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                 abort = true; calledHelp = true;
  31                 vector<string> tempOutNames;
  32                 outputTypes["chimera"] = tempOutNames;
  33                 outputTypes["accnos"] = tempOutNames;
  34                 outputTypes["fasta"] = tempOutNames;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 vector<string> ChimeraSlayerCommand::getRequiredParameters(){
  43         try {
  44                 string AlignArray[] =  {"template","fasta"};
  45                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  46                 return myArray;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "ChimeraSlayerCommand", "getRequiredParameters");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 vector<string> ChimeraSlayerCommand::getRequiredFiles(){
  55         try {
  56                 vector<string> myArray;
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "ChimeraSlayerCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 //***************************************************************************************************************
  65 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
  66         try {
  67                 abort = false; calledHelp = false;
  68
  69                 //allow user to run help
  70                 if(option == "help") { help(); abort = true; calledHelp = true; }
  71
  72                 else {
  73                         //valid paramters for this command
  74                         string Array[] =  {"fasta", "processors","name", "include","trim", "window", "template","numwanted", "ksize", "match","mismatch",
  75                         "divergence", "minsim","mincov","minbs", "minsnp","parents", "iters","outputdir","inputdir", "search","realign" };
  76                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  77
  78                         OptionParser parser(option);
  79                         map<string,string> parameters = parser.getParameters();
  80
  81                         ValidParameters validParameter("chimera.slayer");
  82                         map<string,string>::iterator it;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         vector<string> tempOutNames;
  90                         outputTypes["chimera"] = tempOutNames;
  91                         outputTypes["accnos"] = tempOutNames;
  92                         outputTypes["fasta"] = tempOutNames;
  93
  94                         //if the user changes the input directory command factory will send this info to us in the output parameter
  95                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  96                         if (inputDir == "not found"){   inputDir = "";          }
  97
  98                         //check for required parameters
  99                         fastafile = validParameter.validFile(parameters, "fasta", false);
 100                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("[ERROR]: fasta is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 101                         else {
 102                                 m->splitAtDash(fastafile, fastaFileNames);
 103
 104                                 //go through files and make sure they are good, if not, then disregard them
 105                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 106                                         if (inputDir != "") {
 107                                                 string path = m->hasPath(fastaFileNames[i]);
 108                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 109                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 110                                         }
 111
 112                                         int ableToOpen;
 113                                         ifstream in;
 114
 115                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 116
 117                                         //if you can't open it, try default location
 118                                         if (ableToOpen == 1) {
 119                                                 if (m->getDefaultPath() != "") { //default path is set
 120                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 121                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 122                                                         ifstream in2;
 123                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 124                                                         in2.close();
 125                                                         fastaFileNames[i] = tryPath;
 126                                                 }
 127                                         }
 128
 129                                         if (ableToOpen == 1) {
 130                                                 if (m->getOutputDir() != "") { //default path is set
 131                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 132                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 133                                                         ifstream in2;
 134                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 135                                                         in2.close();
 136                                                         fastaFileNames[i] = tryPath;
 137                                                 }
 138                                         }
 139
 140                                         in.close();
 141
 142                                         if (ableToOpen == 1) {
 143                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 144                                                 //erase from file list
 145                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 146                                                 i--;
 147                                         }
 148                                 }
 149
 150                                 //make sure there is at least one valid file left
 151                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 152                         }
 153
 154
 155                         //check for required parameters
 156                         bool hasName = true;
 157                         namefile = validParameter.validFile(parameters, "name", false);
 158                         if (namefile == "not found") { namefile = "";  hasName = false; }
 159                         else {
 160                                 m->splitAtDash(namefile, nameFileNames);
 161
 162                                 //go through files and make sure they are good, if not, then disregard them
 163                                 for (int i = 0; i < nameFileNames.size(); i++) {
 164                                         if (inputDir != "") {
 165                                                 string path = m->hasPath(nameFileNames[i]);
 166                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 167                                                 if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 168                                         }
 169
 170                                         int ableToOpen;
 171                                         ifstream in;
 172
 173                                         ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 174
 175                                         //if you can't open it, try default location
 176                                         if (ableToOpen == 1) {
 177                                                 if (m->getDefaultPath() != "") { //default path is set
 178                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 179                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 180                                                         ifstream in2;
 181                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 182                                                         in2.close();
 183                                                         nameFileNames[i] = tryPath;
 184                                                 }
 185                                         }
 186
 187                                         if (ableToOpen == 1) {
 188                                                 if (m->getOutputDir() != "") { //default path is set
 189                                                         string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 190                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 191                                                         ifstream in2;
 192                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 193                                                         in2.close();
 194                                                         nameFileNames[i] = tryPath;
 195                                                 }
 196                                         }
 197
 198                                         in.close();
 199
 200                                         if (ableToOpen == 1) {
 201                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 202                                                 //erase from file list
 203                                                 nameFileNames.erase(nameFileNames.begin()+i);
 204                                                 i--;
 205                                         }
 206                                 }
 207
 208                                 //make sure there is at least one valid file left
 209                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 210                         }
 211
 212                         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; }
 213
 214                         //if the user changes the output directory command factory will send this info to us in the output parameter
 215                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 216
 217
 218                         string path;
 219                         it = parameters.find("template");
 220                         //user has given a template file
 221                         if(it != parameters.end()){
 222                                 if (it->second == "self") { templatefile = "self"; }
 223                                 else {
 224                                         path = m->hasPath(it->second);
 225                                         //if the user has not given a path then, add inputdir. else leave path alone.
 226                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
 227
 228                                         templatefile = validParameter.validFile(parameters, "template", true);
 229                                         if (templatefile == "not open") { abort = true; }
 230                                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 231                                 }
 232                         }
 233
 234                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 235                         convert(temp, processors);
 236
 237                         includeAbunds = validParameter.validFile(parameters, "include", false);         if (includeAbunds == "not found") { includeAbunds = "greater"; }
 238                         if ((includeAbunds != "greater") && (includeAbunds != "greaterequal") && (includeAbunds != "all")) { includeAbunds = "greater"; m->mothurOut("Invalid include setting. options are greater, greaterequal or all. using greater."); m->mothurOutEndLine(); }
 239
 240                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 241                         convert(temp, ksize);
 242
 243                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 244                         convert(temp, window);
 245
 246                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 247                         convert(temp, match);
 248
 249                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 250                         convert(temp, mismatch);
 251
 252                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 253                         convert(temp, divR);
 254
 255                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 256                         convert(temp, minSimilarity);
 257
 258                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 259                         convert(temp, minCoverage);
 260
 261                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 262                         convert(temp, minBS);
 263
 264                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "100"; }
 265                         convert(temp, minSNP);
 266
 267                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 268                         convert(temp, parents);
 269
 270                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "f"; }
 271                         realign = m->isTrue(temp);
 272
 273                         temp = validParameter.validFile(parameters, "trim", false);                             if (temp == "not found") { temp = "f"; }
 274                         trim = m->isTrue(temp);
 275
 276                         //temp = validParameter.validFile(parameters, "trimera", false);                                if (temp == "not found") { temp = "f"; }
 277                         //trimera = m->isTrue(temp);
 278
 279                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "distance"; }
 280
 281                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "100"; }
 282                         convert(temp, iters);
 283
 284                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 285                         convert(temp, increment);
 286
 287                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 288                         convert(temp, numwanted);
 289
 290                         if ((search != "distance") && (search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 291                 }
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 295                 exit(1);
 296         }
 297 }
 298 //**********************************************************************************************************************
 299
 300 void ChimeraSlayerCommand::help(){
 301         try {
 302
 303                 m->mothurOut("The chimera.slayer command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 304                 m->mothurOut("This command was modeled after the chimeraSlayer written by the Broad Institute.\n");
 305                 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,
 306                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 307                 m->mothurOut("The name parameter allows you to provide a name file, if you are using template=self. \n");
 308                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n");
 309                 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");
 310                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 311                 #ifdef USE_MPI
 312                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 313                 #endif
 314                 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 piece, default=F. \n");
 315                 //m->mothurOut("The trimera parameter allows you to check both peices of a chimeric sequence for chimeras, thus looking for trimeras and quadmeras. default=F. \n");
 316                 m->mothurOut("The window parameter allows you to specify the window size for searching for chimeras, default=50. \n");
 317                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n");
 318                 m->mothurOut("The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n");
 319                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n");
 320                 m->mothurOut("The match parameter allows you to reward matched bases in blast search, default is 5. \n");
 321                 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");
 322                 m->mothurOut("The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n");
 323                 m->mothurOut("The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n");
 324                 m->mothurOut("The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=100.\n");
 325                 m->mothurOut("The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n");
 326                 m->mothurOut("The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n");
 327                 m->mothurOut("The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n");
 328                 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");
 329                 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");
 330                 m->mothurOut("The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default false.  \n");
 331                 m->mothurOut("The chimera.slayer command should be in the following format: \n");
 332                 m->mothurOut("chimera.slayer(fasta=yourFastaFile, template=yourTemplate, search=yourSearch) \n");
 333                 m->mothurOut("Example: chimera.slayer(fasta=AD.align, template=core_set_aligned.imputed.fasta, search=kmer) \n");
 334                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 335         }
 336         catch(exception& e) {
 337                 m->errorOut(e, "ChimeraSlayerCommand", "help");
 338                 exit(1);
 339         }
 340 }
 341
 342 //***************************************************************************************************************
 343
 344 ChimeraSlayerCommand::~ChimeraSlayerCommand(){  /*      do nothing      */      }
 345
 346 //***************************************************************************************************************
 347
 348 int ChimeraSlayerCommand::execute(){
 349         try{
 350
 351                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 352
 353                 for (int s = 0; s < fastaFileNames.size(); s++) {
 354
 355                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 356
 357                         int start = time(NULL);
 358
 359                         if (templatefile != "self") { //you want to run slayer with a refernce template
 360                                 chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 361                         }else {
 362                                 if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 363                                         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);
 364                                 }else {
 365
 366                                         m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 367
 368                                         //use unique.seqs to create new name and fastafile
 369                                         string inputString = "fasta=" + fastaFileNames[s];
 370                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 371                                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 372
 373                                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 374                                         uniqueCommand->execute();
 375
 376                                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 377
 378                                         delete uniqueCommand;
 379
 380                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 381
 382                                         string nameFile = filenames["name"][0];
 383                                         fastaFileNames[s] = filenames["fasta"][0];
 384
 385                                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, nameFile, search, includeAbunds, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 386                                 }
 387                         }
 388
 389                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 390                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimera";
 391                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 392                         string trimFastaFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.fasta";
 393
 394                         if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0;    }
 395
 396                         if (chimera->getUnaligned()) {
 397                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 398                                 delete chimera;
 399                                 return 0;
 400                         }
 401                         templateSeqsLength = chimera->getLength();
 402
 403                 #ifdef USE_MPI
 404                         int pid, numSeqsPerProcessor;
 405                                 int tag = 2001;
 406                                 vector<unsigned long int> MPIPos;
 407
 408                                 MPI_Status status;
 409                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 410                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 411
 412                                 MPI_File inMPI;
 413                                 MPI_File outMPI;
 414                                 MPI_File outMPIAccnos;
 415                                 MPI_File outMPIFasta;
 416
 417                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 418                                 int inMode=MPI_MODE_RDONLY;
 419
 420                                 char outFilename[1024];
 421                                 strcpy(outFilename, outputFileName.c_str());
 422
 423                                 char outAccnosFilename[1024];
 424                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 425
 426                                 char outFastaFilename[1024];
 427                                 strcpy(outFastaFilename, trimFastaFileName.c_str());
 428
 429                                 char inFileName[1024];
 430                                 strcpy(inFileName, fastaFileNames[s].c_str());
 431
 432                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 433                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 434                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 435                                 if (trim) { MPI_File_open(MPI_COMM_WORLD, outFastaFilename, outMode, MPI_INFO_NULL, &outMPIFasta); }
 436
 437                         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;  }
 438
 439                                 if (pid == 0) { //you are the root process
 440                                         m->mothurOutEndLine();
 441                                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 442                                         m->mothurOutEndLine();
 443
 444                                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 445
 446                                         //print header
 447                                         int length = outTemp.length();
 448                                         char* buf2 = new char[length];
 449                                         memcpy(buf2, outTemp.c_str(), length);
 450
 451                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 452                                         delete buf2;
 453
 454                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 455
 456                                         //send file positions to all processes
 457                                         for(int i = 1; i < processors; i++) {
 458                                                 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 459                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 460                                         }
 461
 462                                         //figure out how many sequences you have to align
 463                                         numSeqsPerProcessor = numSeqs / processors;
 464                                         int startIndex =  pid * numSeqsPerProcessor;
 465                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 466
 467                                         //do your part
 468                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 469
 470                                         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;  }
 471
 472                                 }else{ //you are a child process
 473                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 474                                         MPIPos.resize(numSeqs+1);
 475                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 476
 477                                         //figure out how many sequences you have to align
 478                                         numSeqsPerProcessor = numSeqs / processors;
 479                                         int startIndex =  pid * numSeqsPerProcessor;
 480                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 481
 482                                         //do your part
 483                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 484
 485                                         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;  }
 486                                 }
 487
 488                                 //close files
 489                                 MPI_File_close(&inMPI);
 490                                 MPI_File_close(&outMPI);
 491                                 MPI_File_close(&outMPIAccnos);
 492                                 if (trim) { MPI_File_close(&outMPIFasta); }
 493                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 494
 495                 #else
 496                         ofstream outHeader;
 497                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras.tempHeader";
 498                         m->openOutputFile(tempHeader, outHeader);
 499
 500                         chimera->printHeader(outHeader);
 501                         outHeader.close();
 502
 503                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 504
 505                         for (int i = 0; i < (positions.size()-1); i++) {
 506                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 507                         }
 508
 509                         //break up file
 510                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 511                                 if(processors == 1){
 512                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 513
 514                                         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; }
 515
 516                                 }else{
 517                                         processIDS.resize(0);
 518
 519                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 520
 521                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 522                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 523                                         if (trim) {  rename((trimFastaFileName + toString(processIDS[0]) + ".temp").c_str(), trimFastaFileName.c_str()); }
 524
 525                                         //append output files
 526                                         for(int i=1;i<processors;i++){
 527                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 528                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 529                                         }
 530
 531                                         //append output files
 532                                         for(int i=1;i<processors;i++){
 533                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 534                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 535                                         }
 536
 537                                         if (trim) {
 538                                                 for(int i=1;i<processors;i++){
 539                                                         m->appendFiles((trimFastaFileName + toString(processIDS[i]) + ".temp"), trimFastaFileName);
 540                                                         remove((trimFastaFileName + toString(processIDS[i]) + ".temp").c_str());
 541                                                 }
 542                                         }
 543
 544                                         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; }
 545                                 }
 546
 547                         #else
 548                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 549
 550                                 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; }
 551
 552                         #endif
 553
 554                         m->appendFiles(outputFileName, tempHeader);
 555
 556                         remove(outputFileName.c_str());
 557                         rename(tempHeader.c_str(), outputFileName.c_str());
 558
 559                 #endif
 560                         delete chimera;
 561
 562
 563                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 564
 565                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 566                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 567                         if (trim) {  outputNames.push_back(trimFastaFileName); outputTypes["fasta"].push_back(trimFastaFileName); }
 568
 569                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 570                 }
 571
 572                 m->mothurOutEndLine();
 573                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 574                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 575                 m->mothurOutEndLine();
 576
 577                 return 0;
 578
 579         }
 580         catch(exception& e) {
 581                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 582                 exit(1);
 583         }
 584 }
 585 //**********************************************************************************************************************
 586
 587 int ChimeraSlayerCommand::driver(linePair* filePos, string outputFName, string filename, string accnos, string fasta){
 588         try {
 589                 ofstream out;
 590                 m->openOutputFile(outputFName, out);
 591
 592                 ofstream out2;
 593                 m->openOutputFile(accnos, out2);
 594
 595                 ofstream out3;
 596                 if (trim) {  m->openOutputFile(fasta, out3); }
 597
 598                 ifstream inFASTA;
 599                 m->openInputFile(filename, inFASTA);
 600
 601                 inFASTA.seekg(filePos->start);
 602
 603                 bool done = false;
 604                 int count = 0;
 605
 606                 while (!done) {
 607
 608                         if (m->control_pressed) {       out.close(); out2.close(); if (trim) { out3.close(); } inFASTA.close(); return 1;       }
 609
 610                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 611
 612                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 613
 614                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 615                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 616                                 }else{
 617                                         //find chimeras
 618                                         chimera->getChimeras(candidateSeq);
 619
 620                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 621
 622                                         //print results
 623                                         Sequence* trimmed = chimera->print(out, out2);
 624
 625                                         if (trim) { trimmed->printSequence(out3); delete trimmed; }
 626
 627                                         //do you want to check both pieces for chimeras
 628                                         //if (trimera) {}
 629                                 }
 630                         count++;
 631                         }
 632                         delete candidateSeq;
 633
 634                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 635                                 unsigned long int pos = inFASTA.tellg();
 636                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 637                         #else
 638                                 if (inFASTA.eof()) { break; }
 639                         #endif
 640
 641                         //report progress
 642                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 643                 }
 644                 //report progress
 645                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 646
 647                 out.close();
 648                 out2.close();
 649                 if (trim) { out3.close(); }
 650                 inFASTA.close();
 651
 652                 return count;
 653         }
 654         catch(exception& e) {
 655                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 656                 exit(1);
 657         }
 658 }
 659 //**********************************************************************************************************************
 660 #ifdef USE_MPI
 661 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, MPI_File& outFastaMPI, vector<unsigned long int>& MPIPos){
 662         try {
 663                 MPI_Status status;
 664                 int pid;
 665                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 666
 667                 for(int i=0;i<num;i++){
 668
 669                         if (m->control_pressed) {       return 1;       }
 670
 671                         //read next sequence
 672                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 673
 674                         char* buf4 = new char[length];
 675                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 676
 677                         string tempBuf = buf4;
 678                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 679                         istringstream iss (tempBuf,istringstream::in);
 680
 681                         delete buf4;
 682
 683                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 684
 685                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 686
 687                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 688                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 689                                 }else{
 690
 691                                         //find chimeras
 692                                         chimera->getChimeras(candidateSeq);
 693
 694                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 695
 696                                         //print results
 697                                         Sequence* trimmed = chimera->print(outMPI, outAccMPI);
 698
 699                                         if (trim) {
 700                                                 string outputString = ">" + trimmed->getName() + "\n" + trimmed->getAligned() + "\n";
 701                                                 delete trimmed;
 702
 703                                                 //write to accnos file
 704                                                 int length = outputString.length();
 705                                                 char* buf2 = new char[length];
 706                                                 memcpy(buf2, outputString.c_str(), length);
 707
 708                                                 MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
 709                                                 delete buf2;
 710                                         }
 711
 712                                         //do you want to check both pieces for chimeras
 713                                         //if (trimera) {}
 714
 715                                 }
 716                         }
 717                         delete candidateSeq;
 718
 719                         //report progress
 720                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 721                 }
 722                 //report progress
 723                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 724
 725
 726                 return 0;
 727         }
 728         catch(exception& e) {
 729                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 730                 exit(1);
 731         }
 732 }
 733 #endif
 734
 735 /**************************************************************************************************/
 736
 737 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos, string fasta) {
 738         try {
 739 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 740                 int process = 0;
 741                 int num = 0;
 742
 743                 //loop through and create all the processes you want
 744                 while (process != processors) {
 745                         int pid = fork();
 746
 747                         if (pid > 0) {
 748                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 749                                 process++;
 750                         }else if (pid == 0){
 751                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp", fasta + toString(getpid()) + ".temp");
 752
 753                                 //pass numSeqs to parent
 754                                 ofstream out;
 755                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 756                                 m->openOutputFile(tempFile, out);
 757                                 out << num << endl;
 758                                 out.close();
 759
 760                                 exit(0);
 761                         }else {
 762                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 763                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 764                                 exit(0);
 765                         }
 766                 }
 767
 768                 //force parent to wait until all the processes are done
 769                 for (int i=0;i<processors;i++) {
 770                         int temp = processIDS[i];
 771                         wait(&temp);
 772                 }
 773
 774                 for (int i = 0; i < processIDS.size(); i++) {
 775                         ifstream in;
 776                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 777                         m->openInputFile(tempFile, in);
 778                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 779                         in.close(); remove(tempFile.c_str());
 780                 }
 781
 782                 return num;
 783 #endif
 784         }
 785         catch(exception& e) {
 786                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
 787                 exit(1);
 788         }
 789 }
 790
 791 /**************************************************************************************************/
 792
 793