chimeracheckcommand.cpp

   1 /*
   2  *  chimeracheckcommand.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 "chimeracheckcommand.h"
  11
  12 //***************************************************************************************************************
  13
  14 ChimeraCheckCommand::ChimeraCheckCommand(string option)  {
  15         try {
  16                 abort = false;
  17
  18                 //allow user to run help
  19                 if(option == "help") { help(); abort = true; }
  20
  21                 else {
  22                         //valid paramters for this command
  23                         string Array[] =  {"fasta","processors","increment","template","ksize","svg", "name","outputdir","inputdir" };
  24                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  25
  26                         OptionParser parser(option);
  27                         map<string,string> parameters = parser.getParameters();
  28
  29                         ValidParameters validParameter("chimera.check");
  30                         map<string,string>::iterator it;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //if the user changes the input directory command factory will send this info to us in the output parameter
  38                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  39                         if (inputDir == "not found"){   inputDir = "";          }
  40                         else {
  41                                 it = parameters.find("template");
  42                                 //user has given a template file
  43                                 if(it != parameters.end()){
  44                                         string path = hasPath(it->second);
  45                                         //if the user has not given a path then, add inputdir. else leave path alone.
  46                                         if (path == "") {       parameters["template"] = inputDir + it->second;         }
  47                                 }
  48                         }
  49
  50                         //check for required parameters
  51                         fastafile = validParameter.validFile(parameters, "fasta", false);
  52                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
  53                         else {
  54                                 splitAtDash(fastafile, fastaFileNames);
  55
  56                                 //go through files and make sure they are good, if not, then disregard them
  57                                 for (int i = 0; i < fastaFileNames.size(); i++) {
  58                                         if (inputDir != "") {
  59                                                 string path = hasPath(fastaFileNames[i]);
  60                                                 //if the user has not given a path then, add inputdir. else leave path alone.
  61                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
  62                                         }
  63
  64                                         int ableToOpen;
  65                                         ifstream in;
  66
  67                                         ableToOpen = openInputFile(fastaFileNames[i], in, "noerror");
  68
  69                                         //if you can't open it, try default location
  70                                         if (ableToOpen == 1) {
  71                                                 if (m->getDefaultPath() != "") { //default path is set
  72                                                         string tryPath = m->getDefaultPath() + getSimpleName(fastaFileNames[i]);
  73                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
  74                                                         ableToOpen = openInputFile(tryPath, in, "noerror");
  75                                                         fastaFileNames[i] = tryPath;
  76                                                 }
  77                                         }
  78                                         in.close();
  79
  80                                         if (ableToOpen == 1) {
  81                                                 m->mothurOut("Unable to open " + fastaFileNames[i] +". It will be disregarded."); m->mothurOutEndLine();
  82                                                 //erase from file list
  83                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
  84                                                 i--;
  85                                         }
  86                                 }
  87
  88                                 //make sure there is at least one valid file left
  89                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
  90                         }
  91
  92                         //if the user changes the output directory command factory will send this info to us in the output parameter
  93                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
  94
  95                         templatefile = validParameter.validFile(parameters, "template", true);
  96                         if (templatefile == "not open") { abort = true; }
  97                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("template is a required parameter for the chimera.check command."); m->mothurOutEndLine(); abort = true;  }
  98
  99                         namefile = validParameter.validFile(parameters, "name", false);
 100                         if (namefile == "not found") { namefile = ""; }
 101                         else {
 102                                 splitAtDash(namefile, nameFileNames);
 103
 104                                 //go through files and make sure they are good, if not, then disregard them
 105                                 for (int i = 0; i < nameFileNames.size(); i++) {
 106                                         if (inputDir != "") {
 107                                                 string path = hasPath(nameFileNames[i]);
 108                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 109                                                 if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 110                                         }
 111
 112                                         int ableToOpen;
 113                                         ifstream in;
 114
 115                                         ableToOpen = openInputFile(nameFileNames[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() + getSimpleName(nameFileNames[i]);
 121                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 122                                                         ableToOpen = openInputFile(tryPath, in, "noerror");
 123                                                         nameFileNames[i] = tryPath;
 124                                                 }
 125                                         }
 126                                         in.close();
 127
 128                                         if (ableToOpen == 1) {
 129                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 130                                                 //erase from file list
 131                                                 nameFileNames.erase(nameFileNames.begin()+i);
 132                                                 i--;
 133                                         }
 134
 135                                 }
 136
 137                                 //make sure there is at least one valid file left
 138                                 if (nameFileNames.size() != 0) {
 139                                         if (nameFileNames.size() != fastaFileNames.size()) {
 140                                                  m->mothurOut("Different number of valid name files and fasta files, aborting command."); m->mothurOutEndLine();
 141                                                  abort = true;
 142                                         }
 143                                 }
 144                         }
 145
 146                         string temp = validParameter.validFile(parameters, "processors", false);                if (temp == "not found") { temp = "1"; }
 147                         convert(temp, processors);
 148
 149                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 150                         convert(temp, ksize);
 151
 152                         temp = validParameter.validFile(parameters, "svg", false);                              if (temp == "not found") { temp = "F"; }
 153                         svg = isTrue(temp);
 154                         if (nameFileNames.size() != 0) { svg = true; }
 155
 156                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "10"; }
 157                         convert(temp, increment);
 158                 }
 159         }
 160         catch(exception& e) {
 161                 m->errorOut(e, "ChimeraCheckCommand", "ChimeraCheckCommand");
 162                 exit(1);
 163         }
 164 }
 165 //**********************************************************************************************************************
 166
 167 void ChimeraCheckCommand::help(){
 168         try {
 169
 170                 m->mothurOut("The chimera.check command reads a fastafile and templatefile and outputs potentially chimeric sequences.\n");
 171                 m->mothurOut("This command was created using the algorythms described in CHIMERA_CHECK version 2.7 written by Niels Larsen. \n");
 172                 m->mothurOut("The chimera.check command parameters are fasta, template, processors, ksize, increment, svg and name.\n");
 173                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required. \n");
 174                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 175                 m->mothurOut("The template parameter allows you to enter a template file containing known non-chimeric sequences, and is required. \n");
 176                 m->mothurOut("The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n");
 177                 #ifdef USE_MPI
 178                 m->mothurOut("When using MPI, the processors parameter is set to the number of MPI processes running. \n");
 179                 #endif
 180                 m->mothurOut("The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 10.\n");
 181                 m->mothurOut("The ksize parameter allows you to input kmersize, default is 7. \n");
 182                 m->mothurOut("The svg parameter allows you to specify whether or not you would like a svg file outputted for each query sequence, default is False.\n");
 183                 m->mothurOut("The name parameter allows you to enter a file containing names of sequences you would like .svg files for.\n");
 184                 m->mothurOut("You may enter multiple name files by separating their names with dashes. ie. fasta=abrecovery.svg.names-amzon.svg.names \n");
 185                 m->mothurOut("The chimera.check command should be in the following format: \n");
 186                 m->mothurOut("chimera.check(fasta=yourFastaFile, template=yourTemplateFile, processors=yourProcessors, ksize=yourKmerSize) \n");
 187                 m->mothurOut("Example: chimera.check(fasta=AD.fasta, template=core_set_aligned,imputed.fasta, processors=4, ksize=8) \n");
 188                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 189         }
 190         catch(exception& e) {
 191                 m->errorOut(e, "ChimeraCheckCommand", "help");
 192                 exit(1);
 193         }
 194 }
 195
 196 //***************************************************************************************************************
 197
 198 ChimeraCheckCommand::~ChimeraCheckCommand(){    /*      do nothing      */      }
 199
 200 //***************************************************************************************************************
 201
 202 int ChimeraCheckCommand::execute(){
 203         try{
 204
 205                 if (abort == true) { return 0; }
 206
 207                 for (int i = 0; i < fastaFileNames.size(); i++) {
 208
 209                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 210
 211                         int start = time(NULL);
 212
 213                         string thisNameFile = "";
 214                         if (nameFileNames.size() != 0) { thisNameFile = nameFileNames[i]; }
 215
 216                         chimera = new ChimeraCheckRDP(fastaFileNames[i], templatefile, thisNameFile, svg, increment, ksize, outputDir);
 217
 218                         if (m->control_pressed) { delete chimera;       return 0;       }
 219
 220                         string outputFileName = outputDir + getRootName(getSimpleName(fastaFileNames[i]))  + "chimeracheck.chimeras";
 221                         outputNames.push_back(outputFileName);
 222
 223                 #ifdef USE_MPI
 224
 225                                 int pid, end, numSeqsPerProcessor;
 226                                 int tag = 2001;
 227                                 vector<unsigned long int> MPIPos;
 228
 229                                 MPI_Status status;
 230                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 231                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 232
 233                                 MPI_File inMPI;
 234                                 MPI_File outMPI;
 235
 236                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 237                                 int inMode=MPI_MODE_RDONLY;
 238
 239                                 char outFilename[1024];
 240                                 strcpy(outFilename, outputFileName.c_str());
 241
 242                                 char inFileName[1024];
 243                                 strcpy(inFileName, fastaFileNames[i].c_str());
 244
 245                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 246                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 247
 248                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); } delete chimera; return 0;  }
 249
 250                                 if (pid == 0) { //you are the root process
 251                                         MPIPos = setFilePosFasta(fastaFileNames[i], numSeqs); //fills MPIPos, returns numSeqs
 252
 253                                         //send file positions to all processes
 254                                         for(int j = 1; j < processors; j++) {
 255                                                 MPI_Send(&numSeqs, 1, MPI_INT, j, tag, MPI_COMM_WORLD);
 256                                                 MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, j, tag, MPI_COMM_WORLD);
 257                                         }
 258
 259                                         //figure out how many sequences you have to align
 260                                         numSeqsPerProcessor = numSeqs / processors;
 261                                         int startIndex =  pid * numSeqsPerProcessor;
 262                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 263
 264
 265                                         //align your part
 266                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 267
 268                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); }   delete chimera; return 0;  }
 269
 270                                         //wait on chidren
 271                                         for(int j = 1; j < processors; j++) {
 272                                                 char buf[4];
 273                                                 MPI_Recv(buf, 4, MPI_CHAR, j, tag, MPI_COMM_WORLD, &status);
 274                                         }
 275                                 }else{ //you are a child process
 276                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 277                                         MPIPos.resize(numSeqs+1);
 278                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 279
 280                                         //figure out how many sequences you have to align
 281                                         numSeqsPerProcessor = numSeqs / processors;
 282                                         int startIndex =  pid * numSeqsPerProcessor;
 283                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 284
 285                                         //align your part
 286                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 287
 288                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);   for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 289
 290                                         //tell parent you are done.
 291                                         char buf[4];
 292                                         strcpy(buf, "done");
 293                                         MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 294                                 }
 295
 296                                 //close files
 297                                 MPI_File_close(&inMPI);
 298                                 MPI_File_close(&outMPI);
 299                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 300                 #else
 301
 302                         vector<unsigned long int> positions = divideFile(fastaFileNames[i], processors);
 303
 304                         for (int s = 0; s < (positions.size()-1); s++) {
 305                                 lines.push_back(new linePair(positions[s], positions[(s+1)]));
 306                         }
 307
 308                         //break up file
 309                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 310                                 if(processors == 1){
 311                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 312
 313                                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); } for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear(); delete chimera; return 0; }
 314
 315                                 }else{
 316                                         processIDS.resize(0);
 317
 318                                         numSeqs = createProcesses(outputFileName, fastaFileNames[i]);
 319
 320                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 321
 322                                         //append output files
 323                                         for(int j=1;j<processors;j++){
 324                                                 appendFiles((outputFileName + toString(processIDS[j]) + ".temp"), outputFileName);
 325                                                 remove((outputFileName + toString(processIDS[j]) + ".temp").c_str());
 326                                         }
 327
 328                                         if (m->control_pressed) {
 329                                                 for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }
 330                                                 for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 331                                                 delete chimera;
 332                                                 return 0;
 333                                         }
 334                                 }
 335
 336                         #else
 337                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[i]);
 338
 339                                 if (m->control_pressed) { for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear(); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } delete chimera; return 0; }
 340                         #endif
 341                 #endif
 342                         delete chimera;
 343                         for (int j = 0; j < lines.size(); j++) {  delete lines[j];  }  lines.clear();
 344
 345                         m->mothurOutEndLine(); m->mothurOut("This method does not determine if a sequence is chimeric, but allows you to make that determination based on the IS values."); m->mothurOutEndLine();
 346                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 347
 348                 }
 349
 350                 m->mothurOutEndLine();
 351                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 352                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 353                 m->mothurOutEndLine();
 354
 355                 return 0;
 356
 357         }
 358         catch(exception& e) {
 359                 m->errorOut(e, "ChimeraCheckCommand", "execute");
 360                 exit(1);
 361         }
 362 }
 363 //**********************************************************************************************************************
 364
 365 int ChimeraCheckCommand::driver(linePair* filePos, string outputFName, string filename){
 366         try {
 367                 ofstream out;
 368                 openOutputFile(outputFName, out);
 369
 370                 ofstream out2;
 371
 372                 ifstream inFASTA;
 373                 openInputFile(filename, inFASTA);
 374
 375                 inFASTA.seekg(filePos->start);
 376
 377                 bool done = false;
 378                 int count = 0;
 379
 380                 while (!done) {
 381
 382                         if (m->control_pressed) {       return 1;       }
 383
 384                         Sequence* candidateSeq = new Sequence(inFASTA);  gobble(inFASTA);
 385
 386                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 387                                 //find chimeras
 388                                 chimera->getChimeras(candidateSeq);
 389
 390                                 if (m->control_pressed) {       delete candidateSeq; return 1;  }
 391
 392                                 //print results
 393                                 chimera->print(out, out2);
 394                         }
 395                         delete candidateSeq;
 396
 397                         unsigned long int pos = inFASTA.tellg();
 398                         if ((pos == -1) || (pos >= filePos->end)) { break; }
 399
 400                         //report progress
 401                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 402                 }
 403                 //report progress
 404                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 405
 406                 out.close();
 407                 inFASTA.close();
 408
 409                 return count;
 410         }
 411         catch(exception& e) {
 412                 m->errorOut(e, "ChimeraCheckCommand", "driver");
 413                 exit(1);
 414         }
 415 }
 416 //**********************************************************************************************************************
 417 #ifdef USE_MPI
 418 int ChimeraCheckCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos){
 419         try {
 420                 MPI_File outAccMPI;
 421                 MPI_Status status;
 422                 int pid;
 423                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 424
 425                 for(int i=0;i<num;i++){
 426
 427                         if (m->control_pressed) { return 0; }
 428
 429                         //read next sequence
 430                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 431
 432                         char* buf4 = new char[length];
 433                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 434
 435                         string tempBuf = buf4;
 436                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 437                         istringstream iss (tempBuf,istringstream::in);
 438                         delete buf4;
 439
 440                         Sequence* candidateSeq = new Sequence(iss);  gobble(iss);
 441
 442                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 443                                 //find chimeras
 444                                 chimera->getChimeras(candidateSeq);
 445
 446                                 //print results
 447                                 chimera->print(outMPI, outAccMPI);
 448                         }
 449                         delete candidateSeq;
 450
 451                         //report progress
 452                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 453                 }
 454                 //report progress
 455                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 456
 457                 return 0;
 458         }
 459         catch(exception& e) {
 460                 m->errorOut(e, "ChimeraCheckCommand", "driverMPI");
 461                 exit(1);
 462         }
 463 }
 464 #endif
 465
 466 /**************************************************************************************************/
 467
 468 int ChimeraCheckCommand::createProcesses(string outputFileName, string filename) {
 469         try {
 470 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 471                 int process = 0;
 472                 int num = 0;
 473
 474                 //loop through and create all the processes you want
 475                 while (process != processors) {
 476                         int pid = fork();
 477
 478                         if (pid > 0) {
 479                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 480                                 process++;
 481                         }else if (pid == 0){
 482                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename);
 483
 484                                 //pass numSeqs to parent
 485                                 ofstream out;
 486                                 string tempFile = toString(getpid()) + ".temp";
 487                                 openOutputFile(tempFile, out);
 488                                 out << num << endl;
 489                                 out.close();
 490
 491                                 exit(0);
 492                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 493                 }
 494
 495                 //force parent to wait until all the processes are done
 496                 for (int i=0;i<processors;i++) {
 497                         int temp = processIDS[i];
 498                         wait(&temp);
 499                 }
 500
 501                 for (int i = 0; i < processIDS.size(); i++) {
 502                         ifstream in;
 503                         string tempFile =  toString(processIDS[i]) + ".temp";
 504                         openInputFile(tempFile, in);
 505                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 506                         in.close(); remove(tempFile.c_str());
 507                 }
 508
 509                 return num;
 510 #endif
 511         }
 512         catch(exception& e) {
 513                 m->errorOut(e, "ChimeraCheckCommand", "createProcesses");
 514                 exit(1);
 515         }
 516 }
 517 /**************************************************************************************************/
 518
 519