 if(alignreport != "")                                   {       screenAlignReport(badSeqNames);         }
 505                 if(qualfile != "")                                              {       screenQual(badSeqNames);                        }
 506                 if(taxonomy != "")                                              {       screenTaxonomy(badSeqNames);            }
 507
 508                 if (m->control_pressed) { m->mothurRemove(goodSeqFile);  return 0; }
 509
 510                 #ifdef USE_MPI
 511                         }
 512                 #endif
 513
 514                 m->mothurOutEndLine();
 515                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 516                 m->mothurOut(goodSeqFile); m->mothurOutEndLine();       outputTypes["fasta"].push_back(goodSeqFile);
 517                 m->mothurOut(badAccnosFile); m->mothurOutEndLine();      outputTypes["accnos"].push_back(badAccnosFile);
 518                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 519                 m->mothurOutEndLine();
 520                 m->mothurOutEndLine();
 521
 522                 //set fasta file as new current fastafile
 523                 string current = "";
 524                 itTypes = outputTypes.find("fasta");
 525                 if (itTypes != outputTypes.end()) {
 526                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 527                 }
 528
 529                 itTypes = outputTypes.find("name");
 530                 if (itTypes != outputTypes.end()) {
 531                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 532                 }
 533
 534                 itTypes = outputTypes.find("group");
 535                 if (itTypes != outputTypes.end()) {
 536                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 537                 }
 538
 539                 itTypes = outputTypes.find("qfile");
 540                 if (itTypes != outputTypes.end()) {
 541                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 542                 }
 543
 544                 itTypes = outputTypes.find("taxonomy");
 545                 if (itTypes != outputTypes.end()) {
 546                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 547                 }
 548
 549                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to screen " + toString(numFastaSeqs) + " sequences.");
 550                 m->mothurOutEndLine();
 551
 552                 return 0;
 553         }
 554         catch(exception& e) {
 555                 m->errorOut(e, "ScreenSeqsCommand", "execute");
 556                 exit(1);
 557         }
 558 }
 559
 560 //***************************************************************************************************************
 561
 562 int ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
 563         try {
 564                 ifstream inputNames;
 565                 m->openInputFile(namefile, inputNames);
 566                 set<string> badSeqGroups;
 567                 string seqName, seqList, group;
 568                 set<string>::iterator it;
 569
 570                 string goodNameFile = outputDir + m->getRootName(m->getSimpleName(namefile)) + "good" + m->getExtension(namefile);
 571                 outputNames.push_back(goodNameFile);  outputTypes["name"].push_back(goodNameFile);
 572
 573                 ofstream goodNameOut;   m->openOutputFile(goodNameFile, goodNameOut);
 574
 575                 while(!inputNames.eof()){
 576                         if (m->control_pressed) { goodNameOut.close();  inputNames.close(); m->mothurRemove(goodNameFile);  return 0; }
 577
 578                         inputNames >> seqName >> seqList;
 579                         it = badSeqNames.find(seqName);
 580
 581                         if(it != badSeqNames.end()){
 582                                 badSeqNames.erase(it);
 583
 584                                 if(namefile != ""){
 585                                         int start = 0;
 586                                         for(int i=0;i<seqList.length();i++){
 587                                                 if(seqList[i] == ','){
 588                                                         badSeqGroups.insert(seqList.substr(start,i-start));
 589                                                         start = i+1;
 590                                                 }
 591                                         }
 592                                         badSeqGroups.insert(seqList.substr(start,seqList.length()-start));
 593                                 }
 594                         }
 595                         else{
 596                                 goodNameOut << seqName << '\t' << seqList << endl;
 597                         }
 598                         m->gobble(inputNames);
 599                 }
 600                 inputNames.close();
 601                 goodNameOut.close();
 602
 603                 //we were unable to remove some of the bad sequences
 604                 if (badSeqNames.size() != 0) {
 605                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 606                                 m->mothurOut("Your namefile does not include the sequence " + *it + " please correct.");
 607                                 m->mothurOutEndLine();
 608                         }
 609                 }
 610
 611                 if(groupfile != ""){
 612
 613                         ifstream inputGroups;
 614                         m->openInputFile(groupfile, inputGroups);
 615
 616                         string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
 617                         outputNames.push_back(goodGroupFile);   outputTypes["group"].push_back(goodGroupFile);
 618
 619                         ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 620
 621                         while(!inputGroups.eof()){
 622                                 if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); m->mothurRemove(goodNameFile);  m->mothurRemove(goodGroupFile); return 0; }
 623
 624                                 inputGroups >> seqName >> group;
 625
 626                                 it = badSeqGroups.find(seqName);
 627
 628                                 if(it != badSeqGroups.end()){
 629                                         badSeqGroups.erase(it);
 630                                 }
 631                                 else{
 632                                         goodGroupOut << seqName << '\t' << group << endl;
 633                                 }
 634                                 m->gobble(inputGroups);
 635                         }
 636                         inputGroups.close();
 637                         goodGroupOut.close();
 638
 639                         //we were unable to remove some of the bad sequences
 640                         if (badSeqGroups.size() != 0) {
 641                                 for (it = badSeqGroups.begin(); it != badSeqGroups.end(); it++) {
 642                                         m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 643                                         m->mothurOutEndLine();
 644                                 }
 645                         }
 646                 }
 647
 648
 649                 return 0;
 650
 651         }
 652         catch(exception& e) {
 653                 m->errorOut(e, "ScreenSeqsCommand", "screenNameGroupFile");
 654                 exit(1);
 655         }
 656 }
 657 //***************************************************************************************************************
 658 int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
 659         try {
 660
 661                 vector<int> startPosition;
 662                 vector<int> endPosition;
 663                 vector<int> seqLength;
 664                 vector<int> ambigBases;
 665                 vector<int> longHomoPolymer;
 666
 667 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 668                 vector<unsigned long long> positions = m->divideFile(fastafile, processors);
 669
 670                 for (int i = 0; i < (positions.size()-1); i++) {
 671                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 672                 }
 673 #else
 674                 lines.push_back(new linePair(0, 1000));
 675 #endif
 676
 677 #ifdef USE_MPI
 678                 int pid;
 679                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 680
 681                 if (pid == 0) {
 682                         driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 683 #else
 684                 int numSeqs = 0;
 685                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 686                         if(processors == 1){
 687                                 numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 688                         }else{
 689                                 numSeqs = createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile);
 690                         }
 691
 692                         if (m->control_pressed) {  return 0; }
 693                 #else
 694                         numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 695                         if (m->control_pressed) {  return 0; }
 696                 #endif
 697 #endif
 698                 sort(startPosition.begin(), startPosition.end());
 699                 sort(endPosition.begin(), endPosition.end());
 700                 sort(seqLength.begin(), seqLength.end());
 701                 sort(ambigBases.begin(), ambigBases.end());
 702                 sort(longHomoPolymer.begin(), longHomoPolymer.end());
 703
 704                 //numSeqs is the number of unique seqs, startPosition.size() is the total number of seqs, we want to optimize using all seqs
 705                 int criteriaPercentile  = int(startPosition.size() * (criteria / (float) 100));
 706
 707                 for (int i = 0; i < optimize.size(); i++) {
 708                         if (optimize[i] == "start") { startPos = startPosition[criteriaPercentile]; m->mothurOut("Optimizing start to " + toString(startPos) + "."); m->mothurOutEndLine(); }
 709                         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();}
 710                         else if (optimize[i] == "maxambig") { maxAmbig = ambigBases[criteriaPercentile]; m->mothurOut("Optimizing maxambig to " + toString(maxAmbig) + "."); m->mothurOutEndLine(); }
 711                         else if (optimize[i] == "maxhomop") { maxHomoP = longHomoPolymer[criteriaPercentile]; m->mothurOut("Optimizing maxhomop to " + toString(maxHomoP) + "."); m->mothurOutEndLine(); }
 712                         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(); }
 713                         else if (optimize[i] == "maxlength") { maxLength = seqLength[criteriaPercentile]; m->mothurOut("Optimizing maxlength to " + toString(maxLength) + "."); m->mothurOutEndLine(); }
 714                 }
 715
 716 #ifdef USE_MPI
 717                 }
 718
 719                 MPI_Status status;
 720                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 721                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 722
 723                 if (pid == 0) {
 724                         //send file positions to all processes
 725                         for(int i = 1; i < processors; i++) {
 726                                 MPI_Send(&startPos, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 727                                 MPI_Send(&endPos, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 728                                 MPI_Send(&maxAmbig, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 729                                 MPI_Send(&maxHomoP, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 730                                 MPI_Send(&minLength, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 731                                 MPI_Send(&maxLength, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 732                         }
 733                 }else {
 734                         MPI_Recv(&startPos, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 735                         MPI_Recv(&endPos, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 736                         MPI_Recv(&maxAmbig, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 737                         MPI_Recv(&maxHomoP, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 738                         MPI_Recv(&minLength, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 739                         MPI_Recv(&maxLength, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 740                 }
 741                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 742 #endif
 743                 return 0;
 744         }
 745         catch(exception& e) {
 746                 m->errorOut(e, "ScreenSeqsCommand", "getSummary");
 747                 exit(1);
 748         }
 749 }
 750 /**************************************************************************************/
 751 int ScreenSeqsCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, linePair* filePos) {
 752         try {
 753
 754                 ifstream in;
 755                 m->openInputFile(filename, in);
 756
 757                 in.seekg(filePos->start);
 758
 759                 bool done = false;
 760                 int count = 0;
 761
 762                 while (!done) {
 763
 764                         if (m->control_pressed) { in.close(); return 1; }
 765
 766                         Sequence current(in); m->gobble(in);
 767
 768                         if (current.getName() != "") {
 769                                 int num = 1;
 770                                 if (namefile != "") {
 771                                         //make sure this sequence is in the namefile, else error
 772                                         map<string, int>::iterator it = nameMap.find(current.getName());
 773
 774                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 775                                         else { num = it->second; }
 776                                 }
 777
 778                                 //for each sequence this sequence represents
 779                                 for (int i = 0; i < num; i++) {
 780                                         startPosition.push_back(current.getStartPos());
 781                                         endPosition.push_back(current.getEndPos());
 782                                         seqLength.push_back(current.getNumBases());
 783                                         ambigBases.push_back(current.getAmbigBases());
 784                                         longHomoPolymer.push_back(current.getLongHomoPolymer());
 785                                 }
 786
 787                                 count++;
 788                         }
 789                         //if((count) % 100 == 0){       m->mothurOut("Optimizing sequence: " + toString(count)); m->mothurOutEndLine();         }
 790                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 791                                 unsigned long long pos = in.tellg();
 792                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 793                         #else
 794                                 if (in.eof()) { break; }
 795                         #endif
 796
 797                 }
 798
 799                 in.close();
 800
 801                 return count;
 802         }
 803         catch(exception& e) {
 804                 m->errorOut(e, "ScreenSeqsCommand", "driverCreateSummary");
 805                 exit(1);
 806         }
 807 }
 808 /**************************************************************************************************/
 809 int ScreenSeqsCommand::createProcessesCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename) {
 810         try {
 811 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 812                 int process = 1;
 813                 int num = 0;
 814                 processIDS.clear();
 815
 816                 //loop through and create all the processes you want
 817                 while (process != processors) {
 818                         int pid = fork();
 819
 820                         if (pid > 0) {
 821                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 822                                 process++;
 823                         }else if (pid == 0){
 824                                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[process]);
 825
 826                                 //pass numSeqs to parent
 827                                 ofstream out;
 828                                 string tempFile = fastafile + toString(getpid()) + ".num.temp";
 829                                 m->openOutputFile(tempFile, out);
 830
 831                                 out << num << endl;
 832                                 out << startPosition.size() << endl;
 833                                 for (int k = 0; k < startPosition.size(); k++)          {               out << startPosition[k] << '\t'; }  out << endl;
 834                                 for (int k = 0; k < endPosition.size(); k++)            {               out << endPosition[k] << '\t'; }  out << endl;
 835                                 for (int k = 0; k < seqLength.size(); k++)                      {               out << seqLength[k] << '\t'; }  out << endl;
 836                                 for (int k = 0; k < ambigBases.size(); k++)                     {               out << ambigBases[k] << '\t'; }  out << endl;
 837                                 for (int k = 0; k < longHomoPolymer.size(); k++)        {               out << longHomoPolymer[k] << '\t'; }  out << endl;
 838
 839                                 out.close();
 840
 841                                 exit(0);
 842                         }else {
 843                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 844                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 845                                 exit(0);
 846                         }
 847                 }
 848
 849                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 850
 851                 //force parent to wait until all the processes are done
 852                 for (int i=0;i<processIDS.size();i++) {
 853                         int temp = processIDS[i];
 854                         wait(&temp);
 855                 }
 856
 857                 //parent reads in and combine Filter info
 858                 for (int i = 0; i < processIDS.size(); i++) {
 859                         string tempFilename = fastafile + toString(processIDS[i]) + ".num.temp";
 860                         ifstream in;
 861                         m->openInputFile(tempFilename, in);
 862
 863                         int temp, tempNum;
 864                         in >> tempNum; m->gobble(in); num += tempNum;
 865                         in >> tempNum; m->gobble(in);
 866                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; startPosition.push_back(temp);              }               m->gobble(in);
 867                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; endPosition.push_back(temp);                }               m->gobble(in);
 868                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; seqLength.push_back(temp);                  }               m->gobble(in);
 869                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; ambigBases.push_back(temp);                 }               m->gobble(in);
 870                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; longHomoPolymer.push_back(temp);    }               m->gobble(in);
 871
 872                         in.close();
 873                         m->mothurRemove(tempFilename);
 874                 }
 875
 876                 return num;
 877 #endif
 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 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1274                 int process = 0;
1275                 int num = 0;
1276
1277                 //loop through and create all the processes you want
1278                 while (process != processors) {
1279                         int pid = fork();
1280
1281                         if (pid > 0) {
1282                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1283                                 process++;
1284                         }else if (pid == 0){
1285                                 num = driver(lines[process], goodFileName + toString(getpid()) + ".temp", badAccnos + toString(getpid()) + ".temp", filename, badSeqNames);
1286
1287                                 //pass numSeqs to parent
1288                                 ofstream out;
1289                                 string tempFile = filename + toString(getpid()) + ".num.temp";
1290                                 m->openOutputFile(tempFile, out);
1291                                 out << num << endl;
1292                                 out.close();
1293
1294                                 exit(0);
1295                         }else {
1296                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1297                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1298                                 exit(0);
1299                         }
1300                 }
1301
1302                 //force parent to wait until all the processes are done
1303                 for (int i=0;i<processors;i++) {
1304                         int temp = processIDS[i];
1305                         wait(&temp);
1306                 }
1307
1308                 for (int i = 0; i < processIDS.size(); i++) {
1309                         ifstream in;
1310                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
1311                         m->openInputFile(tempFile, in);
1312                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1313                         in.close(); m->mothurRemove(tempFile);
1314                 }
1315
1316                 return num;
1317 #endif
1318         }
1319         catch(exception& e) {
1320                 m->errorOut(e, "ScreenSeqsCommand", "createProcesses");
1321                 exit(1);
1322         }
1323 }
1324
1325 //***************************************************************************************************************
1326
1327