classifyseqscommand.cpp

   1 /*
   2  *  classifyseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/2/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "classifyseqscommand.h"
  11
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> ClassifySeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptaxonomy);
  18                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  20                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  21                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  22                 CommandParameter psearch("search", "Multiple", "kmer-blast-suffix-distance", "kmer", "", "", "",false,false); parameters.push_back(psearch);
  23                 CommandParameter pksize("ksize", "Number", "", "8", "", "", "",false,false); parameters.push_back(pksize);
  24                 CommandParameter pmethod("method", "Multiple", "bayesian-knn", "bayesian", "", "", "",false,false); parameters.push_back(pmethod);
  25                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  26                 CommandParameter pmatch("match", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pmatch);
  27                 CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pmismatch);
  28                 CommandParameter pgapopen("gapopen", "Number", "", "-2.0", "", "", "",false,false); parameters.push_back(pgapopen);
  29                 CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pgapextend);
  30                 //CommandParameter pflip("flip", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pflip);
  31                 CommandParameter pcutoff("cutoff", "Number", "", "0", "", "", "",false,true); parameters.push_back(pcutoff);
  32                 CommandParameter pprobs("probs", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pprobs);
  33                 CommandParameter piters("iters", "Number", "", "100", "", "", "",false,true); parameters.push_back(piters);
  34                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  35                 CommandParameter pnumwanted("numwanted", "Number", "", "10", "", "", "",false,true); parameters.push_back(pnumwanted);
  36                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  37                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  38
  39                 vector<string> myArray;
  40                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  41                 return myArray;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "ClassifySeqsCommand", "setParameters");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 string ClassifySeqsCommand::getHelpString(){
  50         try {
  51                 string helpString = "";
  52                 helpString += "The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n";
  53                 helpString += "The classify.seqs command parameters are reference, fasta, name, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n";
  54                 helpString += "The reference, fasta and taxonomy parameters are required. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  55                 helpString += "The search parameter allows you to specify the method to find most similar template.  Your options are: suffix, kmer, blast and distance. The default is kmer.\n";
  56                 helpString += "The name parameter allows you add a names file with your fasta file, if you enter multiple fasta files, you must enter matching names files for them.\n";
  57                 helpString += "The group parameter allows you add a group file so you can have the summary totals broken up by group.\n";
  58                 helpString += "The method parameter allows you to specify classification method to use.  Your options are: bayesian and knn. The default is bayesian.\n";
  59                 helpString += "The ksize parameter allows you to specify the kmer size for finding most similar template to candidate.  The default is 8.\n";
  60                 helpString += "The processors parameter allows you to specify the number of processors to use. The default is 1.\n";
  61 #ifdef USE_MPI
  62                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  63 #endif
  64                 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
  65                 helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n";
  66                 helpString += "The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n";
  67                 helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n";
  68                 helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n";
  69                 helpString += "The numwanted parameter allows you to specify the number of sequence matches you want with the knn method.  The default is 10.\n";
  70                 helpString += "The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy.  The default is 0.\n";
  71                 helpString += "The probs parameter shuts off the bootstrapping results for the bayesian method. The default is true, meaning you want the bootstrapping to be shown.\n";
  72                 helpString += "The iters parameter allows you to specify how many iterations to do when calculating the bootstrap confidence score for your taxonomy with the bayesian method.  The default is 100.\n";
  73                 //helpString += "The flip parameter allows you shut off mothur's   The default is T.\n";
  74                 helpString += "The classify.seqs command should be in the following format: \n";
  75                 helpString += "classify.seqs(reference=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n";
  76                 helpString += "Example classify.seqs(fasta=amazon.fasta, reference=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n";
  77                 helpString += "The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n";
  78                 helpString += "The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n";
  79                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  80                 return helpString;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "ClassifySeqsCommand", "getHelpString");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88 ClassifySeqsCommand::ClassifySeqsCommand(){
  89         try {
  90                 abort = true; calledHelp = true;
  91                 setParameters();
  92                 vector<string> tempOutNames;
  93                 outputTypes["taxonomy"] = tempOutNames;
  94                 outputTypes["accnos"] = tempOutNames;
  95                 outputTypes["taxsummary"] = tempOutNames;
  96                 outputTypes["matchdist"] = tempOutNames;
  97         }
  98         catch(exception& e) {
  99                 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
 100                 exit(1);
 101         }
 102 }
 103 //**********************************************************************************************************************
 104 ClassifySeqsCommand::ClassifySeqsCommand(string option)  {
 105         try {
 106                 abort = false; calledHelp = false;
 107                 rdb = ReferenceDB::getInstance();
 108
 109                 //allow user to run help
 110                 if(option == "help") { help(); abort = true; calledHelp = true; }
 111                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 112
 113                 else {
 114                         vector<string> myArray = setParameters();
 115
 116                         OptionParser parser(option);
 117                         map<string, string> parameters = parser.getParameters();
 118
 119                         ValidParameters validParameter("classify.seqs");
 120                         map<string, string>::iterator it;
 121
 122                         //check to make sure all parameters are valid for command
 123                         for (it = parameters.begin(); it != parameters.end(); it++) {
 124                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 125                         }
 126
 127                         //initialize outputTypes
 128                         vector<string> tempOutNames;
 129                         outputTypes["taxonomy"] = tempOutNames;
 130                         outputTypes["taxsummary"] = tempOutNames;
 131                         outputTypes["matchdist"] = tempOutNames;
 132                         outputTypes["accnos"] = tempOutNames;
 133
 134                         //if the user changes the output directory command factory will send this info to us in the output parameter
 135                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 136
 137                         //if the user changes the input directory command factory will send this info to us in the output parameter
 138                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 139                         if (inputDir == "not found"){   inputDir = "";          }
 140                         else {
 141                                 string path;
 142                                 it = parameters.find("reference");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 148                                 }
 149
 150                                 it = parameters.find("taxonomy");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 156                                 }
 157
 158                                 it = parameters.find("group");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 164                                 }
 165                         }
 166
 167                         fastaFileName = validParameter.validFile(parameters, "fasta", false);
 168                         if (fastaFileName == "not found") {
 169                                 //if there is a current fasta file, use it
 170                                 string filename = m->getFastaFile();
 171                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 172                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 173                         }
 174                         else {
 175                                 m->splitAtDash(fastaFileName, fastaFileNames);
 176
 177                                 //go through files and make sure they are good, if not, then disregard them
 178                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 179
 180                                         bool ignore = false;
 181                                         if (fastaFileNames[i] == "current") {
 182                                                 fastaFileNames[i] = m->getFastaFile();
 183                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 184                                                 else {
 185                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 186                                                         //erase from file list
 187                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 188                                                         i--;
 189                                                 }
 190                                         }
 191
 192                                         if (!ignore) {
 193
 194                                                 if (inputDir != "") {
 195                                                         string path = m->hasPath(fastaFileNames[i]);
 196                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 197                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 198                                                 }
 199
 200                                                 int ableToOpen;
 201
 202                                                 ifstream in;
 203                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 204
 205                                                 //if you can't open it, try default location
 206                                                 if (ableToOpen == 1) {
 207                                                         if (m->getDefaultPath() != "") { //default path is set
 208                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 209                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 210                                                                 ifstream in2;
 211                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 212                                                                 in2.close();
 213                                                                 fastaFileNames[i] = tryPath;
 214                                                         }
 215                                                 }
 216
 217                                                 if (ableToOpen == 1) {
 218                                                         if (m->getOutputDir() != "") { //default path is set
 219                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 220                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 221                                                                 ifstream in2;
 222                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 223                                                                 in2.close();
 224                                                                 fastaFileNames[i] = tryPath;
 225                                                         }
 226                                                 }
 227
 228                                                 in.close();
 229
 230                                                 if (ableToOpen == 1) {
 231                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 232                                                         //erase from file list
 233                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 234                                                         i--;
 235                                                 }else {
 236                                                         m->setFastaFile(fastaFileNames[i]);
 237                                                 }
 238                                         }
 239
 240                                 }
 241
 242                                 //make sure there is at least one valid file left
 243                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 244                         }
 245
 246                         namefile = validParameter.validFile(parameters, "name", false);
 247                         if (namefile == "not found") { namefile = "";  }
 248
 249                         else {
 250                                 m->splitAtDash(namefile, namefileNames);
 251
 252                                 //go through files and make sure they are good, if not, then disregard them
 253                                 for (int i = 0; i < namefileNames.size(); i++) {
 254                                         bool ignore = false;
 255                                         if (namefileNames[i] == "current") {
 256                                                 namefileNames[i] = m->getNameFile();
 257                                                 if (namefileNames[i] != "") {  m->mothurOut("Using " + namefileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 258                                                 else {
 259                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 260                                                         //erase from file list
 261                                                         namefileNames.erase(namefileNames.begin()+i);
 262                                                         i--;
 263                                                 }
 264                                         }
 265
 266                                         if (!ignore) {
 267
 268                                                 if (inputDir != "") {
 269                                                         string path = m->hasPath(namefileNames[i]);
 270                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 271                                                         if (path == "") {       namefileNames[i] = inputDir + namefileNames[i];         }
 272                                                 }
 273                                                 int ableToOpen;
 274
 275                                                 ifstream in;
 276                                                 ableToOpen = m->openInputFile(namefileNames[i], in, "noerror");
 277
 278                                                 //if you can't open it, try default location
 279                                                 if (ableToOpen == 1) {
 280                                                         if (m->getDefaultPath() != "") { //default path is set
 281                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(namefileNames[i]);
 282                                                                 m->mothurOut("Unable to open " + namefileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 283                                                                 ifstream in2;
 284                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 285                                                                 in2.close();
 286                                                                 namefileNames[i] = tryPath;
 287                                                         }
 288                                                 }
 289
 290                                                 if (ableToOpen == 1) {
 291                                                         if (m->getOutputDir() != "") { //default path is set
 292                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(namefileNames[i]);
 293                                                                 m->mothurOut("Unable to open " + namefileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 294                                                                 ifstream in2;
 295                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 296                                                                 in2.close();
 297                                                                 namefileNames[i] = tryPath;
 298                                                         }
 299                                                 }
 300                                                 in.close();
 301
 302                                                 if (ableToOpen == 1) {
 303                                                         m->mothurOut("Unable to open " + namefileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();  abort = true;
 304                                                         //erase from file list
 305                                                         namefileNames.erase(namefileNames.begin()+i);
 306                                                         i--;
 307                                                 }else {
 308                                                         m->setNameFile(namefileNames[i]);
 309                                                 }
 310                                         }
 311                                 }
 312                         }
 313
 314                         if (namefile != "") {
 315                                 if (namefileNames.size() != fastaFileNames.size()) { abort = true; m->mothurOut("If you provide a name file, you must have one for each fasta file."); m->mothurOutEndLine(); }
 316                         }
 317
 318                         groupfile = validParameter.validFile(parameters, "group", false);
 319                         if (groupfile == "not found") { groupfile = "";  }
 320                         else {
 321                                 m->splitAtDash(groupfile, groupfileNames);
 322
 323                                 //go through files and make sure they are good, if not, then disregard them
 324                                 for (int i = 0; i < groupfileNames.size(); i++) {
 325                                         if (inputDir != "") {
 326                                                 string path = m->hasPath(groupfileNames[i]);
 327                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 328                                                 if (path == "") {       groupfileNames[i] = inputDir + groupfileNames[i];               }
 329                                         }
 330                                         int ableToOpen;
 331
 332                                         ifstream in;
 333                                         ableToOpen = m->openInputFile(groupfileNames[i], in, "noerror");
 334
 335                                         //if you can't open it, try default location
 336                                         if (ableToOpen == 1) {
 337                                                 if (m->getDefaultPath() != "") { //default path is set
 338                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(groupfileNames[i]);
 339                                                         m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 340                                                         ifstream in2;
 341                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 342                                                         in2.close();
 343                                                         groupfileNames[i] = tryPath;
 344                                                 }
 345                                         }
 346
 347                                         if (ableToOpen == 1) {
 348                                                 if (m->getOutputDir() != "") { //default path is set
 349                                                         string tryPath = m->getOutputDir() + m->getSimpleName(groupfileNames[i]);
 350                                                         m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 351                                                         ifstream in2;
 352                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 353                                                         in2.close();
 354                                                         groupfileNames[i] = tryPath;
 355                                                 }
 356                                         }
 357
 358                                         in.close();
 359
 360                                         if (ableToOpen == 1) {
 361                                                 m->mothurOut("Unable to open " + groupfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); groupfileNames[i] = "";
 362                                                 //erase from file list
 363                                                 groupfileNames.erase(groupfileNames.begin()+i);
 364                                                 i--;
 365                                         }else {
 366                                                 m->setGroupFile(groupfileNames[i]);
 367                                         }
 368                                 }
 369                         }
 370
 371                         if (groupfile != "") {
 372                                 if (groupfileNames.size() != fastaFileNames.size()) { abort = true; m->mothurOut("If you provide a group file, you must have one for each fasta file."); m->mothurOutEndLine(); }
 373                         }else {
 374                                 for (int i = 0; i < fastaFileNames.size(); i++) {  groupfileNames.push_back("");  }
 375                         }
 376
 377                         //check for optional parameter and set defaults
 378                         // ...at some point should added some additional type checking...
 379                         string temp;
 380                         temp = validParameter.validFile(parameters, "ksize", false);            if (temp == "not found"){       temp = "8";                             }
 381                         m->mothurConvert(temp, kmerSize);
 382
 383                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 384                         m->setProcessors(temp);
 385                         m->mothurConvert(temp, processors);
 386
 387                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 388                         save = m->isTrue(temp);
 389                         rdb->save = save;
 390                         if (save) { //clear out old references
 391                                 rdb->clearMemory();
 392                         }
 393
 394                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 395                         templateFileName = validParameter.validFile(parameters, "reference", true);
 396                         if (templateFileName == "not found") {
 397                                 //check for saved reference sequences
 398                                 if (rdb->referenceSeqs.size() != 0) {
 399                                         templateFileName = "saved";
 400                                 }else {
 401                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the classify.seqs command.");
 402                                         m->mothurOutEndLine();
 403                                         abort = true;
 404                                 }
 405                         }else if (templateFileName == "not open") { abort = true; }
 406                         else {  if (save) {     rdb->setSavedReference(templateFileName);       }       }
 407
 408                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 409                         taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true);
 410                         if (taxonomyFileName == "not found") {
 411                                 //check for saved reference sequences
 412                                 if (rdb->wordGenusProb.size() != 0) {
 413                                         taxonomyFileName = "saved";
 414                                 }else {
 415                                         m->mothurOut("[ERROR]: You don't have any saved taxonomy information and the taxonomy parameter is a required for the classify.seqs command.");
 416                                         m->mothurOutEndLine();
 417                                         abort = true;
 418                                 }
 419                         }else if (taxonomyFileName == "not open") { abort = true; }
 420                         else {  if (save) {     rdb->setSavedTaxonomy(taxonomyFileName);        }       }
 421
 422                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
 423
 424                         method = validParameter.validFile(parameters, "method", false);         if (method == "not found"){     method = "bayesian";    }
 425
 426                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 427                         m->mothurConvert(temp, match);
 428
 429                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 430                         m->mothurConvert(temp, misMatch);
 431
 432                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 433                         m->mothurConvert(temp, gapOpen);
 434
 435                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 436                         m->mothurConvert(temp, gapExtend);
 437
 438                         temp = validParameter.validFile(parameters, "numwanted", false);        if (temp == "not found"){       temp = "10";                    }
 439                         m->mothurConvert(temp, numWanted);
 440
 441                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found"){       temp = "0";                             }
 442                         m->mothurConvert(temp, cutoff);
 443
 444                         temp = validParameter.validFile(parameters, "probs", false);            if (temp == "not found"){       temp = "true";                  }
 445                         probs = m->isTrue(temp);
 446
 447                         //temp = validParameter.validFile(parameters, "flip", false);                   if (temp == "not found"){       temp = "T";                             }
 448                         //flip = m->isTrue(temp);
 449                         flip = true;
 450
 451                         temp = validParameter.validFile(parameters, "iters", false);            if (temp == "not found") { temp = "100";                        }
 452                         m->mothurConvert(temp, iters);
 453
 454
 455                         if ((method == "bayesian") && (search != "kmer"))  {
 456                                 m->mothurOut("The bayesian method requires the kmer search." + search + "will be disregarded." ); m->mothurOutEndLine();
 457                                 search = "kmer";
 458                         }
 459
 460                         if (namefileNames.size() == 0){
 461                                 vector<string> files; files.push_back(fastaFileNames[fastaFileNames.size()-1]);
 462                                 parser.getNameFile(files);
 463                         }
 464
 465                 }
 466
 467         }
 468         catch(exception& e) {
 469                 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
 470                 exit(1);
 471         }
 472 }
 473
 474 //**********************************************************************************************************************
 475 ClassifySeqsCommand::~ClassifySeqsCommand(){
 476         if (abort == false) {
 477                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 478         }
 479 }
 480 //**********************************************************************************************************************
 481
 482 int ClassifySeqsCommand::execute(){
 483         try {
 484                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 485
 486                 if(method == "bayesian"){       classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip);             }
 487                 else if(method == "knn"){       classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted, rand());                               }
 488                 else {
 489                         m->mothurOut(search + " is not a valid method option. I will run the command using bayesian.");
 490                         m->mothurOutEndLine();
 491                         classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip);
 492                 }
 493
 494                 if (m->control_pressed) { delete classify; return 0; }
 495
 496                 for (int s = 0; s < fastaFileNames.size(); s++) {
 497
 498                         m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 499
 500                         string baseTName = taxonomyFileName;
 501                         if (taxonomyFileName == "saved") {baseTName = rdb->getSavedTaxonomy();  }
 502
 503                         string RippedTaxName = m->getRootName(m->getSimpleName(baseTName));
 504                         RippedTaxName = m->getExtension(RippedTaxName.substr(0, RippedTaxName.length()-1));
 505                         if (RippedTaxName[0] == '.') { RippedTaxName = RippedTaxName.substr(1, RippedTaxName.length()); }
 506                         RippedTaxName +=  ".";
 507
 508                         if (outputDir == "") { outputDir += m->hasPath(fastaFileNames[s]); }
 509                         string newTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "taxonomy";
 510                         string newaccnosFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "flip.accnos";
 511                         string tempTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "taxonomy.temp";
 512                         string taxSummary = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "tax.summary";
 513
 514                         if ((method == "knn") && (search == "distance")) {
 515                                 string DistName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "match.dist";
 516                                 classify->setDistName(DistName);  outputNames.push_back(DistName); outputTypes["matchdist"].push_back(DistName);
 517                         }
 518
 519                         outputNames.push_back(newTaxonomyFile); outputTypes["taxonomy"].push_back(newTaxonomyFile);
 520                         outputNames.push_back(newaccnosFile); outputTypes["accnos"].push_back(newaccnosFile);
 521                         outputNames.push_back(taxSummary);      outputTypes["taxsummary"].push_back(taxSummary);
 522
 523                         int start = time(NULL);
 524                         int numFastaSeqs = 0;
 525                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 526
 527 #ifdef USE_MPI
 528                                 int pid, numSeqsPerProcessor;
 529                                 int tag = 2001;
 530                                 vector<unsigned long long> MPIPos;
 531
 532                                 MPI_Status status;
 533                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 534                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 535
 536                                 MPI_File inMPI;
 537                                 MPI_File outMPINewTax;
 538                                 MPI_File outMPITempTax;
 539                                 MPI_File outMPIAcc;
 540
 541                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 542                                 int inMode=MPI_MODE_RDONLY;
 543
 544                                 char outNewTax[1024];
 545                                 strcpy(outNewTax, newTaxonomyFile.c_str());
 546
 547                                 char outTempTax[1024];
 548                                 strcpy(outTempTax, tempTaxonomyFile.c_str());
 549
 550                                 char outAcc[1024];
 551                                 strcpy(outAcc, newaccnosFile.c_str());
 552
 553                                 char inFileName[1024];
 554                                 strcpy(inFileName, fastaFileNames[s].c_str());
 555
 556                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 557                                 MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax);
 558                                 MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax);
 559                                 MPI_File_open(MPI_COMM_WORLD, outAcc, outMode, MPI_INFO_NULL, &outMPIAcc);
 560
 561                                 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax);  MPI_File_close(&outMPIAcc);   MPI_File_close(&outMPITempTax);  delete classify;  return 0;  }
 562
 563                                 if (pid == 0) { //you are the root process
 564
 565                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
 566
 567                                         //send file positions to all processes
 568                                         for(int i = 1; i < processors; i++) {
 569                                                 MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 570                                                 MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 571                                         }
 572
 573                                         //figure out how many sequences you have to align
 574                                         numSeqsPerProcessor = numFastaSeqs / processors;
 575                                         int startIndex =  pid * numSeqsPerProcessor;
 576                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 577
 578
 579                                         //align your part
 580                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos);
 581
 582                                         if (m->control_pressed) {  outputTypes.clear(); MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax); MPI_File_close(&outMPIAcc);   MPI_File_close(&outMPITempTax);  for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);        } delete classify; return 0;  }
 583
 584                                         for (int i = 1; i < processors; i++) {
 585                                                 int done;
 586                                                 MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 587                                         }
 588                                 }else{ //you are a child process
 589                                         MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 590                                         MPIPos.resize(numFastaSeqs+1);
 591                                         MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 592
 593                                         //figure out how many sequences you have to align
 594                                         numSeqsPerProcessor = numFastaSeqs / processors;
 595                                         int startIndex =  pid * numSeqsPerProcessor;
 596                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 597
 598
 599                                         //align your part
 600                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos);
 601
 602                                         if (m->control_pressed) {  outputTypes.clear(); MPI_File_close(&inMPI);  MPI_File_close(&outMPINewTax);  MPI_File_close(&outMPIAcc);  MPI_File_close(&outMPITempTax);  delete classify; return 0;  }
 603
 604                                         int done = 0;
 605                                         MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 606                                 }
 607
 608                                 //close files
 609                                 MPI_File_close(&inMPI);
 610                                 MPI_File_close(&outMPINewTax);
 611                                 MPI_File_close(&outMPITempTax);
 612                                 MPI_File_close(&outMPIAcc);
 613                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 614
 615 #else
 616
 617                         vector<unsigned long long> positions;
 618 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 619                         positions = m->divideFile(fastaFileNames[s], processors);
 620                         for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(new linePair(positions[i], positions[(i+1)]));  }
 621 #else
 622                         if (processors == 1) {
 623                                 lines.push_back(new linePair(0, 1000));
 624                         }else {
 625                                 positions = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs);
 626                 if (positions.size() < processors) { processors = positions.size(); }
 627
 628                                 //figure out how many sequences you have to process
 629                                 int numSeqsPerProcessor = numFastaSeqs / processors;
 630                                 for (int i = 0; i < processors; i++) {
 631                                         int startIndex =  i * numSeqsPerProcessor;
 632                                         if(i == (processors - 1)){      numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;   }
 633                                         lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
 634                                 }
 635                         }
 636 #endif
 637                         if(processors == 1){
 638                                 numFastaSeqs = driver(lines[0], newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]);
 639                         }else{
 640                                 numFastaSeqs = createProcesses(newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]);
 641                         }
 642 #endif
 643
 644                         if (!m->isBlank(newaccnosFile)) { m->mothurOutEndLine(); m->mothurOut("[WARNING]: mothur suspects some of your sequences may be reversed, please check " + newaccnosFile + " for the list of the sequences."); m->mothurOutEndLine(); }
 645
 646                 m->mothurOutEndLine();
 647                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 648                 start = time(NULL);
 649
 650
 651
 652                 #ifdef USE_MPI
 653                         if (pid == 0) {  //this part does not need to be paralellized
 654
 655                                 if(namefile != "") { m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();  MPIReadNamesFile(namefileNames[s]);  m->mothurOut("  Done."); m->mothurOutEndLine(); }
 656                 #else
 657                         //read namefile
 658                         if(namefile != "") {
 659
 660                             m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();
 661
 662                                 nameMap.clear(); //remove old names
 663
 664                                 ifstream inNames;
 665                                 m->openInputFile(namefileNames[s], inNames);
 666
 667                                 string firstCol, secondCol;
 668                                 while(!inNames.eof()) {
 669                                         inNames >> firstCol >> secondCol; m->gobble(inNames);
 670
 671                                         vector<string> temp;
 672                                         m->splitAtComma(secondCol, temp);
 673
 674                                         nameMap[firstCol] = temp;
 675                                 }
 676                                 inNames.close();
 677
 678                                 m->mothurOut("  Done."); m->mothurOutEndLine();
 679                         }
 680                 #endif
 681
 682                         string group = "";
 683                         if (groupfile != "") {  group = groupfileNames[s]; }
 684
 685                         PhyloSummary taxaSum(baseTName, group);
 686
 687                         if (m->control_pressed) { outputTypes.clear();  for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } delete classify; return 0; }
 688
 689                         if (namefile == "") {  taxaSum.summarize(tempTaxonomyFile);  }
 690                         else {
 691                                 ifstream in;
 692                                 m->openInputFile(tempTaxonomyFile, in);
 693
 694                                 //read in users taxonomy file and add sequences to tree
 695                                 string name, taxon;
 696
 697                                 while(!in.eof()){
 698                                         in >> name >> taxon; m->gobble(in);
 699
 700                                         itNames = nameMap.find(name);
 701
 702                                         if (itNames == nameMap.end()) {
 703                                                 m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
 704                                         }else{
 705                                                 for (int i = 0; i < itNames->second.size(); i++) {
 706                                                         taxaSum.addSeqToTree(itNames->second[i], taxon);  //add it as many times as there are identical seqs
 707                                                 }
 708                                                 itNames->second.clear();
 709                                                 nameMap.erase(itNames->first);
 710                                         }
 711                                 }
 712                                 in.close();
 713                         }
 714                         m->mothurRemove(tempTaxonomyFile);
 715
 716                         if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } delete classify; return 0; }
 717
 718                         //print summary file
 719                         ofstream outTaxTree;
 720                         m->openOutputFile(taxSummary, outTaxTree);
 721                         taxaSum.print(outTaxTree);
 722                         outTaxTree.close();
 723
 724                         //output taxonomy with the unclassified bins added
 725                         ifstream inTax;
 726                         m->openInputFile(newTaxonomyFile, inTax);
 727
 728                         ofstream outTax;
 729                         string unclass = newTaxonomyFile + ".unclass.temp";
 730                         m->openOutputFile(unclass, outTax);
 731
 732                         //get maxLevel from phylotree so you know how many 'unclassified's to add
 733                         int maxLevel = taxaSum.getMaxLevel();
 734
 735                         //read taxfile - this reading and rewriting is done to preserve the confidence scores.
 736                         string name, taxon;
 737                         while (!inTax.eof()) {
 738                                 if (m->control_pressed) { outputTypes.clear();  for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } m->mothurRemove(unclass); delete classify; return 0; }
 739
 740                                 inTax >> name >> taxon; m->gobble(inTax);
 741
 742                                 string newTax = addUnclassifieds(taxon, maxLevel);
 743
 744                                 outTax << name << '\t' << newTax << endl;
 745                         }
 746                         inTax.close();
 747                         outTax.close();
 748
 749                         m->mothurRemove(newTaxonomyFile);
 750                         rename(unclass.c_str(), newTaxonomyFile.c_str());
 751
 752                         m->mothurOutEndLine();
 753                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 754
 755                         #ifdef USE_MPI
 756                                 }
 757                         #endif
 758
 759                         m->mothurOutEndLine();
 760                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 761                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 762                         m->mothurOutEndLine();
 763                 }
 764
 765                 //set taxonomy file as new current taxonomyfile
 766                 string current = "";
 767                 itTypes = outputTypes.find("taxonomy");
 768                 if (itTypes != outputTypes.end()) {
 769                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 770                 }
 771
 772                 current = "";
 773                 itTypes = outputTypes.find("accnos");
 774                 if (itTypes != outputTypes.end()) {
 775                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 776                 }
 777
 778                 delete classify;
 779
 780                 return 0;
 781         }
 782         catch(exception& e) {
 783                 m->errorOut(e, "ClassifySeqsCommand", "execute");
 784                 exit(1);
 785         }
 786 }
 787
 788 /**************************************************************************************************/
 789 string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) {
 790         try{
 791                 string newTax, taxon;
 792                 int level = 0;
 793
 794                 //keep what you have counting the levels
 795                 while (tax.find_first_of(';') != -1) {
 796                         //get taxon
 797                         taxon = tax.substr(0,tax.find_first_of(';'))+';';
 798                         tax = tax.substr(tax.find_first_of(';')+1, tax.length());
 799                         newTax += taxon;
 800                         level++;
 801                 }
 802
 803                 //add "unclassified" until you reach maxLevel
 804                 while (level < maxlevel) {
 805                         newTax += "unclassified;";
 806                         level++;
 807                 }
 808
 809                 return newTax;
 810         }
 811         catch(exception& e) {
 812                 m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds");
 813                 exit(1);
 814         }
 815 }
 816
 817 /**************************************************************************************************/
 818
 819 int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string accnos, string filename) {
 820         try {
 821
 822                 int num = 0;
 823                 processIDS.clear();
 824
 825 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 826                 int process = 1;
 827
 828                 //loop through and create all the processes you want
 829                 while (process != processors) {
 830                         int pid = fork();
 831
 832                         if (pid > 0) {
 833                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 834                                 process++;
 835                         }else if (pid == 0){
 836                                 num = driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", filename);
 837
 838                                 //pass numSeqs to parent
 839                                 ofstream out;
 840                                 string tempFile = filename + toString(getpid()) + ".num.temp";
 841                                 m->openOutputFile(tempFile, out);
 842                                 out << num << endl;
 843                                 out.close();
 844
 845                                 exit(0);
 846                         }else {
 847                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 848                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 849                                 exit(0);
 850                         }
 851                 }
 852
 853                 //parent does its part
 854                 num = driver(lines[0], taxFileName, tempTaxFile, accnos, filename);
 855
 856                 //force parent to wait until all the processes are done
 857                 for (int i=0;i<processIDS.size();i++) {
 858                         int temp = processIDS[i];
 859                         wait(&temp);
 860                 }
 861
 862                 for (int i = 0; i < processIDS.size(); i++) {
 863                         ifstream in;
 864                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
 865                         m->openInputFile(tempFile, in);
 866                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 867                         in.close(); m->mothurRemove(m->getFullPathName(tempFile));
 868                 }
 869 #else
 870                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 871                 //Windows version shared memory, so be careful when passing variables through the alignData struct.
 872                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 873                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 874
 875                 vector<classifyData*> pDataArray;
 876                 DWORD   dwThreadIdArray[processors-1];
 877                 HANDLE  hThreadArray[processors-1];
 878
 879                 //Create processor worker threads.
 880                 for( int i=0; i<processors-1; i++ ){
 881                         // Allocate memory for thread data.
 882                         string extension = "";
 883                         if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
 884
 885                         classifyData* tempclass = new classifyData((accnos + extension), probs, method, templateFileName, taxonomyFileName, (taxFileName + extension), (tempTaxFile + extension), filename, search, kmerSize, iters, numWanted, m, lines[i]->start, lines[i]->end, match, misMatch, gapOpen, gapExtend, cutoff, i, flip);
 886                         pDataArray.push_back(tempclass);
 887
 888                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
 889                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 890                         hThreadArray[i] = CreateThread(NULL, 0, MyClassThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 891
 892                 }
 893
 894                 //parent does its part
 895                 num = driver(lines[processors-1], taxFileName + toString(processors-1) + ".temp", tempTaxFile + toString(processors-1) + ".temp", accnos + toString(processors-1) + ".temp", filename);
 896                 processIDS.push_back((processors-1));
 897
 898                 //Wait until all threads have terminated.
 899                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 900
 901                 //Close all thread handles and free memory allocations.
 902                 for(int i=0; i < pDataArray.size(); i++){
 903                         num += pDataArray[i]->count;
 904                         CloseHandle(hThreadArray[i]);
 905                         delete pDataArray[i];
 906                 }
 907
 908         #endif
 909         vector<string> nonBlankAccnosFiles;
 910                 if (!(m->isBlank(accnos))) { nonBlankAccnosFiles.push_back(accnos); }
 911                 else { m->mothurRemove(accnos); } //remove so other files can be renamed to it
 912
 913                 for(int i=0;i<processIDS.size();i++){
 914                         appendTaxFiles((taxFileName + toString(processIDS[i]) + ".temp"), taxFileName);
 915                         appendTaxFiles((tempTaxFile + toString(processIDS[i]) + ".temp"), tempTaxFile);
 916             if (!(m->isBlank(accnos + toString(processIDS[i]) + ".temp"))) {
 917                                 nonBlankAccnosFiles.push_back(accnos + toString(processIDS[i]) + ".temp");
 918                         }else { m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));  }
 919
 920                         m->mothurRemove((m->getFullPathName(taxFileName) + toString(processIDS[i]) + ".temp"));
 921                         m->mothurRemove((m->getFullPathName(tempTaxFile) + toString(processIDS[i]) + ".temp"));
 922                 }
 923
 924         //append accnos files
 925                 if (nonBlankAccnosFiles.size() != 0) {
 926                         rename(nonBlankAccnosFiles[0].c_str(), accnos.c_str());
 927
 928                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
 929                                 appendTaxFiles(nonBlankAccnosFiles[h], accnos);
 930                                 m->mothurRemove(nonBlankAccnosFiles[h]);
 931                         }
 932                 }else { //recreate the accnosfile if needed
 933                         ofstream out;
 934                         m->openOutputFile(accnos, out);
 935                         out.close();
 936                 }
 937
 938                 return num;
 939
 940         }
 941         catch(exception& e) {
 942                 m->errorOut(e, "ClassifySeqsCommand", "createProcesses");
 943                 exit(1);
 944         }
 945 }
 946 /**************************************************************************************************/
 947
 948 void ClassifySeqsCommand::appendTaxFiles(string temp, string filename) {
 949         try{
 950
 951                 ofstream output;
 952                 ifstream input;
 953                 m->openOutputFileAppend(filename, output);
 954                 m->openInputFile(temp, input);
 955
 956                 while(char c = input.get()){
 957                         if(input.eof())         {       break;                  }
 958                         else                            {       output << c;    }
 959                 }
 960
 961                 input.close();
 962                 output.close();
 963         }
 964         catch(exception& e) {
 965                 m->errorOut(e, "ClassifySeqsCommand", "appendTaxFiles");
 966                 exit(1);
 967         }
 968 }
 969
 970 //**********************************************************************************************************************
 971
 972 int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempTFName, string accnos, string filename){
 973         try {
 974                 ofstream outTax;
 975                 m->openOutputFile(taxFName, outTax);
 976
 977                 ofstream outTaxSimple;
 978                 m->openOutputFile(tempTFName, outTaxSimple);
 979
 980                 ofstream outAcc;
 981                 m->openOutputFile(accnos, outAcc);
 982
 983                 ifstream inFASTA;
 984                 m->openInputFile(filename, inFASTA);
 985
 986                 string taxonomy;
 987
 988                 inFASTA.seekg(filePos->start);
 989
 990                 bool done = false;
 991                 int count = 0;
 992
 993                 while (!done) {
 994                         if (m->control_pressed) {
 995                                 inFASTA.close();
 996                                 outTax.close();
 997                                 outTaxSimple.close();
 998                                 outAcc.close(); return 0; }
 999
1000                         Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
1001
1002                         if (candidateSeq->getName() != "") {
1003
1004                                 taxonomy = classify->getTaxonomy(candidateSeq);
1005
1006                                 if (m->control_pressed) { delete candidateSeq; return 0; }
1007
1008                                 if (taxonomy == "unknown;") { m->mothurOut("[WARNING]: " + candidateSeq->getName() + " could not be classified. You can use the remove.lineage command with taxon=unknown; to remove such sequences."); m->mothurOutEndLine(); }
1009
1010                                 //output confidence scores or not
1011                                 if (probs) {
1012                                         outTax << candidateSeq->getName() << '\t' << taxonomy << endl;
1013                                 }else{
1014                                         outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
1015                                 }
1016
1017                                 if (classify->getFlipped()) { outAcc << candidateSeq->getName() << endl; }
1018
1019                                 outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
1020
1021                                 count++;
1022                         }
1023                         delete candidateSeq;
1024
1025                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1026                                 unsigned long long pos = inFASTA.tellg();
1027                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
1028                         #else
1029                                 if (inFASTA.eof()) { break; }
1030                         #endif
1031
1032                         //report progress
1033                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1034
1035                 }
1036                 //report progress
1037                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1038
1039                 inFASTA.close();
1040                 outTax.close();
1041                 outTaxSimple.close();
1042                 outAcc.close();
1043
1044                 return count;
1045         }
1046         catch(exception& e) {
1047                 m->errorOut(e, "ClassifySeqsCommand", "driver");
1048                 exit(1);
1049         }
1050 }
1051 //**********************************************************************************************************************
1052 #ifdef USE_MPI
1053 int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, MPI_File& accFile, vector<unsigned long long>& MPIPos){
1054         try {
1055                 MPI_Status statusNew;
1056                 MPI_Status statusTemp;
1057                 MPI_Status statusAcc;
1058                 MPI_Status status;
1059
1060                 int pid;
1061                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
1062
1063                 string taxonomy;
1064                 string outputString;
1065
1066                 for(int i=0;i<num;i++){
1067
1068                         if (m->control_pressed) { return 0; }
1069
1070                         //read next sequence
1071                         int length = MPIPos[start+i+1] - MPIPos[start+i];
1072                         char* buf4 = new char[length];
1073                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
1074
1075                         string tempBuf = buf4;
1076                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
1077                         istringstream iss (tempBuf,istringstream::in);
1078                         delete buf4;
1079
1080                         Sequence* candidateSeq = new Sequence(iss);
1081
1082                         if (candidateSeq->getName() != "") {
1083                                 taxonomy = classify->getTaxonomy(candidateSeq);
1084
1085                                 if (taxonomy == "unknown;") { m->mothurOut("[WARNING]: " + candidateSeq->getName() + " could not be classified. You can use the remove.lineage command with taxon=unknown; to remove such sequences."); m->mothurOutEndLine(); }
1086
1087                                 //output confidence scores or not
1088                                 if (probs) {
1089                                         outputString =  candidateSeq->getName() + "\t" + taxonomy + "\n";
1090                                 }else{
1091                                         outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
1092                                 }
1093
1094                                 int length = outputString.length();
1095                                 char* buf2 = new char[length];
1096                                 memcpy(buf2, outputString.c_str(), length);
1097
1098                                 MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew);
1099                                 delete buf2;
1100
1101                                 outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
1102                                 length = outputString.length();
1103                                 char* buf = new char[length];
1104                                 memcpy(buf, outputString.c_str(), length);
1105
1106                                 MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp);
1107                                 delete buf;
1108
1109                                 if (classify->getFlipped()) {
1110                                         outputString =  candidateSeq->getName() + "\n";
1111                                         length = outputString.length();
1112                                         char* buf3 = new char[length];
1113                                         memcpy(buf3, outputString.c_str(), length);
1114
1115                                         MPI_File_write_shared(accFile, buf3, length, MPI_CHAR, &statusAcc);
1116                                         delete buf3;
1117                                 }
1118
1119                         }
1120                         delete candidateSeq;
1121
1122                         if((i+1) % 100 == 0){   cout << "Classifying sequence " << (i+1) << endl;       }
1123                 }
1124
1125                 if(num % 100 != 0){     cout << "Classifying sequence " << (num) << endl;       }
1126
1127
1128                 return 1;
1129         }
1130         catch(exception& e) {
1131                 m->errorOut(e, "ClassifySeqsCommand", "driverMPI");
1132                 exit(1);
1133         }
1134 }
1135
1136 //**********************************************************************************************************************
1137 int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){
1138         try {
1139
1140                 nameMap.clear(); //remove old names
1141
1142                 MPI_File inMPI;
1143                 MPI_Offset size;
1144                 MPI_Status status;
1145
1146                 //char* inFileName = new char[nameFilename.length()];
1147                 //memcpy(inFileName, nameFilename.c_str(), nameFilename.length());
1148
1149                 char inFileName[1024];
1150                 strcpy(inFileName, nameFilename.c_str());
1151
1152                 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
1153                 MPI_File_get_size(inMPI, &size);
1154                 //delete inFileName;
1155
1156                 char* buffer = new char[size];
1157                 MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
1158
1159                 string tempBuf = buffer;
1160                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
1161                 istringstream iss (tempBuf,istringstream::in);
1162                 delete buffer;
1163
1164                 string firstCol, secondCol;
1165                 while(!iss.eof()) {
1166                         iss >> firstCol >> secondCol; m->gobble(iss);
1167
1168                         vector<string> temp;
1169                         m->splitAtComma(secondCol, temp);
1170
1171                         nameMap[firstCol] = temp;
1172                 }
1173
1174                 MPI_File_close(&inMPI);
1175
1176                 return 1;
1177         }
1178         catch(exception& e) {
1179                 m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile");
1180                 exit(1);
1181         }
1182 }
1183 #endif
1184 /**************************************************************************************************/