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