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                         convert(temp, startPos);
 224
 225                         temp = validParameter.validFile(parameters, "end", false);                      if (temp == "not found") { temp = "-1"; }
 226                         convert(temp, endPos);
 227
 228                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
 229                         convert(temp, maxAmbig);
 230
 231                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "-1"; }
 232                         convert(temp, maxHomoP);
 233
 234                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "-1"; }
 235                         convert(temp, minLength);
 236
 237                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "-1"; }
 238                         convert(temp, maxLength);
 239
 240                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 241                         m->setProcessors(temp);
 242                         convert(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                         convert(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                         positions = m->divideFile(fastafile, processors);
 287                         for (int i = 0; i < (positions.size()-1); i++) {
 288                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 289                         }
 290                 }
 291
 292                 string goodSeqFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "good" + m->getExtension(fastafile);
 293                 string badAccnosFile =  outputDir + m->getRootName(m->getSimpleName(fastafile)) + "bad.accnos";
 294
 295                 int numFastaSeqs = 0;
 296                 set<string> badSeqNames;
 297                 int start = time(NULL);
 298
 299 #ifdef USE_MPI
 300                         int pid, numSeqsPerProcessor;
 301                         int tag = 2001;
 302                         vector<unsigned long long> MPIPos;
 303
 304                         MPI_Status status;
 305                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 306                         MPI_Comm_size(MPI_COMM_WORLD, &processors);
 307
 308                         MPI_File inMPI;
 309                         MPI_File outMPIGood;
 310                         MPI_File outMPIBadAccnos;
 311
 312                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 313                         int inMode=MPI_MODE_RDONLY;
 314
 315                         char outGoodFilename[1024];
 316                         strcpy(outGoodFilename, goodSeqFile.c_str());
 317
 318                         char outBadAccnosFilename[1024];
 319                         strcpy(outBadAccnosFilename, badAccnosFile.c_str());
 320
 321                         char inFileName[1024];
 322                         strcpy(inFileName, fastafile.c_str());
 323
 324                         MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 325                         MPI_File_open(MPI_COMM_WORLD, outGoodFilename, outMode, MPI_INFO_NULL, &outMPIGood);
 326                         MPI_File_open(MPI_COMM_WORLD, outBadAccnosFilename, outMode, MPI_INFO_NULL, &outMPIBadAccnos);
 327
 328                         if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood); MPI_File_close(&outMPIBadAccnos); return 0; }
 329
 330                         if (pid == 0) { //you are the root process
 331
 332                                 MPIPos = m->setFilePosFasta(fastafile, numFastaSeqs); //fills MPIPos, returns numSeqs
 333
 334                                 //send file positions to all processes
 335                                 for(int i = 1; i < processors; i++) {
 336                                         MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 337                                         MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 338                                 }
 339
 340                                 //figure out how many sequences you have to align
 341                                 numSeqsPerProcessor = numFastaSeqs / processors;
 342                                 int startIndex =  pid * numSeqsPerProcessor;
 343                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 344                 //      cout << pid << '\t' << numSeqsPerProcessor << '\t' <<   startIndex << endl;
 345                                 //align your part
 346                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 347                 //cout << pid << " done" << endl;
 348                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos);  return 0; }
 349
 350                                 for (int i = 1; i < processors; i++) {
 351
 352                                         //get bad lists
 353                                         int badSize;
 354                                         MPI_Recv(&badSize, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 355                                         /*for (int j = 0; j < badSize; j++) {
 356                                                 int length;
 357                                                 MPI_Recv(&length, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);  //recv the length of the name
 358                                                 char* buf2 = new char[length];                                                                          //make space to recieve it
 359                                                 MPI_Recv(buf2, length, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);  //get name
 360
 361                                                 string tempBuf = buf2;
 362                                                 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
 363                                                 delete buf2;
 364
 365                                                 badSeqNames.insert(tempBuf);
 366                                         }*/
 367                                 }
 368                         }else{ //you are a child process
 369                                 MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 370                                 MPIPos.resize(numFastaSeqs+1);
 371                                 MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 372
 373                                 //figure out how many sequences you have to align
 374                                 numSeqsPerProcessor = numFastaSeqs / processors;
 375                                 int startIndex =  pid * numSeqsPerProcessor;
 376                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 377                 //cout << pid << '\t' << numSeqsPerProcessor << '\t' <<         startIndex << endl;
 378                                 //align your part
 379                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIGood, outMPIBadAccnos, MPIPos, badSeqNames);
 380 //cout << pid << " done" << endl;
 381                                 if (m->control_pressed) { MPI_File_close(&inMPI);  MPI_File_close(&outMPIGood);  MPI_File_close(&outMPIBadAccnos); return 0; }
 382
 383                                 //send bad list
 384                                 int badSize = badSeqNames.size();
 385                                 MPI_Send(&badSize, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 386
 387                                 /*
 388                                 set<string>::iterator it;
 389                                 for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 390                                         string name = *it;
 391                                         int length = name.length();
 392                                         char* buf2 = new char[length];
 393                                         memcpy(buf2, name.c_str(), length);
 394
 395                                         MPI_Send(&length, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 396                                         MPI_Send(buf2, length, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 397                                 }*/
 398                         }
 399
 400                         //close files
 401                         MPI_File_close(&inMPI);
 402                         MPI_File_close(&outMPIGood);
 403                         MPI_File_close(&outMPIBadAccnos);
 404                         MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 405
 406 #else
 407
 408         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 409                         if(processors == 1){
 410                                 numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 411
 412                                 if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 413
 414                         }else{
 415                                 processIDS.resize(0);
 416
 417                                 numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 418
 419                                 rename((goodSeqFile + toString(processIDS[0]) + ".temp").c_str(), goodSeqFile.c_str());
 420                                 rename((badAccnosFile + toString(processIDS[0]) + ".temp").c_str(), badAccnosFile.c_str());
 421
 422                                 //append alignment and report files
 423                                 for(int i=1;i<processors;i++){
 424                                         m->appendFiles((goodSeqFile + toString(processIDS[i]) + ".temp"), goodSeqFile);
 425                                         m->mothurRemove((goodSeqFile + toString(processIDS[i]) + ".temp"));
 426
 427                                         m->appendFiles((badAccnosFile + toString(processIDS[i]) + ".temp"), badAccnosFile);
 428                                         m->mothurRemove((badAccnosFile + toString(processIDS[i]) + ".temp"));
 429                                 }
 430
 431                                 if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 432
 433                                 //read badSeqs in because root process doesnt know what other "bad" seqs the children found
 434                                 ifstream inBad;
 435                                 int ableToOpen = m->openInputFile(badAccnosFile, inBad, "no error");
 436
 437                                 if (ableToOpen == 0) {
 438                                         badSeqNames.clear();
 439                                         string tempName;
 440                                         while (!inBad.eof()) {
 441                                                 inBad >> tempName; m->gobble(inBad);
 442                                                 badSeqNames.insert(tempName);
 443                                         }
 444                                         inBad.close();
 445                                 }
 446                         }
 447         #else
 448                         numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);
 449
 450                         if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 451
 452         #endif
 453
 454 #endif
 455
 456                 #ifdef USE_MPI
 457                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 458
 459                         if (pid == 0) { //only one process should fix files
 460
 461                                 //read accnos file with all names in it, process 0 just has its names
 462                                 MPI_File inMPIAccnos;
 463                                 MPI_Offset size;
 464
 465                                 char inFileName[1024];
 466                                 strcpy(inFileName, badAccnosFile.c_str());
 467
 468                                 MPI_File_open(MPI_COMM_SELF, inFileName, inMode, MPI_INFO_NULL, &inMPIAccnos);  //comm, filename, mode, info, filepointer
 469                                 MPI_File_get_size(inMPIAccnos, &size);
 470
 471                                 char* buffer = new char[size];
 472                                 MPI_File_read(inMPIAccnos, buffer, size, MPI_CHAR, &status);
 473
 474                                 string tempBuf = buffer;
 475                                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
 476                                 istringstream iss (tempBuf,istringstream::in);
 477
 478                                 delete buffer;
 479                                 MPI_File_close(&inMPIAccnos);
 480
 481                                 badSeqNames.clear();
 482                                 string tempName;
 483                                 while (!iss.eof()) {
 484                                         iss >> tempName; m->gobble(iss);
 485                                         badSeqNames.insert(tempName);
 486                                 }
 487                 #endif
 488
 489                 if(namefile != "" && groupfile != "")   {
 490                         screenNameGroupFile(badSeqNames);
 491                         if (m->control_pressed) {  m->mothurRemove(goodSeqFile); return 0; }
 492                 }else if(namefile != "")        {
 493                         screenNameGroupFile(badSeqNames);
 494                         if (m->control_pressed) {  m->mothurRemove(goodSeqFile);  return 0; }
 495                 }else if(groupfile != "")                               {       screenGroupFile(badSeqNames);           }       // this screens just the group
 496
 497                 if (m->control_pressed) { m->mothurRemove(goodSeqFile);  return 0; }
 498
 499                 if(alignreport != "")                                   {       screenAlignReport(badSeqNames);         }
 500                 if(qualfile != "")                                              {       screenQual(badSeqNames);                        }
 501                 if(taxonomy != "")                                              {       screenTaxonomy(badSeqNames);            }
 502
 503                 if (m->control_pressed) { m->mothurRemove(goodSeqFile);  return 0; }
 504
 505                 #ifdef USE_MPI
 506                         }
 507                 #endif
 508
 509                 m->mothurOutEndLine();
 510                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 511                 m->mothurOut(goodSeqFile); m->mothurOutEndLine();       outputTypes["fasta"].push_back(goodSeqFile);
 512                 m->mothurOut(badAccnosFile); m->mothurOutEndLine();      outputTypes["accnos"].push_back(badAccnosFile);
 513                 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
 514                 m->mothurOutEndLine();
 515                 m->mothurOutEndLine();
 516
 517                 //set fasta file as new current fastafile
 518                 string current = "";
 519                 itTypes = outputTypes.find("fasta");
 520                 if (itTypes != outputTypes.end()) {
 521                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 522                 }
 523
 524                 itTypes = outputTypes.find("name");
 525                 if (itTypes != outputTypes.end()) {
 526                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 527                 }
 528
 529                 itTypes = outputTypes.find("group");
 530                 if (itTypes != outputTypes.end()) {
 531                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 532                 }
 533
 534                 itTypes = outputTypes.find("qfile");
 535                 if (itTypes != outputTypes.end()) {
 536                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 537                 }
 538
 539                 itTypes = outputTypes.find("taxonomy");
 540                 if (itTypes != outputTypes.end()) {
 541                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 542                 }
 543
 544                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to screen " + toString(numFastaSeqs) + " sequences.");
 545                 m->mothurOutEndLine();
 546
 547                 return 0;
 548         }
 549         catch(exception& e) {
 550                 m->errorOut(e, "ScreenSeqsCommand", "execute");
 551                 exit(1);
 552         }
 553 }
 554
 555 //***************************************************************************************************************
 556
 557 int ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
 558         try {
 559                 ifstream inputNames;
 560                 m->openInputFile(namefile, inputNames);
 561                 set<string> badSeqGroups;
 562                 string seqName, seqList, group;
 563                 set<string>::iterator it;
 564
 565                 string goodNameFile = outputDir + m->getRootName(m->getSimpleName(namefile)) + "good" + m->getExtension(namefile);
 566                 outputNames.push_back(goodNameFile);  outputTypes["name"].push_back(goodNameFile);
 567
 568                 ofstream goodNameOut;   m->openOutputFile(goodNameFile, goodNameOut);
 569
 570                 while(!inputNames.eof()){
 571                         if (m->control_pressed) { goodNameOut.close();  inputNames.close(); m->mothurRemove(goodNameFile);  return 0; }
 572
 573                         inputNames >> seqName >> seqList;
 574                         it = badSeqNames.find(seqName);
 575
 576                         if(it != badSeqNames.end()){
 577                                 badSeqNames.erase(it);
 578
 579                                 if(namefile != ""){
 580                                         int start = 0;
 581                                         for(int i=0;i<seqList.length();i++){
 582                                                 if(seqList[i] == ','){
 583                                                         badSeqGroups.insert(seqList.substr(start,i-start));
 584                                                         start = i+1;
 585                                                 }
 586                                         }
 587                                         badSeqGroups.insert(seqList.substr(start,seqList.length()-start));
 588                                 }
 589                         }
 590                         else{
 591                                 goodNameOut << seqName << '\t' << seqList << endl;
 592                         }
 593                         m->gobble(inputNames);
 594                 }
 595                 inputNames.close();
 596                 goodNameOut.close();
 597
 598                 //we were unable to remove some of the bad sequences
 599                 if (badSeqNames.size() != 0) {
 600                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 601                                 m->mothurOut("Your namefile does not include the sequence " + *it + " please correct.");
 602                                 m->mothurOutEndLine();
 603                         }
 604                 }
 605
 606                 if(groupfile != ""){
 607
 608                         ifstream inputGroups;
 609                         m->openInputFile(groupfile, inputGroups);
 610
 611                         string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
 612                         outputNames.push_back(goodGroupFile);   outputTypes["group"].push_back(goodGroupFile);
 613
 614                         ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 615
 616                         while(!inputGroups.eof()){
 617                                 if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); m->mothurRemove(goodNameFile);  m->mothurRemove(goodGroupFile); return 0; }
 618
 619                                 inputGroups >> seqName >> group;
 620
 621                                 it = badSeqGroups.find(seqName);
 622
 623                                 if(it != badSeqGroups.end()){
 624                                         badSeqGroups.erase(it);
 625                                 }
 626                                 else{
 627                                         goodGroupOut << seqName << '\t' << group << endl;
 628                                 }
 629                                 m->gobble(inputGroups);
 630                         }
 631                         inputGroups.close();
 632                         goodGroupOut.close();
 633
 634                         //we were unable to remove some of the bad sequences
 635                         if (badSeqGroups.size() != 0) {
 636                                 for (it = badSeqGroups.begin(); it != badSeqGroups.end(); it++) {
 637                                         m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 638                                         m->mothurOutEndLine();
 639                                 }
 640                         }
 641                 }
 642
 643
 644                 return 0;
 645
 646         }
 647         catch(exception& e) {
 648                 m->errorOut(e, "ScreenSeqsCommand", "screenNameGroupFile");
 649                 exit(1);
 650         }
 651 }
 652 //***************************************************************************************************************
 653 int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
 654         try {
 655
 656                 vector<int> startPosition;
 657                 vector<int> endPosition;
 658                 vector<int> seqLength;
 659                 vector<int> ambigBases;
 660                 vector<int> longHomoPolymer;
 661
 662                 vector<unsigned long long> positions = m->divideFile(fastafile, processors);
 663
 664                 for (int i = 0; i < (positions.size()-1); i++) {
 665                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 666                 }
 667
 668
 669 #ifdef USE_MPI
 670                 int pid;
 671                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 672
 673                 if (pid == 0) {
 674                         driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 675 #else
 676                 int numSeqs = 0;
 677                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 678                         if(processors == 1){
 679                                 numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 680                         }else{
 681                                 numSeqs = createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile);
 682                         }
 683
 684                         if (m->control_pressed) {  return 0; }
 685                 #else
 686                         numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 687                         if (m->control_pressed) {  return 0; }
 688                 #endif
 689 #endif
 690                 sort(startPosition.begin(), startPosition.end());
 691                 sort(endPosition.begin(), endPosition.end());
 692                 sort(seqLength.begin(), seqLength.end());
 693                 sort(ambigBases.begin(), ambigBases.end());
 694                 sort(longHomoPolymer.begin(), longHomoPolymer.end());
 695
 696                 //numSeqs is the number of unique seqs, startPosition.size() is the total number of seqs, we want to optimize using all seqs
 697                 int criteriaPercentile  = int(startPosition.size() * (criteria / (float) 100));
 698
 699                 for (int i = 0; i < optimize.size(); i++) {
 700                         if (optimize[i] == "start") { startPos = startPosition[criteriaPercentile]; m->mothurOut("Optimizing start to " + toString(startPos) + "."); m->mothurOutEndLine(); }
 701                         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();}
 702                         else if (optimize[i] == "maxambig") { maxAmbig = ambigBases[criteriaPercentile]; m->mothurOut("Optimizing maxambig to " + toString(maxAmbig) + "."); m->mothurOutEndLine(); }
 703                         else if (optimize[i] == "maxhomop") { maxHomoP = longHomoPolymer[criteriaPercentile]; m->mothurOut("Optimizing maxhomop to " + toString(maxHomoP) + "."); m->mothurOutEndLine(); }
 704                         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(); }
 705                         else if (optimize[i] == "maxlength") { maxLength = seqLength[criteriaPercentile]; m->mothurOut("Optimizing maxlength to " + toString(maxLength) + "."); m->mothurOutEndLine(); }
 706                 }
 707
 708 #ifdef USE_MPI
 709                 }
 710
 711                 MPI_Status status;
 712                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 713                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 714
 715                 if (pid == 0) {
 716                         //send file positions to all processes
 717                         for(int i = 1; i < processors; i++) {
 718                                 MPI_Send(&startPos, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 719                                 MPI_Send(&endPos, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 720                                 MPI_Send(&maxAmbig, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 721                                 MPI_Send(&maxHomoP, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 722                                 MPI_Send(&minLength, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 723                                 MPI_Send(&maxLength, 1, MPI_INT, i, 2001, MPI_COMM_WORLD);
 724                         }
 725                 }else {
 726                         MPI_Recv(&startPos, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 727                         MPI_Recv(&endPos, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 728                         MPI_Recv(&maxAmbig, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 729                         MPI_Recv(&maxHomoP, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 730                         MPI_Recv(&minLength, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 731                         MPI_Recv(&maxLength, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD, &status);
 732                 }
 733                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 734 #endif
 735                 return 0;
 736         }
 737         catch(exception& e) {
 738                 m->errorOut(e, "ScreenSeqsCommand", "getSummary");
 739                 exit(1);
 740         }
 741 }
 742 /**************************************************************************************/
 743 int ScreenSeqsCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, linePair* filePos) {
 744         try {
 745
 746                 ifstream in;
 747                 m->openInputFile(filename, in);
 748
 749                 in.seekg(filePos->start);
 750
 751                 bool done = false;
 752                 int count = 0;
 753
 754                 while (!done) {
 755
 756                         if (m->control_pressed) { in.close(); return 1; }
 757
 758                         Sequence current(in); m->gobble(in);
 759
 760                         if (current.getName() != "") {
 761                                 int num = 1;
 762                                 if (namefile != "") {
 763                                         //make sure this sequence is in the namefile, else error
 764                                         map<string, int>::iterator it = nameMap.find(current.getName());
 765
 766                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 767                                         else { num = it->second; }
 768                                 }
 769
 770                                 //for each sequence this sequence represents
 771                                 for (int i = 0; i < num; i++) {
 772                                         startPosition.push_back(current.getStartPos());
 773                                         endPosition.push_back(current.getEndPos());
 774                                         seqLength.push_back(current.getNumBases());
 775                                         ambigBases.push_back(current.getAmbigBases());
 776                                         longHomoPolymer.push_back(current.getLongHomoPolymer());
 777                                 }
 778
 779                                 count++;
 780                         }
 781
 782                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 783                                 unsigned long long pos = in.tellg();
 784                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 785                         #else
 786                                 if (in.eof()) { break; }
 787                         #endif
 788
 789                 }
 790
 791                 in.close();
 792
 793                 return count;
 794         }
 795         catch(exception& e) {
 796                 m->errorOut(e, "ScreenSeqsCommand", "driverCreateSummary");
 797                 exit(1);
 798         }
 799 }
 800 /**************************************************************************************************/
 801 int ScreenSeqsCommand::createProcessesCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename) {
 802         try {
 803 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 804                 int process = 1;
 805                 int num = 0;
 806                 processIDS.clear();
 807
 808                 //loop through and create all the processes you want
 809                 while (process != processors) {
 810                         int pid = fork();
 811
 812                         if (pid > 0) {
 813                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 814                                 process++;
 815                         }else if (pid == 0){
 816                                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[process]);
 817
 818                                 //pass numSeqs to parent
 819                                 ofstream out;
 820                                 string tempFile = fastafile + toString(getpid()) + ".num.temp";
 821                                 m->openOutputFile(tempFile, out);
 822
 823                                 out << num << endl;
 824                                 out << startPosition.size() << endl;
 825                                 for (int k = 0; k < startPosition.size(); k++)          {               out << startPosition[k] << '\t'; }  out << endl;
 826                                 for (int k = 0; k < endPosition.size(); k++)            {               out << endPosition[k] << '\t'; }  out << endl;
 827                                 for (int k = 0; k < seqLength.size(); k++)                      {               out << seqLength[k] << '\t'; }  out << endl;
 828                                 for (int k = 0; k < ambigBases.size(); k++)                     {               out << ambigBases[k] << '\t'; }  out << endl;
 829                                 for (int k = 0; k < longHomoPolymer.size(); k++)        {               out << longHomoPolymer[k] << '\t'; }  out << endl;
 830
 831                                 out.close();
 832
 833                                 exit(0);
 834                         }else {
 835                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 836                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 837                                 exit(0);
 838                         }
 839                 }
 840
 841                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 842
 843                 //force parent to wait until all the processes are done
 844                 for (int i=0;i<processIDS.size();i++) {
 845                         int temp = processIDS[i];
 846                         wait(&temp);
 847                 }
 848
 849                 //parent reads in and combine Filter info
 850                 for (int i = 0; i < processIDS.size(); i++) {
 851                         string tempFilename = fastafile + toString(processIDS[i]) + ".num.temp";
 852                         ifstream in;
 853                         m->openInputFile(tempFilename, in);
 854
 855                         int temp, tempNum;
 856                         in >> tempNum; m->gobble(in); num += tempNum;
 857                         in >> tempNum; m->gobble(in);
 858                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; startPosition.push_back(temp);              }               m->gobble(in);
 859                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; endPosition.push_back(temp);                }               m->gobble(in);
 860                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; seqLength.push_back(temp);                  }               m->gobble(in);
 861                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; ambigBases.push_back(temp);                 }               m->gobble(in);
 862                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; longHomoPolymer.push_back(temp);    }               m->gobble(in);
 863
 864                         in.close();
 865                         m->mothurRemove(tempFilename);
 866                 }
 867
 868                 return num;
 869 #endif
 870         }
 871         catch(exception& e) {
 872                 m->errorOut(e, "ScreenSeqsCommand", "createProcessesCreateSummary");
 873                 exit(1);
 874         }
 875 }
 876
 877 //***************************************************************************************************************
 878
 879 int ScreenSeqsCommand::screenGroupFile(set<string> badSeqNames){
 880         try {
 881                 ifstream inputGroups;
 882                 m->openInputFile(groupfile, inputGroups);
 883                 string seqName, group;
 884                 set<string>::iterator it;
 885
 886                 string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
 887                 outputNames.push_back(goodGroupFile);  outputTypes["group"].push_back(goodGroupFile);
 888                 ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
 889
 890                 while(!inputGroups.eof()){
 891                         if (m->control_pressed) { goodGroupOut.close(); inputGroups.close(); m->mothurRemove(goodGroupFile); return 0; }
 892
 893                         inputGroups >> seqName >> group;
 894                         it = badSeqNames.find(seqName);
 895
 896                         if(it != badSeqNames.end()){
 897                                 badSeqNames.erase(it);
 898                         }
 899                         else{
 900                                 goodGroupOut << seqName << '\t' << group << endl;
 901                         }
 902                         m->gobble(inputGroups);
 903                 }
 904
 905                 if (m->control_pressed) { goodGroupOut.close();  inputGroups.close(); m->mothurRemove(goodGroupFile);  return 0; }
 906
 907                 //we were unable to remove some of the bad sequences
 908                 if (badSeqNames.size() != 0) {
 909                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 910                                 m->mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
 911                                 m->mothurOutEndLine();
 912                         }
 913                 }
 914
 915                 inputGroups.close();
 916                 goodGroupOut.close();
 917
 918                 if (m->control_pressed) { m->mothurRemove(goodGroupFile);   }
 919
 920                 return 0;
 921
 922         }
 923         catch(exception& e) {
 924                 m->errorOut(e, "ScreenSeqsCommand", "screenGroupFile");
 925                 exit(1);
 926         }
 927 }
 928
 929 //***************************************************************************************************************
 930
 931 int ScreenSeqsCommand::screenAlignReport(set<string> badSeqNames){
 932         try {
 933                 ifstream inputAlignReport;
 934                 m->openInputFile(alignreport, inputAlignReport);
 935                 string seqName, group;
 936                 set<string>::iterator it;
 937
 938                 string goodAlignReportFile = outputDir + m->getRootName(m->getSimpleName(alignreport)) + "good" + m->getExtension(alignreport);
 939                 outputNames.push_back(goodAlignReportFile);  outputTypes["alignreport"].push_back(goodAlignReportFile);
 940                 ofstream goodAlignReportOut;    m->openOutputFile(goodAlignReportFile, goodAlignReportOut);
 941
 942                 while (!inputAlignReport.eof()) {               //      need to copy header
 943                         char c = inputAlignReport.get();
 944                         goodAlignReportOut << c;
 945                         if (c == 10 || c == 13){        break;  }
 946                 }
 947
 948                 while(!inputAlignReport.eof()){
 949                         if (m->control_pressed) { goodAlignReportOut.close(); inputAlignReport.close(); m->mothurRemove(goodAlignReportFile); return 0; }
 950
 951                         inputAlignReport >> seqName;
 952                         it = badSeqNames.find(seqName);
 953                         string line;
 954                         while (!inputAlignReport.eof()) {               //      need to copy header
 955                                 char c = inputAlignReport.get();
 956                                 line += c;
 957                                 if (c == 10 || c == 13){        break;  }
 958                         }
 959
 960                         if(it != badSeqNames.end()){
 961                                 badSeqNames.erase(it);
 962                         }
 963                         else{
 964                                 goodAlignReportOut << seqName << '\t' << line;
 965                         }
 966                         m->gobble(inputAlignReport);
 967                 }
 968
 969                 if (m->control_pressed) { goodAlignReportOut.close();  inputAlignReport.close(); m->mothurRemove(goodAlignReportFile);  return 0; }
 970
 971                 //we were unable to remove some of the bad sequences
 972                 if (badSeqNames.size() != 0) {
 973                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
 974                                 m->mothurOut("Your alignreport file does not include the sequence " + *it + " please correct.");
 975                                 m->mothurOutEndLine();
 976                         }
 977                 }
 978
 979                 inputAlignReport.close();
 980                 goodAlignReportOut.close();
 981
 982                 if (m->control_pressed) {  m->mothurRemove(goodAlignReportFile);  return 0; }
 983
 984                 return 0;
 985
 986         }
 987         catch(exception& e) {
 988                 m->errorOut(e, "ScreenSeqsCommand", "screenAlignReport");
 989                 exit(1);
 990         }
 991
 992 }
 993 //***************************************************************************************************************
 994
 995 int ScreenSeqsCommand::screenTaxonomy(set<string> badSeqNames){
 996         try {
 997                 ifstream input;
 998                 m->openInputFile(taxonomy, input);
 999                 string seqName, tax;
1000                 set<string>::iterator it;
1001
1002                 string goodTaxFile = outputDir + m->getRootName(m->getSimpleName(taxonomy)) + "good" + m->getExtension(taxonomy);
1003                 outputNames.push_back(goodTaxFile);  outputTypes["taxonomy"].push_back(goodTaxFile);
1004                 ofstream goodTaxOut;    m->openOutputFile(goodTaxFile, goodTaxOut);
1005
1006                 while(!input.eof()){
1007                         if (m->control_pressed) { goodTaxOut.close(); input.close(); m->mothurRemove(goodTaxFile); return 0; }
1008
1009                         input >> seqName >> tax;
1010                         it = badSeqNames.find(seqName);
1011
1012                         if(it != badSeqNames.end()){ badSeqNames.erase(it); }
1013                         else{
1014                                 goodTaxOut << seqName << '\t' << tax << endl;
1015                         }
1016                         m->gobble(input);
1017                 }
1018
1019                 if (m->control_pressed) { goodTaxOut.close(); input.close(); m->mothurRemove(goodTaxFile); return 0; }
1020
1021                 //we were unable to remove some of the bad sequences
1022                 if (badSeqNames.size() != 0) {
1023                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
1024                                 m->mothurOut("Your taxonomy file does not include the sequence " + *it + " please correct.");
1025                                 m->mothurOutEndLine();
1026                         }
1027                 }
1028
1029                 input.close();
1030                 goodTaxOut.close();
1031
1032                 if (m->control_pressed) {  m->mothurRemove(goodTaxFile);  return 0; }
1033
1034                 return 0;
1035
1036         }
1037         catch(exception& e) {
1038                 m->errorOut(e, "ScreenSeqsCommand", "screenTaxonomy");
1039                 exit(1);
1040         }
1041
1042 }
1043 //***************************************************************************************************************
1044
1045 int ScreenSeqsCommand::screenQual(set<string> badSeqNames){
1046         try {
1047                 ifstream in;
1048                 m->openInputFile(qualfile, in);
1049                 set<string>::iterator it;
1050
1051                 string goodQualFile = outputDir + m->getRootName(m->getSimpleName(qualfile)) + "good" + m->getExtension(qualfile);
1052                 outputNames.push_back(goodQualFile);  outputTypes["qfile"].push_back(goodQualFile);
1053                 ofstream goodQual;      m->openOutputFile(goodQualFile, goodQual);
1054
1055                 while(!in.eof()){
1056
1057                         if (m->control_pressed) { goodQual.close(); in.close(); m->mothurRemove(goodQualFile); return 0; }
1058
1059                         string saveName = "";
1060                         string name = "";
1061                         string scores = "";
1062
1063                         in >> name;
1064
1065                         if (name.length() != 0) {
1066                                 saveName = name.substr(1);
1067                                 while (!in.eof())       {
1068                                         char c = in.get();
1069                                         if (c == 10 || c == 13){        break;  }
1070                                         else { name += c; }
1071                                 }
1072                                 m->gobble(in);
1073                         }
1074
1075                         while(in){
1076                                 char letter= in.get();
1077                                 if(letter == '>'){      in.putback(letter);     break;  }
1078                                 else{ scores += letter; }
1079                         }
1080
1081                         m->gobble(in);
1082
1083                         it = badSeqNames.find(saveName);
1084
1085                         if(it != badSeqNames.end()){
1086                                 badSeqNames.erase(it);
1087                         }else{
1088                                 goodQual << name << endl << scores;
1089                         }
1090
1091                         m->gobble(in);
1092                 }
1093
1094                 in.close();
1095                 goodQual.close();
1096
1097                 //we were unable to remove some of the bad sequences
1098                 if (badSeqNames.size() != 0) {
1099                         for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
1100                                 m->mothurOut("Your qual file does not include the sequence " + *it + " please correct.");
1101                                 m->mothurOutEndLine();
1102                         }
1103                 }
1104
1105                 if (m->control_pressed) {  m->mothurRemove(goodQualFile);  return 0; }
1106
1107                 return 0;
1108
1109         }
1110         catch(exception& e) {
1111                 m->errorOut(e, "ScreenSeqsCommand", "screenQual");
1112                 exit(1);
1113         }
1114
1115 }
1116 //**********************************************************************************************************************
1117
1118 int ScreenSeqsCommand::driver(linePair* filePos, string goodFName, string badAccnosFName, string filename, set<string>& badSeqNames){
1119         try {
1120                 ofstream goodFile;
1121                 m->openOutputFile(goodFName, goodFile);
1122
1123                 ofstream badAccnosFile;
1124                 m->openOutputFile(badAccnosFName, badAccnosFile);
1125
1126                 ifstream inFASTA;
1127                 m->openInputFile(filename, inFASTA);
1128
1129                 inFASTA.seekg(filePos->start);
1130
1131                 bool done = false;
1132                 int count = 0;
1133
1134                 while (!done) {
1135
1136                         if (m->control_pressed) {  return 0; }
1137
1138                         Sequence currSeq(inFASTA); m->gobble(inFASTA);
1139                         if (currSeq.getName() != "") {
1140                                 bool goodSeq = 1;               //      innocent until proven guilty
1141                                 if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
1142                                 if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
1143                                 if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
1144                                 if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
1145                                 if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
1146                                 if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
1147
1148                                 if(goodSeq == 1){
1149                                         currSeq.printSequence(goodFile);
1150                                 }
1151                                 else{
1152                                         badAccnosFile << currSeq.getName() << endl;
1153                                         badSeqNames.insert(currSeq.getName());
1154                                 }
1155                         count++;
1156                         }
1157
1158                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1159                                 unsigned long long pos = inFASTA.tellg();
1160                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
1161                         #else
1162                                 if (inFASTA.eof()) { break; }
1163                         #endif
1164
1165                         //report progress
1166                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1167                 }
1168                 //report progress
1169                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1170
1171
1172                 goodFile.close();
1173                 inFASTA.close();
1174                 badAccnosFile.close();
1175
1176                 return count;
1177         }
1178         catch(exception& e) {
1179                 m->errorOut(e, "ScreenSeqsCommand", "driver");
1180                 exit(1);
1181         }
1182 }
1183 //**********************************************************************************************************************
1184 #ifdef USE_MPI
1185 int ScreenSeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& goodFile, MPI_File& badAccnosFile, vector<unsigned long long>& MPIPos, set<string>& badSeqNames){
1186         try {
1187                 string outputString = "";
1188                 MPI_Status statusGood;
1189                 MPI_Status statusBadAccnos;
1190                 MPI_Status status;
1191                 int pid;
1192                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
1193
1194                 for(int i=0;i<num;i++){
1195
1196                         if (m->control_pressed) {  return 0; }
1197
1198                         //read next sequence
1199                         int length = MPIPos[start+i+1] - MPIPos[start+i];
1200
1201                         char* buf4 = new char[length];
1202                         memcpy(buf4, outputString.c_str(), length);
1203
1204                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
1205
1206                         string tempBuf = buf4;  delete buf4;
1207                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
1208                         istringstream iss (tempBuf,istringstream::in);
1209
1210                         Sequence currSeq(iss);
1211
1212                         //process seq
1213                         if (currSeq.getName() != "") {
1214                                 bool goodSeq = 1;               //      innocent until proven guilty
1215                                 if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
1216                                 if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
1217                                 if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
1218                                 if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
1219                                 if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
1220                                 if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
1221
1222                                 if(goodSeq == 1){
1223                                         outputString =  ">" + currSeq.getName() + "\n" + currSeq.getAligned() + "\n";
1224
1225                                         //print good seq
1226                                         length = outputString.length();
1227                                         char* buf2 = new char[length];
1228                                         memcpy(buf2, outputString.c_str(), length);
1229
1230                                         MPI_File_write_shared(goodFile, buf2, length, MPI_CHAR, &statusGood);
1231                                         delete buf2;
1232                                 }
1233                                 else{
1234
1235                                         badSeqNames.insert(currSeq.getName());
1236
1237                                         //write to bad accnos file
1238                                         outputString = currSeq.getName() + "\n";
1239
1240                                         length = outputString.length();
1241                                         char* buf3 = new char[length];
1242                                         memcpy(buf3, outputString.c_str(), length);
1243
1244                                         MPI_File_write_shared(badAccnosFile, buf3, length, MPI_CHAR, &statusBadAccnos);
1245                                         delete buf3;
1246                                 }
1247                         }
1248
1249                         //report progress
1250                         if((i) % 100 == 0){     m->mothurOut("Processing sequence: " + toString(i)); m->mothurOutEndLine();             }
1251                 }
1252
1253                 return 1;
1254         }
1255         catch(exception& e) {
1256                 m->errorOut(e, "ScreenSeqsCommand", "driverMPI");
1257                 exit(1);
1258         }
1259 }
1260 #endif
1261 /**************************************************************************************************/
1262
1263 int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, string filename, set<string>& badSeqNames) {
1264         try {
1265 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1266                 int process = 0;
1267                 int num = 0;
1268
1269                 //loop through and create all the processes you want
1270                 while (process != processors) {
1271                         int pid = fork();
1272
1273                         if (pid > 0) {
1274                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1275                                 process++;
1276                         }else if (pid == 0){
1277                                 num = driver(lines[process], goodFileName + toString(getpid()) + ".temp", badAccnos + toString(getpid()) + ".temp", filename, badSeqNames);
1278
1279                                 //pass numSeqs to parent
1280                                 ofstream out;
1281                                 string tempFile = filename + toString(getpid()) + ".num.temp";
1282                                 m->openOutputFile(tempFile, out);
1283                                 out << num << endl;
1284                                 out.close();
1285
1286                                 exit(0);
1287                         }else {
1288                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1289                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1290                                 exit(0);
1291                         }
1292                 }
1293
1294                 //force parent to wait until all the processes are done
1295                 for (int i=0;i<processors;i++) {
1296                         int temp = processIDS[i];
1297                         wait(&temp);
1298                 }
1299
1300                 for (int i = 0; i < processIDS.size(); i++) {
1301                         ifstream in;
1302                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
1303                         m->openInputFile(tempFile, in);
1304                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1305                         in.close(); m->mothurRemove(tempFile);
1306                 }
1307
1308                 return num;
1309 #endif
1310         }
1311         catch(exception& e) {
1312                 m->errorOut(e, "ScreenSeqsCommand", "createProcesses");
1313                 exit(1);
1314         }
1315 }
1316
1317 //***************************************************************************************************************
1318
1319