chimeraseqscommand.cpp

   1 /*
   2  *  chimeraseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 6/29/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeraseqscommand.h"
  11 #include "bellerophon.h"
  12 #include "pintail.h"
  13 #include "ccode.h"
  14 #include "chimeracheckrdp.h"
  15 #include "chimeraslayer.h"
  16
  17
  18 //***************************************************************************************************************
  19
  20 ChimeraSeqsCommand::ChimeraSeqsCommand(string option)  {
  21         try {
  22                 abort = false;
  23
  24                 //allow user to run help
  25                 if(option == "help") { help(); abort = true; }
  26
  27                 else {
  28                         //valid paramters for this command
  29                         string Array[] =  {"fasta", "filter", "correction", "processors", "method", "window", "increment", "template", "conservation", "quantile", "mask",
  30                         "numwanted", "ksize", "svg", "name", "match","mismatch", "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  31                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  32
  33                         OptionParser parser(option);
  34                         map<string,string> parameters = parser.getParameters();
  35
  36                         ValidParameters validParameter;
  37                         map<string,string>::iterator it;
  38
  39                         //check to make sure all parameters are valid for command
  40                         for (it = parameters.begin(); it != parameters.end(); it++) {
  41                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  42                         }
  43
  44                         //if the user changes the input directory command factory will send this info to us in the output parameter
  45                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  46                         if (inputDir == "not found"){   inputDir = "";          }
  47                         else {
  48                                 string path;
  49                                 it = parameters.find("fasta");
  50                                 //user has given a template file
  51                                 if(it != parameters.end()){
  52                                         path = hasPath(it->second);
  53                                         //if the user has not given a path then, add inputdir. else leave path alone.
  54                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  55                                 }
  56
  57                                 it = parameters.find("template");
  58                                 //user has given a template file
  59                                 if(it != parameters.end()){
  60                                         path = hasPath(it->second);
  61                                         //if the user has not given a path then, add inputdir. else leave path alone.
  62                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  63                                 }
  64
  65                                 it = parameters.find("conservation");
  66                                 //user has given a template file
  67                                 if(it != parameters.end()){
  68                                         path = hasPath(it->second);
  69                                         //if the user has not given a path then, add inputdir. else leave path alone.
  70                                         if (path == "") {       parameters["conservation"] = inputDir + it->second;             }
  71                                 }
  72
  73                                 it = parameters.find("quantile");
  74                                 //user has given a template file
  75                                 if(it != parameters.end()){
  76                                         path = hasPath(it->second);
  77                                         //if the user has not given a path then, add inputdir. else leave path alone.
  78                                         if (path == "") {       parameters["quantile"] = inputDir + it->second;         }
  79                                 }
  80
  81                                 it = parameters.find("name");
  82                                 //user has given a template file
  83                                 if(it != parameters.end()){
  84                                         path = hasPath(it->second);
  85                                         //if the user has not given a path then, add inputdir. else leave path alone.
  86                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  87                                 }
  88                         }
  89
  90
  91                         //check for required parameters
  92                         fastafile = validParameter.validFile(parameters, "fasta", true);
  93                         if (fastafile == "not open") { abort = true; }
  94                         else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.seqs command."); m->mothurOutEndLine(); abort = true;  }
  95
  96                         //if the user changes the output directory command factory will send this info to us in the output parameter
  97                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  98                                 outputDir = "";
  99                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
 100                         }
 101
 102                         templatefile = validParameter.validFile(parameters, "template", true);
 103                         if (templatefile == "not open") { abort = true; }
 104                         else if (templatefile == "not found") { templatefile = "";  }
 105
 106                         consfile = validParameter.validFile(parameters, "conservation", true);
 107                         if (consfile == "not open") { abort = true; }
 108                         else if (consfile == "not found") { consfile = "";  }
 109
 110                         quanfile = validParameter.validFile(parameters, "quantile", true);
 111                         if (quanfile == "not open") { abort = true; }
 112                         else if (quanfile == "not found") { quanfile = "";  }
 113
 114                         namefile = validParameter.validFile(parameters, "name", true);
 115                         if (namefile == "not open") { abort = true; }
 116                         else if (namefile == "not found") { namefile = "";  }
 117
 118                         maskfile = validParameter.validFile(parameters, "mask", false);
 119                         if (maskfile == "not found") { maskfile = "";  }
 120                         else if (maskfile != "default")  {
 121                                 if (inputDir != "") {
 122                                         string path = hasPath(maskfile);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       maskfile = inputDir + maskfile;         }
 125                                 }
 126
 127                                 ifstream in;
 128                                 int     ableToOpen = openInputFile(maskfile, in);
 129                                 if (ableToOpen == 1) { abort = true; }
 130                                 in.close();
 131                         }
 132
 133                         method = validParameter.validFile(parameters, "method", false);                 if (method == "not found") { method = "pintail"; }
 134
 135                         string temp;
 136                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 137                         filter = isTrue(temp);
 138
 139                         temp = validParameter.validFile(parameters, "correction", false);               if (temp == "not found") { temp = "T"; }
 140                         correction = isTrue(temp);
 141
 142                         temp = validParameter.validFile(parameters, "processors", false);               if (temp == "not found") { temp = "1"; }
 143                         convert(temp, processors);
 144
 145                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 146                         convert(temp, ksize);
 147
 148                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 149                         svg = isTrue(temp);
 150
 151                         temp = validParameter.validFile(parameters, "window", false);
 152                         if ((temp == "not found") && (method == "chimeraslayer")) { temp = "50"; }
 153                         else if (temp == "not found") { temp = "0"; }
 154                         convert(temp, window);
 155
 156                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 157                         convert(temp, match);
 158
 159                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 160                         convert(temp, mismatch);
 161
 162                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 163                         convert(temp, divR);
 164
 165                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 166                         convert(temp, minSimilarity);
 167
 168                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 169                         convert(temp, minCoverage);
 170
 171                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 172                         convert(temp, minBS);
 173
 174                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "10"; }
 175                         convert(temp, minSNP);
 176
 177                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 178                         convert(temp, parents);
 179
 180                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "f"; }
 181                         realign = isTrue(temp);
 182
 183                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "distance"; }
 184
 185                         temp = validParameter.validFile(parameters, "iters", false);
 186                         if ((temp == "not found") && (method == "chimeraslayer")) { temp = "100"; }
 187                         else if (temp == "not found") { temp = "1000"; }
 188                         convert(temp, iters);
 189
 190                         temp = validParameter.validFile(parameters, "increment", false);
 191                         if ((temp == "not found") && (method == "chimeracheck")) { temp = "10"; }
 192                         else if ((temp == "not found") && (method == "chimeraslayer")) { temp = "5"; }
 193                         else if (temp == "not found") { temp = "25"; }
 194                         convert(temp, increment);
 195
 196                         temp = validParameter.validFile(parameters, "numwanted", false);
 197                         if ((temp == "not found") && (method == "chimeraslayer")) { temp = "15"; }
 198                         else if (temp == "not found") { temp = "20"; }
 199                         convert(temp, numwanted);
 200
 201                         if ((search != "distance") && (search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 202
 203                         if (((method != "bellerophon")) && (templatefile == "")) { m->mothurOut("You must provide a template file with the pintail, ccode, chimeraslayer or chimeracheck methods."); m->mothurOutEndLine(); abort = true;  }
 204
 205
 206                 }
 207         }
 208         catch(exception& e) {
 209                 m->errorOut(e, "ChimeraSeqsCommand", "ChimeraSeqsCommand");
 210                 exit(1);
 211         }
 212 }
 213 //**********************************************************************************************************************
 214
 215 void ChimeraSeqsCommand::help(){
 216         try {
 217
 218                 //"fasta", "filter", "correction", "processors", "method", "window", "increment", "template", "conservation", "quantile", "mask", "numwanted", "ksize", "svg", "name"
 219                 //m->mothurOut("chimera.seqs ASSUMES that your sequences are ALIGNED and if using a template that the template file sequences are the same length as the fasta file sequences.\n\n");
 220                 m->mothurOut("The chimera.seqs command reads a fastafile and creates list of potentially chimeric sequences.\n");
 221                 m->mothurOut("The chimera.seqs command parameters are fasta, filter, correction, processors, mask, method, window, increment, template, conservation, quantile, numwanted, ksize, svg, name, iters, search, realign.\n");
 222                 m->mothurOut("The fasta parameter is always required and template is required if using pintail, ccode or chimeracheck.\n");
 223                 m->mothurOut("The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter. \n");
 224                 m->mothurOut("The correction parameter allows you to put more emphasis on the distance between highly similar sequences and less emphasis on the differences between remote homologs.\n");
 225                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 226                 m->mothurOut("The method parameter allows you to specify the method for finding chimeric sequences.  The default is pintail. Options include bellerophon, ccode and chimeracheck \n");
 227                 m->mothurOut("The mask parameter allows you to specify a file containing one sequence you wish to use as a mask for the your sequences. \n");
 228                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras. \n");
 229                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences.\n");
 230                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences. \n");
 231                 m->mothurOut("The conservation parameter allows you to enter a frequency file containing the highest bases frequency at each place in the alignment.\n");
 232                 m->mothurOut("The quantile parameter allows you to enter a file containing quantiles for a template files sequences.\n");
 233                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with.\n");
 234                 m->mothurOut("The ksize parameter allows you to input kmersize. \n");
 235                 m->mothurOut("The svg parameter allows you to specify whether or not you would like a svg file outputted for each query sequence.\n");
 236                 m->mothurOut("The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n");
 237                 m->mothurOut("The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method.\n");
 238                 m->mothurOut("The minsim parameter allows you .... \n");
 239                 m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
 240                 m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
 241                 m->mothurOut("The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 10) \n");
 242                 m->mothurOut("The search parameter allows you to specify search method for finding the closest parent. Choices are distance, blast, and kmer, default distance.  -used only by chimeraslayer. \n");
 243                 m->mothurOut("The realign parameter allows you to realign the query to the potential paretns. Choices are true or false, default false.  -used only by chimeraslayer. \n");
 244                 m->mothurOut("NOT ALL PARAMETERS ARE USED BY ALL METHODS. Please look below for method specifics.\n\n");
 245                 m->mothurOut("Details for each method: \n");
 246                 m->mothurOut("\tpintail: \n");
 247                 m->mothurOut("\t\tparameters: fasta=required, template=required, filter=F, mask=no mask, processors=1, window=300, increment=25, conservation=not required, but will improve speed, quantile=not required, but will greatly improve speed. \n");
 248                 m->mothurOut("\t\tIf you have run chimera.seqs using pintail a .quan and .freq file will be created for your template, if you have not provided them for use in future command executions.\n");
 249                 m->mothurOut("\tbellerophon: \n");
 250                 m->mothurOut("\t\tparameters: fasta=required, filter=F, processors=1, window=1/4 length of seq, increment=25, correction=T. \n");
 251                 m->mothurOut("\tccode: \n");
 252                 m->mothurOut("\t\tparameters: fasta=required, template=required, filter=F, mask=no mask, processors=1, window=10% of length, numwanted=20\n");
 253                 m->mothurOut("\tchimeracheck: \n");
 254                 m->mothurOut("\t\tparameters: fasta=required, template=required, processors=1, increment=10, ksize=7, svg=F, name=none\n\n");
 255                 m->mothurOut("\tchimeraslayer: \n");
 256                 m->mothurOut("\t\tparameters: fasta=required, template=required, processors=1, increment=10, mask=no mask, numwanted=10, match=5, mismatch=-4, divergence=1.0, minsim=90, parents=5, iters=1000, window=100. \n\n");
 257                 m->mothurOut("The chimera.seqs command should be in the following format: \n");
 258                 m->mothurOut("chimera.seqs(fasta=yourFastaFile, filter=yourFilter, correction=yourCorrection, processors=yourProcessors, method=bellerophon) \n");
 259                 m->mothurOut("Example: chimera.seqs(fasta=AD.align, filter=True, correction=true, method=bellerophon, window=200) \n");
 260                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 261         }
 262         catch(exception& e) {
 263                 m->errorOut(e, "ChimeraSeqsCommand", "help");
 264                 exit(1);
 265         }
 266 }
 267
 268 //***************************************************************************************************************
 269
 270 ChimeraSeqsCommand::~ChimeraSeqsCommand(){      /*      do nothing      */      }
 271
 272 //***************************************************************************************************************
 273
 274 int ChimeraSeqsCommand::execute(){
 275         try{
 276
 277                 if (abort == true) { return 0; }
 278
 279                 int start = time(NULL);
 280
 281                 if (method == "bellerophon")                    {               chimera = new Bellerophon(fastafile, outputDir);                        }
 282                 else if (method == "pintail")                   {               chimera = new Pintail(fastafile, outputDir);                            }
 283                 else if (method == "ccode")                             {               chimera = new Ccode(fastafile, outputDir);                                      }
 284                 else if (method == "chimeracheck")              {               chimera = new ChimeraCheckRDP(fastafile, outputDir);            }
 285                 else if (method == "chimeraslayer")             {               chimera = new ChimeraSlayer(search, realign, fastafile);        }
 286                 else { m->mothurOut("Not a valid method."); m->mothurOutEndLine(); return 0;            }
 287
 288                 //set user options
 289                 if (maskfile == "default") { m->mothurOut("I am using the default 236627 EU009184.1 Shigella dysenteriae str. FBD013."); m->mothurOutEndLine();  }
 290
 291                 chimera->setCons(consfile);
 292                 chimera->setQuantiles(quanfile);
 293                 chimera->setMask(maskfile);
 294                 chimera->setFilter(filter);
 295                 chimera->setCorrection(correction);
 296                 chimera->setProcessors(processors);
 297                 chimera->setWindow(window);
 298                 chimera->setIncrement(increment);
 299                 chimera->setNumWanted(numwanted);
 300                 chimera->setKmerSize(ksize);
 301                 chimera->setSVG(svg);
 302                 chimera->setName(namefile);
 303                 chimera->setMatch(match);
 304                 chimera->setMisMatch(mismatch);
 305                 chimera->setDivR(divR);
 306                 chimera->setParents(parents);
 307                 chimera->setMinSim(minSimilarity);
 308                 chimera->setMinCoverage(minCoverage);
 309                 chimera->setMinBS(minBS);
 310                 chimera->setMinSNP(minSNP);
 311                 chimera->setIters(iters);
 312
 313
 314                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + method + maskfile + ".chimeras";
 315                 string accnosFileName = outputDir + getRootName(getSimpleName(fastafile)) + method + maskfile + ".accnos";
 316                 bool hasAccnos = true;
 317
 318                 if (method == "bellerophon") {//run bellerophon separately since you need to read entire fastafile to run it
 319                         chimera->getChimeras();
 320
 321                         if (m->control_pressed) { delete chimera;       return 0;       }
 322
 323                         ofstream out;
 324                         openOutputFile(outputFileName, out);
 325
 326                         ofstream out2;
 327                         openOutputFile(accnosFileName, out2);
 328
 329                         chimera->print(out, out2);
 330                         out.close();
 331                         out2.close();
 332
 333                         if (m->control_pressed) { remove(accnosFileName.c_str()); remove(outputFileName.c_str()); delete chimera;       return 0;       }
 334
 335                         //delete accnos file if its blank
 336                         if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 337
 338                         m->mothurOutEndLine();
 339                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 340                         m->mothurOut(outputFileName); m->mothurOutEndLine();
 341                         if (hasAccnos) {  m->mothurOut(accnosFileName); m->mothurOutEndLine();  }
 342                         m->mothurOutEndLine();
 343
 344                         delete chimera;
 345                         return 0;
 346                 }
 347
 348                 //reads template
 349                 chimera->setTemplateFile(templatefile);
 350
 351                 if (m->control_pressed) { delete chimera;       return 0;       }
 352
 353                 if  (method != "chimeracheck") {
 354                         if (chimera->getUnaligned()) {
 355                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 356                                 delete chimera;
 357                                 return 0;
 358                         }
 359                 }
 360
 361                 //some methods need to do prep work before processing the chimeras
 362                 chimera->doPrep();
 363
 364                 if (m->control_pressed) { delete chimera;       return 0;       }
 365
 366                 templateSeqsLength = chimera->getLength();
 367
 368                 ofstream outHeader;
 369                 string tempHeader = outputDir + getRootName(getSimpleName(fastafile)) + method + maskfile + ".chimeras.tempHeader";
 370                 openOutputFile(tempHeader, outHeader);
 371
 372                 chimera->printHeader(outHeader);
 373                 outHeader.close();
 374
 375
 376                 //break up file
 377                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 378                         if(processors == 1){
 379                                 ifstream inFASTA;
 380                                 openInputFile(fastafile, inFASTA);
 381                                 numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 382                                 inFASTA.close();
 383
 384                                 lines.push_back(new linePair(0, numSeqs));
 385
 386                                 driver(lines[0], outputFileName, fastafile, accnosFileName);
 387
 388                                 if (m->control_pressed) {
 389                                         remove(outputFileName.c_str());
 390                                         remove(tempHeader.c_str());
 391                                         remove(accnosFileName.c_str());
 392                                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 393                                         delete chimera;
 394                                         return 0;
 395                                 }
 396
 397                                 //delete accnos file if its blank
 398                                 if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 399
 400                         }else{
 401                                 vector<int> positions;
 402                                 processIDS.resize(0);
 403
 404                                 ifstream inFASTA;
 405                                 openInputFile(fastafile, inFASTA);
 406
 407                                 string input;
 408                                 while(!inFASTA.eof()){
 409                                         input = getline(inFASTA);
 410                                         if (input.length() != 0) {
 411                                                 if(input[0] == '>'){    long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);  }
 412                                         }
 413                                 }
 414                                 inFASTA.close();
 415
 416                                 numSeqs = positions.size();
 417
 418                                 int numSeqsPerProcessor = numSeqs / processors;
 419
 420                                 for (int i = 0; i < processors; i++) {
 421                                         long int startPos = positions[ i * numSeqsPerProcessor ];
 422                                         if(i == processors - 1){
 423                                                 numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;
 424                                         }
 425                                         lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 426                                 }
 427
 428
 429                                 createProcesses(outputFileName, fastafile, accnosFileName);
 430
 431                                 rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 432
 433                                 //append output files
 434                                 for(int i=1;i<processors;i++){
 435                                         appendOutputFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 436                                         remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 437                                 }
 438
 439                                 vector<string> nonBlankAccnosFiles;
 440                                 //delete blank accnos files generated with multiple processes
 441                                 for(int i=0;i<processors;i++){
 442                                         if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
 443                                                 nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
 444                                         }else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());  }
 445                                 }
 446
 447                                 //append accnos files
 448                                 if (nonBlankAccnosFiles.size() != 0) {
 449                                         rename(nonBlankAccnosFiles[0].c_str(), accnosFileName.c_str());
 450
 451                                         for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
 452                                                 appendOutputFiles(nonBlankAccnosFiles[h], accnosFileName);
 453                                                 remove(nonBlankAccnosFiles[h].c_str());
 454                                         }
 455                                 }else{ hasAccnos = false;  }
 456
 457                                 if (m->control_pressed) {
 458                                         remove(outputFileName.c_str());
 459                                         remove(accnosFileName.c_str());
 460                                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 461                                         delete chimera;
 462                                         return 0;
 463                                 }
 464
 465                         }
 466
 467                 #else
 468                         ifstream inFASTA;
 469                         openInputFile(candidateFileNames[s], inFASTA);
 470                         numSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 471                         inFASTA.close();
 472                         lines.push_back(new linePair(0, numSeqs));
 473
 474                         driver(lines[0], outputFileName, fastafile, accnosFileName);
 475
 476                         if (m->control_pressed) {
 477                                         remove(outputFileName.c_str());
 478                                         remove(tempHeader.c_str());
 479                                         remove(accnosFileName.c_str());
 480                                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 481                                         delete chimera;
 482                                         return 0;
 483                         }
 484
 485                         //delete accnos file if its blank
 486                         if (isBlank(accnosFileName)) {  remove(accnosFileName.c_str());  hasAccnos = false; }
 487                 #endif
 488
 489                 //m->mothurOut("Output File Names: ");
 490                 //if ((filter) && (method == "bellerophon")) { m->mothurOut(
 491                 //if (outputDir == "") { fastafile = getRootName(fastafile) + "filter.fasta"; }
 492                 //      else                             { fastafile = outputDir + getRootName(getSimpleName(fastafile)) + "filter.fasta"; }
 493
 494                 appendOutputFiles(tempHeader, outputFileName);
 495
 496                 remove(outputFileName.c_str());
 497                 rename(tempHeader.c_str(), outputFileName.c_str());
 498
 499                 delete chimera;
 500
 501                 if (method == "chimeracheck") { remove(accnosFileName.c_str());  m->mothurOutEndLine(); m->mothurOut("This method does not determine if a sequence is chimeric, but allows you to make that determination based on the IS values."); m->mothurOutEndLine();  }
 502
 503                 m->mothurOutEndLine();
 504                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 505                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 506                 if (hasAccnos) {  m->mothurOut(accnosFileName); m->mothurOutEndLine();  }
 507                 m->mothurOutEndLine();
 508
 509                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 510
 511                 m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 512
 513                 return 0;
 514
 515         }
 516         catch(exception& e) {
 517                 m->errorOut(e, "ChimeraSeqsCommand", "execute");
 518                 exit(1);
 519         }
 520 }//**********************************************************************************************************************
 521
 522 int ChimeraSeqsCommand::driver(linePair* line, string outputFName, string filename, string accnos){
 523         try {
 524                 ofstream out;
 525                 openOutputFile(outputFName, out);
 526
 527                 ofstream out2;
 528                 openOutputFile(accnos, out2);
 529
 530                 ifstream inFASTA;
 531                 openInputFile(filename, inFASTA);
 532
 533                 inFASTA.seekg(line->start);
 534
 535                 for(int i=0;i<line->numSeqs;i++){
 536
 537                         if (m->control_pressed) {       return 1;       }
 538
 539                         Sequence* candidateSeq = new Sequence(inFASTA);  gobble(inFASTA);
 540
 541                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 542
 543                                 if ((candidateSeq->getAligned().length() != templateSeqsLength) && (method != "chimeracheck")) {  //chimeracheck does not require seqs to be aligned
 544                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 545                                 }else{
 546                                         //find chimeras
 547                                         chimera->getChimeras(candidateSeq);
 548
 549                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 550
 551                                         //print results
 552                                         chimera->print(out, out2);
 553                                 }
 554                         }
 555                         delete candidateSeq;
 556
 557                         //report progress
 558                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1)); m->mothurOutEndLine();           }
 559                 }
 560                 //report progress
 561                 if((line->numSeqs) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(line->numSeqs)); m->mothurOutEndLine();         }
 562
 563                 out.close();
 564                 out2.close();
 565                 inFASTA.close();
 566
 567                 return 0;
 568         }
 569         catch(exception& e) {
 570                 m->errorOut(e, "ChimeraSeqsCommand", "driver");
 571                 exit(1);
 572         }
 573 }
 574
 575 /**************************************************************************************************/
 576
 577 int ChimeraSeqsCommand::createProcesses(string outputFileName, string filename, string accnos) {
 578         try {
 579 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 580                 int process = 0;
 581                 //              processIDS.resize(0);
 582
 583                 //loop through and create all the processes you want
 584                 while (process != processors) {
 585                         int pid = fork();
 586
 587                         if (pid > 0) {
 588                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 589                                 process++;
 590                         }else if (pid == 0){
 591                                 driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp");
 592                                 exit(0);
 593                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 594                 }
 595
 596                 //force parent to wait until all the processes are done
 597                 for (int i=0;i<processors;i++) {
 598                         int temp = processIDS[i];
 599                         wait(&temp);
 600                 }
 601
 602                 return 0;
 603 #endif
 604         }
 605         catch(exception& e) {
 606                 m->errorOut(e, "ChimeraSeqsCommand", "createProcesses");
 607                 exit(1);
 608         }
 609 }
 610
 611 /**************************************************************************************************/
 612
 613 void ChimeraSeqsCommand::appendOutputFiles(string temp, string filename) {
 614         try{
 615
 616                 ofstream output;
 617                 ifstream input;
 618
 619                 openOutputFileAppend(temp, output);
 620                 openInputFile(filename, input, "noerror");
 621
 622                 while(char c = input.get()){
 623                         if(input.eof())         {       break;                  }
 624                         else                            {       output << c;    }
 625                 }
 626
 627                 input.close();
 628                 output.close();
 629         }
 630         catch(exception& e) {
 631                 m->errorOut(e, "ChimeraSeqsCommand", "appendOuputFiles");
 632                 exit(1);
 633         }
 634 }
 635 //**********************************************************************************************************************
 636
 637