chimeraslayercommand.cpp

   1 /*
   2  *  chimeraslayercommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/31/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeraslayercommand.h"
  11 #include "chimeraslayer.h"
  12 #include "deconvolutecommand.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> ChimeraSlayerCommand::getValidParameters(){
  16         try {
  17                 string AlignArray[] =  {"fasta", "processors","trim", "name","window", "include","template","numwanted", "ksize", "match","mismatch",
  18                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  19                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  20                 return myArray;
  21         }
  22         catch(exception& e) {
  23                 m->errorOut(e, "ChimeraSlayerCommand", "getValidParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 ChimeraSlayerCommand::ChimeraSlayerCommand(){
  29         try {
  30                 vector<string> tempOutNames;
  31                 outputTypes["chimera"] = tempOutNames;
  32                 outputTypes["accnos"] = tempOutNames;
  33                 outputTypes["fasta"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> ChimeraSlayerCommand::getRequiredParameters(){
  42         try {
  43                 string AlignArray[] =  {"template","fasta"};
  44                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "ChimeraSlayerCommand", "getRequiredParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 vector<string> ChimeraSlayerCommand::getRequiredFiles(){
  54         try {
  55                 vector<string> myArray;
  56                 return myArray;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "ChimeraSlayerCommand", "getRequiredFiles");
  60                 exit(1);
  61         }
  62 }
  63 //***************************************************************************************************************
  64 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
  65         try {
  66                 abort = false;
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string Array[] =  {"fasta", "processors","name", "include","trim", "window", "template","numwanted", "ksize", "match","mismatch",
  74                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  75                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  76
  77                         OptionParser parser(option);
  78                         map<string,string> parameters = parser.getParameters();
  79
  80                         ValidParameters validParameter("chimera.slayer");
  81                         map<string,string>::iterator it;
  82
  83                         //check to make sure all parameters are valid for command
  84                         for (it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  86                         }
  87
  88                         vector<string> tempOutNames;
  89                         outputTypes["chimera"] = tempOutNames;
  90                         outputTypes["accnos"] = tempOutNames;
  91                         outputTypes["fasta"] = tempOutNames;
  92
  93                         //if the user changes the input directory command factory will send this info to us in the output parameter
  94                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  95                         if (inputDir == "not found"){   inputDir = "";          }
  96
  97                         //check for required parameters
  98                         fastafile = validParameter.validFile(parameters, "fasta", false);
  99                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("[ERROR]: fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 100                         else {
 101                                 m->splitAtDash(fastafile, fastaFileNames);
 102
 103                                 //go through files and make sure they are good, if not, then disregard them
 104                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 105                                         if (inputDir != "") {
 106                                                 string path = m->hasPath(fastaFileNames[i]);
 107                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 108                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 109                                         }
 110
 111                                         int ableToOpen;
 112                                         ifstream in;
 113
 114                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 115
 116                                         //if you can't open it, try default location
 117                                         if (ableToOpen == 1) {
 118                                                 if (m->getDefaultPath() != "") { //default path is set
 119                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 120                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 121                                                         ifstream in2;
 122                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 123                                                         in2.close();
 124                                                         fastaFileNames[i] = tryPath;
 125                                                 }
 126                                         }
 127
 128                                         if (ableToOpen == 1) {
 129                                                 if (m->getOutputDir() != "") { //default path is set
 130                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 131                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 132                                                         ifstream in2;
 133                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 134                                                         in2.close();
 135                                                         fastaFileNames[i] = tryPath;
 136                                                 }
 137                                         }
 138
 139                                         in.close();
 140
 141                                         if (ableToOpen == 1) {
 142                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 143                                                 //erase from file list
 144                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 145                                                 i--;
 146                                         }
 147                                 }
 148
 149                                 //make sure there is at least one valid file left
 150                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 151                         }
 152
 153
 154                         //check for required parameters
 155                         bool hasName = true;
 156                         namefile = validParameter.validFile(parameters, "name", false);
 157                         if (namefile == "not found") { namefile = "";  hasName = false; }
 158                         else {
 159                                 m->splitAtDash(namefile, nameFileNames);
 160
 161                                 //go through files and make sure they are good, if not, then disregard them
 162                                 for (int i = 0; i < nameFileNames.size(); i++) {
 163                                         if (inputDir != "") {
 164                                                 string path = m->hasPath(nameFileNames[i]);
 165                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 166                                                 if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 167                                         }
 168
 169                                         int ableToOpen;
 170                                         ifstream in;
 171
 172                                         ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 173
 174                                         //if you can't open it, try default location
 175                                         if (ableToOpen == 1) {
 176                                                 if (m->getDefaultPath() != "") { //default path is set
 177                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 178                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 179                                                         ifstream in2;
 180                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 181                                                         in2.close();
 182                                                         nameFileNames[i] = tryPath;
 183                                                 }
 184                                         }
 185
 186                                         if (ableToOpen == 1) {
 187                                                 if (m->getOutputDir() != "") { //default path is set
 188                                                         string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 189                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 190                                                         ifstream in2;
 191                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 192                                                         in2.close();
 193                                                         nameFileNames[i] = tryPath;
 194                                                 }
 195                                         }
 196
 197                                         in.close();
 198
 199                                         if (ableToOpen == 1) {
 200                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 201                                                 //erase from file list
 202                                                 nameFileNames.erase(nameFileNames.begin()+i);
 203                                                 i--;
 204                                         }
 205                                 }
 206
 207                                 //make sure there is at least one valid file left
 208                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 209                         }
 210
 211                         if (hasName && (nameFileNames.size() != fastaFileNames.size())) { m->mothurOut("[ERROR]: The number of namefiles does not match the number of fastafiles, please correct."); m->mothurOutEndLine(); abort=true; }
 212
 213                         //if the user changes the output directory command factory will send this info to us in the output parameter
 214                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 215
 216
 217                         string path;
 218                         it = parameters.find("template");
 219                         //user has given a template file
 220                         if(it != parameters.end()){
 221                                 if (it->second == "self") { templatefile = "self"; }
 222                                 else {
 223                                         path = m->hasPath(it->second);
 224                                         //if the user has not given a path then, add inputdir. else leave path alone.
 225                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
 226
 227                                         templatefile = validParameter.validFile(parameters, "template", true);
 228                                         if (templatefile == "not open") { abort = true; }
 229                                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 230                                 }
 231                         }
 232
 233                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 234                         convert(temp, processors);
 235
 236                         includeAbunds = validParameter.validFile(parameters, "include", false);         if (includeAbunds == "not found") { includeAbunds = "greater"; }
 237                         if ((includeAbunds != "greater") && (includeAbunds != "greaterequal") && (includeAbunds != "all")) { includeAbunds = "greater"; m->mothurOut("Invalid include setting. options are greater, greaterequal or all. using greater."); m->mothurOutEndLine(); }
 238
 239                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 240                         convert(temp, ksize);
 241
 242                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 243                         convert(temp, window);
 244
 245                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 246                         convert(temp, match);
 247
 248                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 249                         convert(temp, mismatch);
 250
 251                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 252                         convert(temp, divR);
 253
 254                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 255                         convert(temp, minSimilarity);
 256
 257                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 258                         convert(temp, minCoverage);
 259
 260                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 261                         convert(temp, minBS);
 262
 263                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "100"; }
 264                         convert(temp, minSNP);
 265
 266                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 267                         convert(temp, parents);
 268
 269                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "f"; }
 270                         realign = m->isTrue(temp);
 271
 272                         temp = validParameter.validFile(parameters, "trim", false);                             if (temp == "not found") { temp = "f"; }
 273                         trim = m->isTrue(temp);
 274
 275                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "distance"; }
 276
 277                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "100"; }
 278                         convert(temp, iters);
 279
 280                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 281                         convert(temp, increment);
 282
 283                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 284                         convert(temp, numwanted);
 285
 286                         if ((search != "distance") && (search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 287                 }
 288         }
 289         catch(exception& e) {
 290                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 291                 exit(1);
 292         }
 293 }
 294 //**********************************************************************************************************************
 295
 296 void ChimeraSlayerCommand::help(){
 297         try {
 298
 299                 m->mothurOut("The chimera.slayer command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 300                 m->mothurOut("This command was modeled after the chimeraSlayer written by the Broad Institute.\n");
 301                 m->mothurOut("The chimera.slayer command parameters are fasta, name, template, processors, trim, ksize, window, match, mismatch, divergence. minsim, mincov, minbs, minsnp, parents, search, iters, increment and numwanted.\n"); //realign,
 302                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 303                 m->mothurOut("The name parameter allows you to provide a name file, if you are using template=self. \n");
 304                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n");
 305                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. You may also set template=self, in this case the abundant sequences will be used as potential parents. \n");
 306                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 307                 #ifdef USE_MPI
 308                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 309                 #endif
 310                 m->mothurOut("The trim parameter allows you to output a new fasta file containing your sequences with the chimeric ones trimmed to include only their longest peice, default=F. \n");
 311                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=50. \n");
 312                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n");
 313                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n");
 314                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n");
 315                 m->mothurOut("The match parameter allows you to reward matched bases in blast search, default is 5. \n");
 316                 m->mothurOut("The parents parameter allows you to select the number of potential parents to investigate from the numwanted best matches after rating them, default is 3. \n");
 317                 m->mothurOut("The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n");
 318                 m->mothurOut("The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n");
 319                 m->mothurOut("The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=100.\n");
 320                 m->mothurOut("The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n");
 321                 m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
 322                 m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
 323                 m->mothurOut("The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 100) \n");
 324                 m->mothurOut("The search parameter allows you to specify search method for finding the closest parent. Choices are distance, blast, and kmer, default distance. \n");
 325                 m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false.  \n");
 326                 m->mothurOut("The chimera.slayer command should be in the following format: \n");
 327                 m->mothurOut("chimera.slayer(fasta=yourFastaFile, template=yourTemplate, search=yourSearch) \n");
 328                 m->mothurOut("Example: chimera.slayer(fasta=AD.align, template=core_set_aligned.imputed.fasta, search=kmer) \n");
 329                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 330         }
 331         catch(exception& e) {
 332                 m->errorOut(e, "ChimeraSlayerCommand", "help");
 333                 exit(1);
 334         }
 335 }
 336
 337 //***************************************************************************************************************
 338
 339 ChimeraSlayerCommand::~ChimeraSlayerCommand(){  /*      do nothing      */      }
 340
 341 //***************************************************************************************************************
 342
 343 int ChimeraSlayerCommand::execute(){
 344         try{
 345
 346                 if (abort == true) { return 0; }
 347
 348                 for (int s = 0; s < fastaFileNames.size(); s++) {
 349
 350                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 351
 352                         int start = time(NULL);
 353
 354                         if (templatefile != "self") { //you want to run slayer with a refernce template
 355                                 chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 356                         }else {
 357                                 if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 358                                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, nameFileNames[s], search, includeAbunds, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 359                                 }else {
 360
 361                                         m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 362
 363                                         //use unique.seqs to create new name and fastafile
 364                                         string inputString = "fasta=" + fastaFileNames[s];
 365                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 366                                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 367
 368                                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 369                                         uniqueCommand->execute();
 370
 371                                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 372
 373                                         delete uniqueCommand;
 374
 375                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 376
 377                                         string nameFile = filenames["name"][0];
 378                                         fastaFileNames[s] = filenames["fasta"][0];
 379
 380                                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, nameFile, search, includeAbunds, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 381                                 }
 382                         }
 383
 384                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 385                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimera";
 386                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 387                         string trimFastaFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.fasta";
 388
 389                         if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0;    }
 390
 391                         if (chimera->getUnaligned()) {
 392                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 393                                 delete chimera;
 394                                 return 0;
 395                         }
 396                         templateSeqsLength = chimera->getLength();
 397
 398                 #ifdef USE_MPI
 399                         int pid, numSeqsPerProcessor;
 400                                 int tag = 2001;
 401                                 vector<unsigned long int> MPIPos;
 402
 403                                 MPI_Status status;
 404                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 405                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 406
 407                                 MPI_File inMPI;
 408                                 MPI_File outMPI;
 409                                 MPI_File outMPIAccnos;
 410                                 MPI_File outMPIFasta;
 411
 412                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 413                                 int inMode=MPI_MODE_RDONLY;
 414
 415                                 char outFilename[1024];
 416                                 strcpy(outFilename, outputFileName.c_str());
 417
 418                                 char outAccnosFilename[1024];
 419                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 420
 421                                 char outFastaFilename[1024];
 422                                 strcpy(outFastaFilename, trimFastaFileName.c_str());
 423
 424                                 char inFileName[1024];
 425                                 strcpy(inFileName, fastaFileNames[s].c_str());
 426
 427                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 428                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 429                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 430                                 if (trim) { MPI_File_open(MPI_COMM_WORLD, outFastaFilename, outMode, MPI_INFO_NULL, &outMPIFasta); }
 431
 432                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) {  MPI_File_close(&outMPIFasta);  } MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }   delete chimera; return 0;  }
 433
 434                                 if (pid == 0) { //you are the root process
 435                                         m->mothurOutEndLine();
 436                                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 437                                         m->mothurOutEndLine();
 438
 439                                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 440
 441                                         //print header
 442                                         int length = outTemp.length();
 443                                         char* buf2 = new char[length];
 444                                         memcpy(buf2, outTemp.c_str(), length);
 445
 446                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 447                                         delete buf2;
 448
 449                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 450
 451                                         //send file positions to all processes
 452                                         for(int i = 1; i < processors; i++) {
 453                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 454                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 455                                         }
 456
 457                                         //figure out how many sequences you have to align
 458                                         numSeqsPerProcessor = numSeqs / processors;
 459                                         int startIndex =  pid * numSeqsPerProcessor;
 460                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 461
 462                                         //do your part
 463                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 464
 465                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  remove(outputFileName.c_str());  remove(accnosFileName.c_str());  delete chimera; return 0;  }
 466
 467                                 }else{ //you are a child process
 468                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 469                                         MPIPos.resize(numSeqs+1);
 470                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 471
 472                                         //figure out how many sequences you have to align
 473                                         numSeqsPerProcessor = numSeqs / processors;
 474                                         int startIndex =  pid * numSeqsPerProcessor;
 475                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 476
 477                                         //do your part
 478                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 479
 480                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 481                                 }
 482
 483                                 //close files
 484                                 MPI_File_close(&inMPI);
 485                                 MPI_File_close(&outMPI);
 486                                 MPI_File_close(&outMPIAccnos);
 487                                 if (trim) { MPI_File_close(&outMPIFasta); }
 488                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 489
 490                 #else
 491                         ofstream outHeader;
 492                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras.tempHeader";
 493                         m->openOutputFile(tempHeader, outHeader);
 494
 495                         chimera->printHeader(outHeader);
 496                         outHeader.close();
 497
 498                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 499
 500                         for (int i = 0; i < (positions.size()-1); i++) {
 501                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 502                         }
 503
 504                         //break up file
 505                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 506                                 if(processors == 1){
 507                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 508
 509                                         if (m->control_pressed) { outputTypes.clear(); if (trim) { remove(trimFastaFileName.c_str()); } remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {      remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 510
 511                                 }else{
 512                                         processIDS.resize(0);
 513
 514                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 515
 516                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 517                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 518                                         if (trim) {  rename((trimFastaFileName + toString(processIDS[0]) + ".temp").c_str(), trimFastaFileName.c_str()); }
 519
 520                                         //append output files
 521                                         for(int i=1;i<processors;i++){
 522                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 523                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 524                                         }
 525
 526                                         //append output files
 527                                         for(int i=1;i<processors;i++){
 528                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 529                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 530                                         }
 531
 532                                         if (trim) {
 533                                                 for(int i=1;i<processors;i++){
 534                                                         m->appendFiles((trimFastaFileName + toString(processIDS[i]) + ".temp"), trimFastaFileName);
 535                                                         remove((trimFastaFileName + toString(processIDS[i]) + ".temp").c_str());
 536                                                 }
 537                                         }
 538
 539                                         if (m->control_pressed) { outputTypes.clear(); if (trim) { remove(trimFastaFileName.c_str()); } remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 540                                 }
 541
 542                         #else
 543                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 544
 545                                 if (m->control_pressed) { outputTypes.clear(); if (trim) { remove(trimFastaFileName.c_str()); } remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {      remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 546
 547                         #endif
 548
 549                         m->appendFiles(outputFileName, tempHeader);
 550
 551                         remove(outputFileName.c_str());
 552                         rename(tempHeader.c_str(), outputFileName.c_str());
 553
 554                 #endif
 555                         delete chimera;
 556
 557
 558                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 559
 560                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 561                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 562                         if (trim) {  outputNames.push_back(trimFastaFileName); outputTypes["fasta"].push_back(trimFastaFileName); }
 563
 564                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 565                 }
 566
 567                 m->mothurOutEndLine();
 568                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 569                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 570                 m->mothurOutEndLine();
 571
 572                 return 0;
 573
 574         }
 575         catch(exception& e) {
 576                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 577                 exit(1);
 578         }
 579 }
 580 //**********************************************************************************************************************
 581
 582 int ChimeraSlayerCommand::driver(linePair* filePos, string outputFName, string filename, string accnos, string fasta){
 583         try {
 584                 ofstream out;
 585                 m->openOutputFile(outputFName, out);
 586
 587                 ofstream out2;
 588                 m->openOutputFile(accnos, out2);
 589
 590                 ofstream out3;
 591                 if (trim) {  m->openOutputFile(fasta, out3); }
 592
 593                 ifstream inFASTA;
 594                 m->openInputFile(filename, inFASTA);
 595
 596                 inFASTA.seekg(filePos->start);
 597
 598                 bool done = false;
 599                 int count = 0;
 600
 601                 while (!done) {
 602
 603                         if (m->control_pressed) {       out.close(); out2.close(); if (trim) { out3.close(); } inFASTA.close(); return 1;       }
 604
 605                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 606
 607                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 608
 609                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 610                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 611                                 }else{
 612                                         //find chimeras
 613                                         chimera->getChimeras(candidateSeq);
 614
 615                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 616
 617                                         //print results
 618                                         Sequence* trimmed = chimera->print(out, out2);
 619
 620                                         if (trim) { trimmed->printSequence(out3); delete trimmed; }
 621                                 }
 622                         count++;
 623                         }
 624                         delete candidateSeq;
 625
 626                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 627                                 unsigned long int pos = inFASTA.tellg();
 628                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 629                         #else
 630                                 if (inFASTA.eof()) { break; }
 631                         #endif
 632
 633                         //report progress
 634                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 635                 }
 636                 //report progress
 637                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 638
 639                 out.close();
 640                 out2.close();
 641                 if (trim) { out3.close(); }
 642                 inFASTA.close();
 643
 644                 return count;
 645         }
 646         catch(exception& e) {
 647                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 648                 exit(1);
 649         }
 650 }
 651 //**********************************************************************************************************************
 652 #ifdef USE_MPI
 653 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, MPI_File& outFastaMPI, vector<unsigned long int>& MPIPos){
 654         try {
 655                 MPI_Status status;
 656                 int pid;
 657                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 658
 659                 for(int i=0;i<num;i++){
 660
 661                         if (m->control_pressed) {       return 1;       }
 662
 663                         //read next sequence
 664                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 665
 666                         char* buf4 = new char[length];
 667                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 668
 669                         string tempBuf = buf4;
 670                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 671                         istringstream iss (tempBuf,istringstream::in);
 672
 673                         delete buf4;
 674
 675                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 676
 677                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 678
 679                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 680                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 681                                 }else{
 682
 683                                         //find chimeras
 684                                         chimera->getChimeras(candidateSeq);
 685
 686                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 687
 688                                         //print results
 689                                         Sequence* trimmed = chimera->print(outMPI, outAccMPI);
 690
 691                                         if (trim) {
 692                                                 string outputString = ">" + trimmed->getName() + "\n" + trimmed->getAligned() + "\n";
 693                                                 delete trimmed;
 694
 695                                                 //write to accnos file
 696                                                 int length = outputString.length();
 697                                                 char* buf2 = new char[length];
 698                                                 memcpy(buf2, outputString.c_str(), length);
 699
 700                                                 MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
 701                                                 delete buf2;
 702                                         }
 703
 704                                 }
 705                         }
 706                         delete candidateSeq;
 707
 708                         //report progress
 709                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 710                 }
 711                 //report progress
 712                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 713
 714
 715                 return 0;
 716         }
 717         catch(exception& e) {
 718                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 719                 exit(1);
 720         }
 721 }
 722 #endif
 723
 724 /**************************************************************************************************/
 725
 726 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos, string fasta) {
 727         try {
 728 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 729                 int process = 0;
 730                 int num = 0;
 731
 732                 //loop through and create all the processes you want
 733                 while (process != processors) {
 734                         int pid = fork();
 735
 736                         if (pid > 0) {
 737                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 738                                 process++;
 739                         }else if (pid == 0){
 740                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp", fasta + toString(getpid()) + ".temp");
 741
 742                                 //pass numSeqs to parent
 743                                 ofstream out;
 744                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 745                                 m->openOutputFile(tempFile, out);
 746                                 out << num << endl;
 747                                 out.close();
 748
 749                                 exit(0);
 750                         }else {
 751                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 752                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 753                                 exit(0);
 754                         }
 755                 }
 756
 757                 //force parent to wait until all the processes are done
 758                 for (int i=0;i<processors;i++) {
 759                         int temp = processIDS[i];
 760                         wait(&temp);
 761                 }
 762
 763                 for (int i = 0; i < processIDS.size(); i++) {
 764                         ifstream in;
 765                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 766                         m->openInputFile(tempFile, in);
 767                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 768                         in.close(); remove(tempFile.c_str());
 769                 }
 770
 771                 return num;
 772 #endif
 773         }
 774         catch(exception& e) {
 775                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
 776                 exit(1);
 777         }
 778 }
 779
 780 /**************************************************************************************************/
 781
 782