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", "", "", "NameCount", "none", "none",false,false); parameters.push_back(pname);
  21         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "none", "none",false,false); parameters.push_back(pcount);
  22                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none",false,false); parameters.push_back(pgroup);
  23
  24                 CommandParameter psearch("search", "Multiple", "kmer-blast-suffix-distance-align", "kmer", "", "", "",false,false); parameters.push_back(psearch);
  25                 CommandParameter pksize("ksize", "Number", "", "8", "", "", "",false,false); parameters.push_back(pksize);
  26                 CommandParameter pmethod("method", "Multiple", "wang-knn-zap", "wang", "", "", "",false,false); parameters.push_back(pmethod);
  27                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  28                 CommandParameter pmatch("match", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pmatch);
  29                 CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pmismatch);
  30                 CommandParameter pgapopen("gapopen", "Number", "", "-2.0", "", "", "",false,false); parameters.push_back(pgapopen);
  31                 CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pgapextend);
  32                 //CommandParameter pflip("flip", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pflip);
  33                 CommandParameter pcutoff("cutoff", "Number", "", "0", "", "", "",false,true); parameters.push_back(pcutoff);
  34                 CommandParameter pprobs("probs", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pprobs);
  35                 CommandParameter piters("iters", "Number", "", "100", "", "", "",false,true); parameters.push_back(piters);
  36                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  37         CommandParameter pshortcuts("shortcuts", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pshortcuts);
  38                 CommandParameter pnumwanted("numwanted", "Number", "", "10", "", "", "",false,true); parameters.push_back(pnumwanted);
  39                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  40                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  41
  42                 vector<string> myArray;
  43                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "ClassifySeqsCommand", "setParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 string ClassifySeqsCommand::getHelpString(){
  53         try {
  54                 string helpString = "";
  55                 helpString += "The classify.seqs command reads a fasta file containing sequences and creates a .taxonomy file and a .tax.summary file.\n";
  56                 helpString += "The classify.seqs command parameters are reference, fasta, name, group, count, search, ksize, method, taxonomy, processors, match, mismatch, gapopen, gapextend, numwanted and probs.\n";
  57                 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";
  58                 helpString += "The search parameter allows you to specify the method to find most similar template.  Your options are: suffix, kmer, blast, align and distance. The default is kmer.\n";
  59                 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";
  60                 helpString += "The group parameter allows you add a group file so you can have the summary totals broken up by group.\n";
  61         helpString += "The count parameter allows you add a count file so you can have the summary totals broken up by group.\n";
  62                 helpString += "The method parameter allows you to specify classification method to use.  Your options are: wang, knn and zap. The default is wang.\n";
  63                 helpString += "The ksize parameter allows you to specify the kmer size for finding most similar template to candidate.  The default is 8.\n";
  64                 helpString += "The processors parameter allows you to specify the number of processors to use. The default is 1.\n";
  65 #ifdef USE_MPI
  66                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  67 #endif
  68                 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.";
  69                 helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n";
  70                 helpString += "The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n";
  71                 helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n";
  72                 helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n";
  73                 helpString += "The numwanted parameter allows you to specify the number of sequence matches you want with the knn method.  The default is 10.\n";
  74                 helpString += "The cutoff parameter allows you to specify a bootstrap confidence threshold for your taxonomy.  The default is 0.\n";
  75                 helpString += "The probs parameter shuts off the bootstrapping results for the wang and zap method. The default is true, meaning you want the bootstrapping to be shown.\n";
  76                 helpString += "The iters parameter allows you to specify how many iterations to do when calculating the bootstrap confidence score for your taxonomy with the wang method.  The default is 100.\n";
  77                 //helpString += "The flip parameter allows you shut off mothur's   The default is T.\n";
  78                 helpString += "The classify.seqs command should be in the following format: \n";
  79                 helpString += "classify.seqs(reference=yourTemplateFile, fasta=yourFastaFile, method=yourClassificationMethod, search=yourSearchmethod, ksize=yourKmerSize, taxonomy=yourTaxonomyFile, processors=yourProcessors) \n";
  80                 helpString += "Example classify.seqs(fasta=amazon.fasta, reference=core.filtered, method=knn, search=gotoh, ksize=8, processors=2)\n";
  81                 helpString += "The .taxonomy file consists of 2 columns: 1 = your sequence name, 2 = the taxonomy for your sequence. \n";
  82                 helpString += "The .tax.summary is a summary of the different taxonomies represented in your fasta file. \n";
  83                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  84                 return helpString;
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "ClassifySeqsCommand", "getHelpString");
  88                 exit(1);
  89         }
  90 }
  91 //**********************************************************************************************************************
  92 string ClassifySeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  93         try {
  94         string outputFileName = "";
  95                 map<string, vector<string> >::iterator it;
  96
  97         //is this a type this command creates
  98         it = outputTypes.find(type);
  99         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
 100         else {
 101             if (type == "taxonomy") {  outputFileName =  "taxonomy"; }
 102             else if (type == "accnos") {  outputFileName =  "flip.accnos"; }
 103             else if (type == "taxsummary") {  outputFileName =  "tax.summary"; }
 104             else if (type == "matchdist") {  outputFileName =  "match.dist"; }
 105             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
 106         }
 107         return outputFileName;
 108         }
 109         catch(exception& e) {
 110                 m->errorOut(e, "ClassifySeqsCommand", "getOutputFileNameTag");
 111                 exit(1);
 112         }
 113 }
 114 //**********************************************************************************************************************
 115 ClassifySeqsCommand::ClassifySeqsCommand(){
 116         try {
 117                 abort = true; calledHelp = true;
 118                 setParameters();
 119                 vector<string> tempOutNames;
 120                 outputTypes["taxonomy"] = tempOutNames;
 121                 outputTypes["accnos"] = tempOutNames;
 122                 outputTypes["taxsummary"] = tempOutNames;
 123                 outputTypes["matchdist"] = tempOutNames;
 124         }
 125         catch(exception& e) {
 126                 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
 127                 exit(1);
 128         }
 129 }
 130 //**********************************************************************************************************************
 131 ClassifySeqsCommand::ClassifySeqsCommand(string option)  {
 132         try {
 133                 abort = false; calledHelp = false;
 134                 rdb = ReferenceDB::getInstance(); hasName = false; hasCount=false;
 135
 136                 //allow user to run help
 137                 if(option == "help") { help(); abort = true; calledHelp = true; }
 138                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 139
 140                 else {
 141                         vector<string> myArray = setParameters();
 142
 143                         OptionParser parser(option);
 144                         map<string, string> parameters = parser.getParameters();
 145
 146                         ValidParameters validParameter("classify.seqs");
 147                         map<string, string>::iterator it;
 148
 149                         //check to make sure all parameters are valid for command
 150                         for (it = parameters.begin(); it != parameters.end(); it++) {
 151                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 152                         }
 153
 154                         //initialize outputTypes
 155                         vector<string> tempOutNames;
 156                         outputTypes["taxonomy"] = tempOutNames;
 157                         outputTypes["taxsummary"] = tempOutNames;
 158                         outputTypes["matchdist"] = tempOutNames;
 159                         outputTypes["accnos"] = tempOutNames;
 160
 161                         //if the user changes the output directory command factory will send this info to us in the output parameter
 162                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 163
 164                         //if the user changes the input directory command factory will send this info to us in the output parameter
 165                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 166                         if (inputDir == "not found"){   inputDir = "";          }
 167                         else {
 168                                 string path;
 169                                 it = parameters.find("reference");
 170                                 //user has given a template file
 171                                 if(it != parameters.end()){
 172                                         path = m->hasPath(it->second);
 173                                         //if the user has not given a path then, add inputdir. else leave path alone.
 174                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 175                                 }
 176
 177                                 it = parameters.find("taxonomy");
 178                                 //user has given a template file
 179                                 if(it != parameters.end()){
 180                                         path = m->hasPath(it->second);
 181                                         //if the user has not given a path then, add inputdir. else leave path alone.
 182                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 183                                 }
 184
 185                                 it = parameters.find("group");
 186                                 //user has given a template file
 187                                 if(it != parameters.end()){
 188                                         path = m->hasPath(it->second);
 189                                         //if the user has not given a path then, add inputdir. else leave path alone.
 190                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 191                                 }
 192
 193                 it = parameters.find("count");
 194                                 //user has given a template file
 195                                 if(it != parameters.end()){
 196                                         path = m->hasPath(it->second);
 197                                         //if the user has not given a path then, add inputdir. else leave path alone.
 198                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 199                                 }
 200                         }
 201
 202                         fastaFileName = validParameter.validFile(parameters, "fasta", false);
 203                         if (fastaFileName == "not found") {
 204                                 //if there is a current fasta file, use it
 205                                 string filename = m->getFastaFile();
 206                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 207                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 208                         }
 209                         else {
 210                                 m->splitAtDash(fastaFileName, fastaFileNames);
 211
 212                                 //go through files and make sure they are good, if not, then disregard them
 213                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 214
 215                                         bool ignore = false;
 216                                         if (fastaFileNames[i] == "current") {
 217                                                 fastaFileNames[i] = m->getFastaFile();
 218                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 219                                                 else {
 220                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 221                                                         //erase from file list
 222                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 223                                                         i--;
 224                                                 }
 225                                         }
 226
 227                                         if (!ignore) {
 228
 229                                                 if (inputDir != "") {
 230                                                         string path = m->hasPath(fastaFileNames[i]);
 231                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 232                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 233                                                 }
 234
 235                                                 int ableToOpen;
 236
 237                                                 ifstream in;
 238                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 239
 240                                                 //if you can't open it, try default location
 241                                                 if (ableToOpen == 1) {
 242                                                         if (m->getDefaultPath() != "") { //default path is set
 243                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 244                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 245                                                                 ifstream in2;
 246                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 247                                                                 in2.close();
 248                                                                 fastaFileNames[i] = tryPath;
 249                                                         }
 250                                                 }
 251
 252                                                 if (ableToOpen == 1) {
 253                                                         if (m->getOutputDir() != "") { //default path is set
 254                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 255                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 256                                                                 ifstream in2;
 257                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 258                                                                 in2.close();
 259                                                                 fastaFileNames[i] = tryPath;
 260                                                         }
 261                                                 }
 262
 263                                                 in.close();
 264
 265                                                 if (ableToOpen == 1) {
 266                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 267                                                         //erase from file list
 268                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 269                                                         i--;
 270                                                 }else {
 271                                                         m->setFastaFile(fastaFileNames[i]);
 272                                                 }
 273                                         }
 274
 275                                 }
 276
 277                                 //make sure there is at least one valid file left
 278                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 279                         }
 280
 281                         namefile = validParameter.validFile(parameters, "name", false);
 282                         if (namefile == "not found") { namefile = "";  }
 283
 284                         else {
 285                                 m->splitAtDash(namefile, namefileNames);
 286
 287                                 //go through files and make sure they are good, if not, then disregard them
 288                                 for (int i = 0; i < namefileNames.size(); i++) {
 289                                         bool ignore = false;
 290                                         if (namefileNames[i] == "current") {
 291                                                 namefileNames[i] = m->getNameFile();
 292                                                 if (namefileNames[i] != "") {  m->mothurOut("Using " + namefileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 293                                                 else {
 294                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 295                                                         //erase from file list
 296                                                         namefileNames.erase(namefileNames.begin()+i);
 297                                                         i--;
 298                                                 }
 299                                         }
 300
 301                                         if (!ignore) {
 302
 303                                                 if (inputDir != "") {
 304                                                         string path = m->hasPath(namefileNames[i]);
 305                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 306                                                         if (path == "") {       namefileNames[i] = inputDir + namefileNames[i];         }
 307                                                 }
 308                                                 int ableToOpen;
 309
 310                                                 ifstream in;
 311                                                 ableToOpen = m->openInputFile(namefileNames[i], in, "noerror");
 312
 313                                                 //if you can't open it, try default location
 314                                                 if (ableToOpen == 1) {
 315                                                         if (m->getDefaultPath() != "") { //default path is set
 316                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(namefileNames[i]);
 317                                                                 m->mothurOut("Unable to open " + namefileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 318                                                                 ifstream in2;
 319                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 320                                                                 in2.close();
 321                                                                 namefileNames[i] = tryPath;
 322                                                         }
 323                                                 }
 324
 325                                                 if (ableToOpen == 1) {
 326                                                         if (m->getOutputDir() != "") { //default path is set
 327                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(namefileNames[i]);
 328                                                                 m->mothurOut("Unable to open " + namefileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 329                                                                 ifstream in2;
 330                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 331                                                                 in2.close();
 332                                                                 namefileNames[i] = tryPath;
 333                                                         }
 334                                                 }
 335                                                 in.close();
 336
 337                                                 if (ableToOpen == 1) {
 338                                                         m->mothurOut("Unable to open " + namefileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();  abort = true;
 339                                                         //erase from file list
 340                                                         namefileNames.erase(namefileNames.begin()+i);
 341                                                         i--;
 342                                                 }else {
 343                                                         m->setNameFile(namefileNames[i]);
 344                                                 }
 345                                         }
 346                                 }
 347                         }
 348
 349             if (namefileNames.size() != 0) { hasName = true; }
 350
 351                         if (namefile != "") {
 352                                 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(); }
 353                         }
 354
 355             //check for required parameters
 356                         countfile = validParameter.validFile(parameters, "count", false);
 357                         if (countfile == "not found") {
 358                 countfile = "";
 359                         }else {
 360                                 m->splitAtDash(countfile, countfileNames);
 361
 362                                 //go through files and make sure they are good, if not, then disregard them
 363                                 for (int i = 0; i < countfileNames.size(); i++) {
 364
 365                                         bool ignore = false;
 366                                         if (countfileNames[i] == "current") {
 367                                                 countfileNames[i] = m->getCountTableFile();
 368                                                 if (countfileNames[i] != "") {  m->mothurOut("Using " + countfileNames[i] + " as input file for the count parameter where you had given current."); m->mothurOutEndLine(); }
 369                                                 else {
 370                                                         m->mothurOut("You have no current count file, ignoring current."); m->mothurOutEndLine(); ignore=true;
 371                                                         //erase from file list
 372                                                         countfileNames.erase(countfileNames.begin()+i);
 373                                                         i--;
 374                                                 }
 375                                         }
 376
 377                                         if (!ignore) {
 378
 379                                                 if (inputDir != "") {
 380                                                         string path = m->hasPath(countfileNames[i]);
 381                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 382                                                         if (path == "") {       countfileNames[i] = inputDir + countfileNames[i];               }
 383                                                 }
 384
 385                                                 int ableToOpen;
 386                                                 ifstream in;
 387
 388                                                 ableToOpen = m->openInputFile(countfileNames[i], in, "noerror");
 389
 390                                                 //if you can't open it, try default location
 391                                                 if (ableToOpen == 1) {
 392                                                         if (m->getDefaultPath() != "") { //default path is set
 393                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(countfileNames[i]);
 394                                                                 m->mothurOut("Unable to open " + countfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 395                                                                 ifstream in2;
 396                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 397                                                                 in2.close();
 398                                                                 countfileNames[i] = tryPath;
 399                                                         }
 400                                                 }
 401
 402                                                 if (ableToOpen == 1) {
 403                                                         if (m->getOutputDir() != "") { //default path is set
 404                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(countfileNames[i]);
 405                                                                 m->mothurOut("Unable to open " + countfileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 406                                                                 ifstream in2;
 407                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 408                                                                 in2.close();
 409                                                                 countfileNames[i] = tryPath;
 410                                                         }
 411                                                 }
 412
 413                                                 in.close();
 414
 415                                                 if (ableToOpen == 1) {
 416                                                         m->mothurOut("Unable to open " + countfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 417                                                         //erase from file list
 418                                                         countfileNames.erase(countfileNames.begin()+i);
 419                                                         i--;
 420                                                 }else {
 421                                                         m->setCountTableFile(countfileNames[i]);
 422                                                 }
 423                                         }
 424                                 }
 425                         }
 426
 427             if (countfileNames.size() != 0) { hasCount = true; if (countfileNames.size() != fastaFileNames.size()) {m->mothurOut("If you provide a count file, you must have one for each fasta file."); m->mothurOutEndLine(); } }
 428
 429                         //make sure there is at least one valid file left
 430             if (hasName && hasCount) { m->mothurOut("[ERROR]: You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 431
 432                         groupfile = validParameter.validFile(parameters, "group", false);
 433                         if (groupfile == "not found") { groupfile = "";  }
 434                         else {
 435                                 m->splitAtDash(groupfile, groupfileNames);
 436
 437                                 //go through files and make sure they are good, if not, then disregard them
 438                                 for (int i = 0; i < groupfileNames.size(); i++) {
 439                                         if (inputDir != "") {
 440                                                 string path = m->hasPath(groupfileNames[i]);
 441                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 442                                                 if (path == "") {       groupfileNames[i] = inputDir + groupfileNames[i];               }
 443                                         }
 444                                         int ableToOpen;
 445
 446                                         ifstream in;
 447                                         ableToOpen = m->openInputFile(groupfileNames[i], in, "noerror");
 448
 449                                         //if you can't open it, try default location
 450                                         if (ableToOpen == 1) {
 451                                                 if (m->getDefaultPath() != "") { //default path is set
 452                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(groupfileNames[i]);
 453                                                         m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 454                                                         ifstream in2;
 455                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 456                                                         in2.close();
 457                                                         groupfileNames[i] = tryPath;
 458                                                 }
 459                                         }
 460
 461                                         if (ableToOpen == 1) {
 462                                                 if (m->getOutputDir() != "") { //default path is set
 463                                                         string tryPath = m->getOutputDir() + m->getSimpleName(groupfileNames[i]);
 464                                                         m->mothurOut("Unable to open " + groupfileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 465                                                         ifstream in2;
 466                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 467                                                         in2.close();
 468                                                         groupfileNames[i] = tryPath;
 469                                                 }
 470                                         }
 471
 472                                         in.close();
 473
 474                                         if (ableToOpen == 1) {
 475                                                 m->mothurOut("Unable to open " + groupfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine(); groupfileNames[i] = "";
 476                                                 //erase from file list
 477                                                 groupfileNames.erase(groupfileNames.begin()+i);
 478                                                 i--;
 479                                         }else {
 480                                                 m->setGroupFile(groupfileNames[i]);
 481                                         }
 482                                 }
 483                         }
 484
 485                         if (groupfile != "") {
 486                                 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(); }
 487                 if (hasCount) { m->mothurOut("[ERROR]: You must enter ONLY ONE of the following: count or group."); m->mothurOutEndLine(); abort = true; }
 488                         }else {
 489                                 for (int i = 0; i < fastaFileNames.size(); i++) {  groupfileNames.push_back("");  }
 490                         }
 491
 492                         //check for optional parameter and set defaults
 493                         // ...at some point should added some additional type checking...
 494                         string temp;
 495                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 496                         m->setProcessors(temp);
 497                         m->mothurConvert(temp, processors);
 498
 499                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 500                         save = m->isTrue(temp);
 501                         rdb->save = save;
 502                         if (save) { //clear out old references
 503                                 rdb->clearMemory();
 504                         }
 505
 506                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 507                         templateFileName = validParameter.validFile(parameters, "reference", true);
 508                         if (templateFileName == "not found") {
 509                                 //check for saved reference sequences
 510                                 if (rdb->referenceSeqs.size() != 0) {
 511                                         templateFileName = "saved";
 512                                 }else {
 513                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the classify.seqs command.");
 514                                         m->mothurOutEndLine();
 515                                         abort = true;
 516                                 }
 517                         }else if (templateFileName == "not open") { abort = true; }
 518                         else {  if (save) {     rdb->setSavedReference(templateFileName);       }       }
 519
 520                         //this has to go after save so that if the user sets save=t and provides no reference we abort
 521                         taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true);
 522                         if (taxonomyFileName == "not found") {
 523                                 //check for saved reference sequences
 524                                 if (rdb->wordGenusProb.size() != 0) {
 525                                         taxonomyFileName = "saved";
 526                                 }else {
 527                                         m->mothurOut("[ERROR]: You don't have any saved taxonomy information and the taxonomy parameter is a required for the classify.seqs command.");
 528                                         m->mothurOutEndLine();
 529                                         abort = true;
 530                                 }
 531                         }else if (taxonomyFileName == "not open") { abort = true; }
 532                         else {  if (save) {     rdb->setSavedTaxonomy(taxonomyFileName);        }       }
 533
 534                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
 535
 536                         method = validParameter.validFile(parameters, "method", false);         if (method == "not found"){     method = "wang";        }
 537
 538             temp = validParameter.validFile(parameters, "ksize", false);                if (temp == "not found"){
 539                 temp = "8";
 540                 if (method == "zap") { temp = "7"; }
 541             }
 542                         m->mothurConvert(temp, kmerSize);
 543
 544                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 545                         m->mothurConvert(temp, match);
 546
 547                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 548                         m->mothurConvert(temp, misMatch);
 549
 550                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 551                         m->mothurConvert(temp, gapOpen);
 552
 553                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 554                         m->mothurConvert(temp, gapExtend);
 555
 556                         temp = validParameter.validFile(parameters, "numwanted", false);        if (temp == "not found"){       temp = "10";                    }
 557                         m->mothurConvert(temp, numWanted);
 558
 559                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found"){       temp = "0";                             }
 560                         m->mothurConvert(temp, cutoff);
 561
 562                         temp = validParameter.validFile(parameters, "probs", false);            if (temp == "not found"){       temp = "true";                  }
 563                         probs = m->isTrue(temp);
 564
 565             temp = validParameter.validFile(parameters, "shortcuts", false);    if (temp == "not found"){       temp = "true";                  }
 566                         writeShortcuts = m->isTrue(temp);
 567
 568                         //temp = validParameter.validFile(parameters, "flip", false);                   if (temp == "not found"){       temp = "T";                             }
 569                         //flip = m->isTrue(temp);
 570                         flip = true;
 571
 572                         temp = validParameter.validFile(parameters, "iters", false);            if (temp == "not found") { temp = "100";                        }
 573                         m->mothurConvert(temp, iters);
 574
 575
 576                         if ((method == "wang") && (search != "kmer"))  {
 577                                 m->mothurOut("The wang method requires the kmer search. " + search + " will be disregarded, and kmer will be used." ); m->mothurOutEndLine();
 578                                 search = "kmer";
 579                         }
 580
 581             if ((method == "zap") && ((search != "kmer") && (search != "align")))  {
 582                                 m->mothurOut("The zap method requires the kmer or align search. " + search + " will be disregarded, and kmer will be used." ); m->mothurOutEndLine();
 583                                 search = "kmer";
 584                         }
 585
 586             if (!abort) {
 587                 if (!hasCount) {
 588                     if (namefileNames.size() == 0){
 589                         if (fastaFileNames.size() != 0) {
 590                             vector<string> files; files.push_back(fastaFileNames[fastaFileNames.size()-1]);
 591                             parser.getNameFile(files);
 592                         }
 593                     }
 594                 }
 595             }
 596         }
 597         }
 598         catch(exception& e) {
 599                 m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
 600                 exit(1);
 601         }
 602 }
 603
 604 //**********************************************************************************************************************
 605 ClassifySeqsCommand::~ClassifySeqsCommand(){
 606         if (abort == false) {
 607                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 608         }
 609 }
 610 //**********************************************************************************************************************
 611
 612 int ClassifySeqsCommand::execute(){
 613         try {
 614                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 615
 616         string outputMethodTag = method + ".";
 617                 if(method == "wang"){   classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip, writeShortcuts);     }
 618                 else if(method == "knn"){       classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted, rand());                               }
 619         else if(method == "zap"){
 620             outputMethodTag = search + "_" + outputMethodTag;
 621             if (search == "kmer") {   classify = new KmerTree(templateFileName, taxonomyFileName, kmerSize, cutoff); }
 622             else {  classify = new AlignTree(templateFileName, taxonomyFileName, cutoff);  }
 623         }
 624                 else {
 625                         m->mothurOut(search + " is not a valid method option. I will run the command using wang.");
 626                         m->mothurOutEndLine();
 627                         classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip, writeShortcuts);
 628                 }
 629
 630                 if (m->control_pressed) { delete classify; return 0; }
 631
 632                 for (int s = 0; s < fastaFileNames.size(); s++) {
 633
 634                         m->mothurOut("Classifying sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 635
 636                         string baseTName = m->getSimpleName(taxonomyFileName);
 637                         if (taxonomyFileName == "saved") {  baseTName = rdb->getSavedTaxonomy();        }
 638
 639             //set rippedTaxName to
 640                         string RippedTaxName = "";
 641             bool foundDot = false;
 642             for (int i = baseTName.length()-1; i >= 0; i--) {
 643                 if (foundDot && (baseTName[i] != '.')) {  RippedTaxName = baseTName[i] + RippedTaxName; }
 644                 else if (foundDot && (baseTName[i] == '.')) {  break; }
 645                 else if (!foundDot && (baseTName[i] == '.')) {  foundDot = true; }
 646             }
 647             if (RippedTaxName != "") { RippedTaxName +=  "."; }
 648
 649                         if (outputDir == "") { outputDir += m->hasPath(fastaFileNames[s]); }
 650                         string newTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + outputMethodTag + getOutputFileNameTag("taxonomy");
 651                         string newaccnosFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + outputMethodTag +getOutputFileNameTag("accnos");
 652                         string tempTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "taxonomy.temp";
 653                         string taxSummary = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + outputMethodTag + getOutputFileNameTag("taxsummary");
 654
 655                         if ((method == "knn") && (search == "distance")) {
 656                                 string DistName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("matchdist");
 657                                 classify->setDistName(DistName);  outputNames.push_back(DistName); outputTypes["matchdist"].push_back(DistName);
 658                         }
 659
 660                         outputNames.push_back(newTaxonomyFile); outputTypes["taxonomy"].push_back(newTaxonomyFile);
 661                         outputNames.push_back(taxSummary);      outputTypes["taxsummary"].push_back(taxSummary);
 662
 663                         int start = time(NULL);
 664                         int numFastaSeqs = 0;
 665                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 666
 667 #ifdef USE_MPI
 668                                 int pid, numSeqsPerProcessor;
 669                                 int tag = 2001;
 670                                 vector<unsigned long long> MPIPos;
 671
 672                                 MPI_Status status;
 673                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 674                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 675
 676                                 MPI_File inMPI;
 677                                 MPI_File outMPINewTax;
 678                                 MPI_File outMPITempTax;
 679                                 MPI_File outMPIAcc;
 680
 681                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 682                                 int inMode=MPI_MODE_RDONLY;
 683
 684                                 char outNewTax[1024];
 685                                 strcpy(outNewTax, newTaxonomyFile.c_str());
 686
 687                                 char outTempTax[1024];
 688                                 strcpy(outTempTax, tempTaxonomyFile.c_str());
 689
 690                                 char outAcc[1024];
 691                                 strcpy(outAcc, newaccnosFile.c_str());
 692
 693                                 char inFileName[1024];
 694                                 strcpy(inFileName, fastaFileNames[s].c_str());
 695
 696                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 697                                 MPI_File_open(MPI_COMM_WORLD, outNewTax, outMode, MPI_INFO_NULL, &outMPINewTax);
 698                                 MPI_File_open(MPI_COMM_WORLD, outTempTax, outMode, MPI_INFO_NULL, &outMPITempTax);
 699                                 MPI_File_open(MPI_COMM_WORLD, outAcc, outMode, MPI_INFO_NULL, &outMPIAcc);
 700
 701                                 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;  }
 702
 703                                 if (pid == 0) { //you are the root process
 704
 705                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
 706
 707                                         //send file positions to all processes
 708                                         for(int i = 1; i < processors; i++) {
 709                                                 MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 710                                                 MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 711                                         }
 712
 713                                         //figure out how many sequences you have to align
 714                                         numSeqsPerProcessor = numFastaSeqs / processors;
 715                                         int startIndex =  pid * numSeqsPerProcessor;
 716                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 717
 718
 719                                         //align your part
 720                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos);
 721
 722                                         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;  }
 723
 724                                         for (int i = 1; i < processors; i++) {
 725                                                 int done;
 726                                                 MPI_Recv(&done, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 727                                         }
 728                                 }else{ //you are a child process
 729                                         MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 730                                         MPIPos.resize(numFastaSeqs+1);
 731                                         MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 732
 733                                         //figure out how many sequences you have to align
 734                                         numSeqsPerProcessor = numFastaSeqs / processors;
 735                                         int startIndex =  pid * numSeqsPerProcessor;
 736                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor;         }
 737
 738
 739                                         //align your part
 740                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPINewTax, outMPITempTax, outMPIAcc, MPIPos);
 741
 742                                         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;  }
 743
 744                                         int done = 0;
 745                                         MPI_Send(&done, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 746                                 }
 747
 748                                 //close files
 749                                 MPI_File_close(&inMPI);
 750                                 MPI_File_close(&outMPINewTax);
 751                                 MPI_File_close(&outMPITempTax);
 752                                 MPI_File_close(&outMPIAcc);
 753                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 754
 755 #else
 756
 757                         vector<unsigned long long> positions;
 758 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 759                         positions = m->divideFile(fastaFileNames[s], processors);
 760                         for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(new linePair(positions[i], positions[(i+1)]));  }
 761 #else
 762                         if (processors == 1) {
 763                                 lines.push_back(new linePair(0, 1000));
 764                         }else {
 765                                 positions = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs);
 766                 if (positions.size() < processors) { processors = positions.size(); }
 767
 768                                 //figure out how many sequences you have to process
 769                                 int numSeqsPerProcessor = numFastaSeqs / processors;
 770                                 for (int i = 0; i < processors; i++) {
 771                                         int startIndex =  i * numSeqsPerProcessor;
 772                                         if(i == (processors - 1)){      numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;   }
 773                                         lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
 774                                 }
 775                         }
 776 #endif
 777                         if(processors == 1){
 778                                 numFastaSeqs = driver(lines[0], newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]);
 779                         }else{
 780                                 numFastaSeqs = createProcesses(newTaxonomyFile, tempTaxonomyFile, newaccnosFile, fastaFileNames[s]);
 781                         }
 782 #endif
 783
 784                         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();
 785                 outputNames.push_back(newaccnosFile); outputTypes["accnos"].push_back(newaccnosFile);
 786             }else { m->mothurRemove(newaccnosFile); }
 787
 788                 m->mothurOutEndLine();
 789                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to classify " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 790                 start = time(NULL);
 791
 792
 793
 794                 #ifdef USE_MPI
 795                         if (pid == 0) {  //this part does not need to be paralellized
 796
 797                                 if(namefile != "") { m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();  MPIReadNamesFile(namefileNames[s]);  m->mothurOut("  Done."); m->mothurOutEndLine(); }
 798                 #else
 799                         //read namefile
 800                         if(namefile != "") {
 801
 802                             m->mothurOut("Reading " + namefileNames[s] + "..."); cout.flush();
 803                                 nameMap.clear(); //remove old names
 804                                 m->readNames(namefileNames[s], nameMap);
 805                                 m->mothurOut("  Done."); m->mothurOutEndLine();
 806                         }
 807                 #endif
 808
 809                 string group = "";
 810                 GroupMap* groupMap = NULL;
 811                 CountTable* ct = NULL;
 812                 PhyloSummary* taxaSum;
 813                 if (hasCount) {
 814                     ct = new CountTable();
 815                     ct->readTable(countfileNames[s]);
 816                     taxaSum = new PhyloSummary(taxonomyFileName, ct);
 817                     taxaSum->summarize(tempTaxonomyFile);
 818                 }else {
 819                     if (groupfile != "") {  group = groupfileNames[s]; groupMap = new GroupMap(group); groupMap->readMap(); }
 820
 821                     taxaSum = new PhyloSummary(taxonomyFileName, groupMap);
 822
 823                     if (m->control_pressed) { outputTypes.clear(); if (ct != NULL) { delete ct; }  if (groupMap != NULL) { delete groupMap; } delete taxaSum; for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]);        } delete classify; return 0; }
 824
 825                     if (namefile == "") {  taxaSum->summarize(tempTaxonomyFile);  }
 826                     else {
 827                         ifstream in;
 828                         m->openInputFile(tempTaxonomyFile, in);
 829
 830                         //read in users taxonomy file and add sequences to tree
 831                         string name, taxon;
 832
 833                         while(!in.eof()){
 834                             if (m->control_pressed) { outputTypes.clear();  if (ct != NULL) { delete ct; } if (groupMap != NULL) { delete groupMap; } delete taxaSum; for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]);        } delete classify; return 0; }
 835
 836                             in >> name >> taxon; m->gobble(in);
 837
 838                             itNames = nameMap.find(name);
 839
 840                             if (itNames == nameMap.end()) {
 841                                 m->mothurOut(name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
 842                             }else{
 843                                 for (int i = 0; i < itNames->second.size(); i++) {
 844                                     taxaSum->addSeqToTree(itNames->second[i], taxon);  //add it as many times as there are identical seqs
 845                                 }
 846                                 itNames->second.clear();
 847                                 nameMap.erase(itNames->first);
 848                             }
 849                         }
 850                         in.close();
 851                     }
 852                 }
 853                         m->mothurRemove(tempTaxonomyFile);
 854
 855                         if (m->control_pressed) {  outputTypes.clear(); if (ct != NULL) { delete ct; } if (groupMap != NULL) { delete groupMap; } for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } delete classify; return 0; }
 856
 857                         //print summary file
 858                         ofstream outTaxTree;
 859                         m->openOutputFile(taxSummary, outTaxTree);
 860                         taxaSum->print(outTaxTree);
 861                         outTaxTree.close();
 862
 863                         //output taxonomy with the unclassified bins added
 864                         ifstream inTax;
 865                         m->openInputFile(newTaxonomyFile, inTax);
 866
 867                         ofstream outTax;
 868                         string unclass = newTaxonomyFile + ".unclass.temp";
 869                         m->openOutputFile(unclass, outTax);
 870
 871                         //get maxLevel from phylotree so you know how many 'unclassified's to add
 872                         int maxLevel = taxaSum->getMaxLevel();
 873
 874                         //read taxfile - this reading and rewriting is done to preserve the confidence scores.
 875                         string name, taxon;
 876                         while (!inTax.eof()) {
 877                                 if (m->control_pressed) { outputTypes.clear(); if (ct != NULL) { delete ct; }  if (groupMap != NULL) { delete groupMap; } delete taxaSum; for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } m->mothurRemove(unclass); delete classify; return 0; }
 878
 879                                 inTax >> name >> taxon; m->gobble(inTax);
 880
 881                                 string newTax = addUnclassifieds(taxon, maxLevel);
 882
 883                                 outTax << name << '\t' << newTax << endl;
 884                         }
 885                         inTax.close();
 886                         outTax.close();
 887
 888             if (ct != NULL) { delete ct; }
 889             if (groupMap != NULL) { delete groupMap; } delete taxaSum;
 890                         m->mothurRemove(newTaxonomyFile);
 891                         rename(unclass.c_str(), newTaxonomyFile.c_str());
 892
 893                         m->mothurOutEndLine();
 894                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numFastaSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 895
 896                         #ifdef USE_MPI
 897                                 }
 898                         #endif
 899                 }
 900         delete classify;
 901
 902         m->mothurOutEndLine();
 903         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 904         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 905         m->mothurOutEndLine();
 906
 907                 //set taxonomy file as new current taxonomyfile
 908                 string current = "";
 909                 itTypes = outputTypes.find("taxonomy");
 910                 if (itTypes != outputTypes.end()) {
 911                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 912                 }
 913
 914                 current = "";
 915                 itTypes = outputTypes.find("accnos");
 916                 if (itTypes != outputTypes.end()) {
 917                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 918                 }
 919
 920
 921
 922                 return 0;
 923         }
 924         catch(exception& e) {
 925                 m->errorOut(e, "ClassifySeqsCommand", "execute");
 926                 exit(1);
 927         }
 928 }
 929
 930 /**************************************************************************************************/
 931 string ClassifySeqsCommand::addUnclassifieds(string tax, int maxlevel) {
 932         try{
 933                 string newTax, taxon;
 934                 int level = 0;
 935
 936                 //keep what you have counting the levels
 937                 while (tax.find_first_of(';') != -1) {
 938                         //get taxon
 939                         taxon = tax.substr(0,tax.find_first_of(';'))+';';
 940                         tax = tax.substr(tax.find_first_of(';')+1, tax.length());
 941                         newTax += taxon;
 942                         level++;
 943                 }
 944
 945                 //add "unclassified" until you reach maxLevel
 946                 while (level < maxlevel) {
 947                         newTax += "unclassified;";
 948                         level++;
 949                 }
 950
 951                 return newTax;
 952         }
 953         catch(exception& e) {
 954                 m->errorOut(e, "ClassifySeqsCommand", "addUnclassifieds");
 955                 exit(1);
 956         }
 957 }
 958
 959 /**************************************************************************************************/
 960
 961 int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile, string accnos, string filename) {
 962         try {
 963
 964                 int num = 0;
 965                 processIDS.clear();
 966
 967 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 968                 int process = 1;
 969
 970                 //loop through and create all the processes you want
 971                 while (process != processors) {
 972                         int pid = fork();
 973
 974                         if (pid > 0) {
 975                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 976                                 process++;
 977                         }else if (pid == 0){
 978                                 num = driver(lines[process], taxFileName + toString(getpid()) + ".temp", tempTaxFile + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", filename);
 979
 980                                 //pass numSeqs to parent
 981                                 ofstream out;
 982                                 string tempFile = filename + toString(getpid()) + ".num.temp";
 983                                 m->openOutputFile(tempFile, out);
 984                                 out << num << endl;
 985                                 out.close();
 986
 987                                 exit(0);
 988                         }else {
 989                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 990                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 991                                 exit(0);
 992                         }
 993                 }
 994
 995                 //parent does its part
 996                 num = driver(lines[0], taxFileName, tempTaxFile, accnos, filename);
 997
 998                 //force parent to wait until all the processes are done
 999                 for (int i=0;i<processIDS.size();i++) {
1000                         int temp = processIDS[i];
1001                         wait(&temp);
1002                 }
1003
1004                 for (int i = 0; i < processIDS.size(); i++) {
1005                         ifstream in;
1006                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
1007                         m->openInputFile(tempFile, in);
1008                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1009                         in.close(); m->mothurRemove(m->getFullPathName(tempFile));
1010                 }
1011 #else
1012                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1013                 //Windows version shared memory, so be careful when passing variables through the alignData struct.
1014                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1015                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1016
1017                 vector<classifyData*> pDataArray;
1018                 DWORD   dwThreadIdArray[processors-1];
1019                 HANDLE  hThreadArray[processors-1];
1020
1021                 //Create processor worker threads.
1022                 for( int i=0; i<processors-1; i++ ){
1023                         // Allocate memory for thread data.
1024                         string extension = "";
1025                         if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
1026
1027                         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, writeShortcuts);
1028                         pDataArray.push_back(tempclass);
1029
1030                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
1031                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1032                         hThreadArray[i] = CreateThread(NULL, 0, MyClassThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
1033
1034                 }
1035
1036                 //parent does its part
1037                 num = driver(lines[processors-1], taxFileName + toString(processors-1) + ".temp", tempTaxFile + toString(processors-1) + ".temp", accnos + toString(processors-1) + ".temp", filename);
1038                 processIDS.push_back((processors-1));
1039
1040                 //Wait until all threads have terminated.
1041                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
1042
1043                 //Close all thread handles and free memory allocations.
1044                 for(int i=0; i < pDataArray.size(); i++){
1045                         num += pDataArray[i]->count;
1046                         CloseHandle(hThreadArray[i]);
1047                         delete pDataArray[i];
1048                 }
1049
1050         #endif
1051         vector<string> nonBlankAccnosFiles;
1052                 if (!(m->isBlank(accnos))) { nonBlankAccnosFiles.push_back(accnos); }
1053                 else { m->mothurRemove(accnos); } //remove so other files can be renamed to it
1054
1055                 for(int i=0;i<processIDS.size();i++){
1056                         m->appendFiles((taxFileName + toString(processIDS[i]) + ".temp"), taxFileName);
1057                         m->appendFiles((tempTaxFile + toString(processIDS[i]) + ".temp"), tempTaxFile);
1058             if (!(m->isBlank(accnos + toString(processIDS[i]) + ".temp"))) {
1059                                 nonBlankAccnosFiles.push_back(accnos + toString(processIDS[i]) + ".temp");
1060                         }else { m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));  }
1061
1062                         m->mothurRemove((m->getFullPathName(taxFileName) + toString(processIDS[i]) + ".temp"));
1063                         m->mothurRemove((m->getFullPathName(tempTaxFile) + toString(processIDS[i]) + ".temp"));
1064                 }
1065
1066         //append accnos files
1067                 if (nonBlankAccnosFiles.size() != 0) {
1068                         rename(nonBlankAccnosFiles[0].c_str(), accnos.c_str());
1069
1070                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
1071                                 m->appendFiles(nonBlankAccnosFiles[h], accnos);
1072                                 m->mothurRemove(nonBlankAccnosFiles[h]);
1073                         }
1074                 }else { //recreate the accnosfile if needed
1075                         ofstream out;
1076                         m->openOutputFile(accnos, out);
1077                         out.close();
1078                 }
1079
1080                 return num;
1081
1082         }
1083         catch(exception& e) {
1084                 m->errorOut(e, "ClassifySeqsCommand", "createProcesses");
1085                 exit(1);
1086         }
1087 }
1088 //**********************************************************************************************************************
1089
1090 int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempTFName, string accnos, string filename){
1091         try {
1092                 ofstream outTax;
1093                 m->openOutputFile(taxFName, outTax);
1094
1095                 ofstream outTaxSimple;
1096                 m->openOutputFile(tempTFName, outTaxSimple);
1097
1098                 ofstream outAcc;
1099                 m->openOutputFile(accnos, outAcc);
1100
1101                 ifstream inFASTA;
1102                 m->openInputFile(filename, inFASTA);
1103
1104                 string taxonomy;
1105
1106                 inFASTA.seekg(filePos->start);
1107
1108                 bool done = false;
1109                 int count = 0;
1110
1111                 while (!done) {
1112                         if (m->control_pressed) {
1113                                 inFASTA.close();
1114                                 outTax.close();
1115                                 outTaxSimple.close();
1116                                 outAcc.close(); return 0; }
1117
1118                         Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
1119
1120                         if (candidateSeq->getName() != "") {
1121
1122                                 taxonomy = classify->getTaxonomy(candidateSeq);
1123
1124                                 if (m->control_pressed) { delete candidateSeq; return 0; }
1125
1126                                 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(); }
1127
1128                                 //output confidence scores or not
1129                                 if (probs) {
1130                                         outTax << candidateSeq->getName() << '\t' << taxonomy << endl;
1131                                 }else{
1132                                         outTax << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
1133                                 }
1134
1135                                 if (classify->getFlipped()) { outAcc << candidateSeq->getName() << endl; }
1136
1137                                 outTaxSimple << candidateSeq->getName() << '\t' << classify->getSimpleTax() << endl;
1138
1139                                 count++;
1140                         }
1141                         delete candidateSeq;
1142
1143                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1144                                 unsigned long long pos = inFASTA.tellg();
1145                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
1146                         #else
1147                                 if (inFASTA.eof()) { break; }
1148                         #endif
1149
1150                         //report progress
1151                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1152
1153                 }
1154                 //report progress
1155                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1156
1157                 inFASTA.close();
1158                 outTax.close();
1159                 outTaxSimple.close();
1160                 outAcc.close();
1161
1162                 return count;
1163         }
1164         catch(exception& e) {
1165                 m->errorOut(e, "ClassifySeqsCommand", "driver");
1166                 exit(1);
1167         }
1168 }
1169 //**********************************************************************************************************************
1170 #ifdef USE_MPI
1171 int ClassifySeqsCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& newFile, MPI_File& tempFile, MPI_File& accFile, vector<unsigned long long>& MPIPos){
1172         try {
1173                 MPI_Status statusNew;
1174                 MPI_Status statusTemp;
1175                 MPI_Status statusAcc;
1176                 MPI_Status status;
1177
1178                 int pid;
1179                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
1180
1181                 string taxonomy;
1182                 string outputString;
1183
1184                 for(int i=0;i<num;i++){
1185
1186                         if (m->control_pressed) { return 0; }
1187
1188                         //read next sequence
1189                         int length = MPIPos[start+i+1] - MPIPos[start+i];
1190                         char* buf4 = new char[length];
1191                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
1192
1193                         string tempBuf = buf4;
1194                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
1195                         istringstream iss (tempBuf,istringstream::in);
1196                         delete buf4;
1197
1198                         Sequence* candidateSeq = new Sequence(iss);
1199
1200                         if (candidateSeq->getName() != "") {
1201                                 taxonomy = classify->getTaxonomy(candidateSeq);
1202
1203                                 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(); }
1204
1205                                 //output confidence scores or not
1206                                 if (probs) {
1207                                         outputString =  candidateSeq->getName() + "\t" + taxonomy + "\n";
1208                                 }else{
1209                                         outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
1210                                 }
1211
1212                                 int length = outputString.length();
1213                                 char* buf2 = new char[length];
1214                                 memcpy(buf2, outputString.c_str(), length);
1215
1216                                 MPI_File_write_shared(newFile, buf2, length, MPI_CHAR, &statusNew);
1217                                 delete buf2;
1218
1219                                 outputString =  candidateSeq->getName() + "\t" + classify->getSimpleTax() + "\n";
1220                                 length = outputString.length();
1221                                 char* buf = new char[length];
1222                                 memcpy(buf, outputString.c_str(), length);
1223
1224                                 MPI_File_write_shared(tempFile, buf, length, MPI_CHAR, &statusTemp);
1225                                 delete buf;
1226
1227                                 if (classify->getFlipped()) {
1228                                         outputString =  candidateSeq->getName() + "\n";
1229                                         length = outputString.length();
1230                                         char* buf3 = new char[length];
1231                                         memcpy(buf3, outputString.c_str(), length);
1232
1233                                         MPI_File_write_shared(accFile, buf3, length, MPI_CHAR, &statusAcc);
1234                                         delete buf3;
1235                                 }
1236
1237                         }
1238                         delete candidateSeq;
1239
1240                         if((i+1) % 100 == 0){   cout << "Classifying sequence " << (i+1) << endl;       }
1241                 }
1242
1243                 if(num % 100 != 0){     cout << "Classifying sequence " << (num) << endl;       }
1244
1245
1246                 return 1;
1247         }
1248         catch(exception& e) {
1249                 m->errorOut(e, "ClassifySeqsCommand", "driverMPI");
1250                 exit(1);
1251         }
1252 }
1253
1254 //**********************************************************************************************************************
1255 int ClassifySeqsCommand::MPIReadNamesFile(string nameFilename){
1256         try {
1257
1258                 nameMap.clear(); //remove old names
1259
1260                 MPI_File inMPI;
1261                 MPI_Offset size;
1262                 MPI_Status status;
1263
1264                 //char* inFileName = new char[nameFilename.length()];
1265                 //memcpy(inFileName, nameFilename.c_str(), nameFilename.length());
1266
1267                 char inFileName[1024];
1268                 strcpy(inFileName, nameFilename.c_str());
1269
1270                 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
1271                 MPI_File_get_size(inMPI, &size);
1272                 //delete inFileName;
1273
1274                 char* buffer = new char[size];
1275                 MPI_File_read(inMPI, buffer, size, MPI_CHAR, &status);
1276
1277                 string tempBuf = buffer;
1278                 if (tempBuf.length() > size) { tempBuf = tempBuf.substr(0, size);  }
1279                 istringstream iss (tempBuf,istringstream::in);
1280                 delete buffer;
1281
1282                 string firstCol, secondCol;
1283                 while(!iss.eof()) {
1284                         iss >> firstCol >> secondCol; m->gobble(iss);
1285
1286                         vector<string> temp;
1287                         m->splitAtComma(secondCol, temp);
1288
1289                         nameMap[firstCol] = temp;
1290                 }
1291
1292                 MPI_File_close(&inMPI);
1293
1294                 return 1;
1295         }
1296         catch(exception& e) {
1297                 m->errorOut(e, "ClassifySeqsCommand", "MPIReadNamesFile");
1298                 exit(1);
1299         }
1300 }
1301 #endif
1302 /**************************************************************************************************/