filterseqscommand.cpp

   1 /*
   2  *  filterseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Thomas Ryabin on 5/4/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "filterseqscommand.h"
  11 #include "sequence.hpp"
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> FilterSeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter phard("hard", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(phard);
  19                 CommandParameter ptrump("trump", "String", "", "*", "", "", "",false,false); parameters.push_back(ptrump);
  20                 CommandParameter psoft("soft", "Number", "", "0", "", "", "",false,false); parameters.push_back(psoft);
  21                 CommandParameter pvertical("vertical", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pvertical);
  22                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "FilterSeqsCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string FilterSeqsCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n";
  40                 helpString += "The filter.seqs command parameters are fasta, trump, soft, hard, processors and vertical. \n";
  41                 helpString += "The fasta parameter is required, unless you have a valid current fasta file. You may enter several fasta files to build the filter from and filter, by separating their names with -'s.\n";
  42                 helpString += "For example: fasta=abrecovery.fasta-amazon.fasta \n";
  43                 helpString += "The trump option will remove a column if the trump character is found at that position in any sequence of the alignment. Default=*, meaning no trump. \n";
  44                 helpString += "A soft mask removes any column where the dominant base (i.e. A, T, G, C, or U) does not occur in at least a designated percentage of sequences. Default=0.\n";
  45                 helpString += "The hard parameter allows you to enter a file containing the filter you want to use.\n";
  46                 helpString += "The vertical parameter removes columns where all sequences contain a gap character. The default is T.\n";
  47                 helpString += "The processors parameter allows you to specify the number of processors to use. The default is 1.\n";
  48                 helpString += "The filter.seqs command should be in the following format: \n";
  49                 helpString += "filter.seqs(fasta=yourFastaFile, trump=yourTrump) \n";
  50                 helpString += "Example filter.seqs(fasta=abrecovery.fasta, trump=.).\n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "FilterSeqsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 FilterSeqsCommand::FilterSeqsCommand(){
  61         try {
  62                 abort = true; calledHelp = true;
  63                 setParameters();
  64                 vector<string> tempOutNames;
  65                 outputTypes["fasta"] = tempOutNames;
  66                 outputTypes["filter"] = tempOutNames;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
  70                 exit(1);
  71         }
  72 }
  73 /**************************************************************************************/
  74 FilterSeqsCommand::FilterSeqsCommand(string option)  {
  75         try {
  76                 abort = false; calledHelp = false;
  77                 filterFileName = "";
  78
  79                 //allow user to run help
  80                 if(option == "help") { help(); abort = true; calledHelp = true; }
  81                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  82
  83                 else {
  84                         vector<string> myArray = setParameters();
  85
  86                         OptionParser parser(option);
  87                         map<string,string> parameters = parser.getParameters();
  88
  89                         ValidParameters validParameter("filter.seqs");
  90                         map<string,string>::iterator it;
  91
  92                         //check to make sure all parameters are valid for command
  93                         for (it = parameters.begin(); it != parameters.end(); it++) {
  94                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  95                         }
  96
  97                         //initialize outputTypes
  98                         vector<string> tempOutNames;
  99                         outputTypes["fasta"] = tempOutNames;
 100                         outputTypes["filter"] = tempOutNames;
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 string path;
 107                                 it = parameters.find("fasta");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 113                                 }
 114
 115                                 it = parameters.find("hard");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["hard"] = inputDir + it->second;             }
 121                                 }
 122                         }
 123
 124                         //check for required parameters
 125                         fasta = validParameter.validFile(parameters, "fasta", false);
 126                         if (fasta == "not found") {
 127                                 fasta = m->getFastaFile();
 128                                 if (fasta != "") { fastafileNames.push_back(fasta);  m->mothurOut("Using " + fasta + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 129                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 130                         }
 131                         else {
 132                                 m->splitAtDash(fasta, fastafileNames);
 133
 134                                 //go through files and make sure they are good, if not, then disregard them
 135                                 for (int i = 0; i < fastafileNames.size(); i++) {
 136
 137                                         bool ignore = false;
 138                                         if (fastafileNames[i] == "current") {
 139                                                 fastafileNames[i] = m->getFastaFile();
 140                                                 if (fastafileNames[i] != "") {  m->mothurOut("Using " + fastafileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 141                                                 else {
 142                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 143                                                         //erase from file list
 144                                                         fastafileNames.erase(fastafileNames.begin()+i);
 145                                                         i--;
 146                                                 }
 147                                         }
 148
 149                                         if (!ignore) {
 150                                                 if (inputDir != "") {
 151                                                         string path = m->hasPath(fastafileNames[i]);
 152                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                                         if (path == "") {       fastafileNames[i] = inputDir + fastafileNames[i];               }
 154                                                 }
 155
 156                                                 ifstream in;
 157                                                 int ableToOpen = m->openInputFile(fastafileNames[i], in, "noerror");
 158
 159                                                 //if you can't open it, try default location
 160                                                 if (ableToOpen == 1) {
 161                                                         if (m->getDefaultPath() != "") { //default path is set
 162                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastafileNames[i]);
 163                                                                 m->mothurOut("Unable to open " + fastafileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 164                                                                 ifstream in2;
 165                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 166                                                                 in2.close();
 167                                                                 fastafileNames[i] = tryPath;
 168                                                         }
 169                                                 }
 170
 171                                                 //if you can't open it, try default location
 172                                                 if (ableToOpen == 1) {
 173                                                         if (m->getOutputDir() != "") { //default path is set
 174                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastafileNames[i]);
 175                                                                 m->mothurOut("Unable to open " + fastafileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 176                                                                 ifstream in2;
 177                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 178                                                                 in2.close();
 179                                                                 fastafileNames[i] = tryPath;
 180                                                         }
 181                                                 }
 182
 183                                                 in.close();
 184
 185                                                 if (ableToOpen == 1) {
 186                                                         m->mothurOut("Unable to open " + fastafileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 187                                                         //erase from file list
 188                                                         fastafileNames.erase(fastafileNames.begin()+i);
 189                                                         i--;
 190                                                 }else{
 191                                                         string simpleName = m->getSimpleName(fastafileNames[i]);
 192                                                         filterFileName += simpleName.substr(0, simpleName.find_first_of('.'));
 193                                                         m->setFastaFile(fastafileNames[i]);
 194                                                 }
 195                                                 in.close();
 196                                         }
 197                                 }
 198
 199                                 //make sure there is at least one valid file left
 200                                 if (fastafileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 201                         }
 202
 203                         if (!abort) {
 204                                 //if the user changes the output directory command factory will send this info to us in the output parameter
 205                                 outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 206                                         outputDir = "";
 207                                         outputDir += m->hasPath(fastafileNames[0]); //if user entered a file with a path then preserve it
 208                                 }
 209                         }
 210                         //check for optional parameter and set defaults
 211                         // ...at some point should added some additional type checking...
 212
 213                         string temp;
 214                         hard = validParameter.validFile(parameters, "hard", true);                              if (hard == "not found") { hard = ""; }
 215                         else if (hard == "not open") { hard = ""; abort = true; }
 216
 217                         temp = validParameter.validFile(parameters, "trump", false);                    if (temp == "not found") { temp = "*"; }
 218                         trump = temp[0];
 219
 220                         temp = validParameter.validFile(parameters, "soft", false);                             if (temp == "not found") { soft = 0; }
 221                         else {  soft = (float)atoi(temp.c_str()) / 100.0;  }
 222
 223                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 224                         m->setProcessors(temp);
 225                         convert(temp, processors);
 226
 227                         vertical = validParameter.validFile(parameters, "vertical", false);
 228                         if (vertical == "not found") {
 229                                 if ((hard == "") && (trump == '*') && (soft == 0)) { vertical = "T"; } //you have not given a hard file or set the trump char.
 230                                 else { vertical = "F";  }
 231                         }
 232
 233                         numSeqs = 0;
 234                 }
 235
 236         }
 237         catch(exception& e) {
 238                 m->errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
 239                 exit(1);
 240         }
 241 }
 242 /**************************************************************************************/
 243
 244 int FilterSeqsCommand::execute() {
 245         try {
 246
 247                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 248
 249                 ifstream inFASTA;
 250                 m->openInputFile(fastafileNames[0], inFASTA);
 251
 252                 Sequence testSeq(inFASTA);
 253                 alignmentLength = testSeq.getAlignLength();
 254                 inFASTA.close();
 255
 256                 ////////////create filter/////////////////
 257                 m->mothurOut("Creating Filter... "); m->mothurOutEndLine();
 258
 259                 filter = createFilter();
 260
 261                 m->mothurOutEndLine();  m->mothurOutEndLine();
 262
 263                 if (m->control_pressed) { outputTypes.clear(); return 0; }
 264
 265                 #ifdef USE_MPI
 266                         int pid;
 267                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 268
 269                         if (pid == 0) { //only one process should output the filter
 270                 #endif
 271
 272                 ofstream outFilter;
 273
 274                 //prevent giantic file name
 275                 string filterFile;
 276                 if (fastafileNames.size() > 3) { filterFile = outputDir + "merge.filter"; }
 277                 else {  filterFile = outputDir + filterFileName + ".filter";  }
 278
 279                 m->openOutputFile(filterFile, outFilter);
 280                 outFilter << filter << endl;
 281                 outFilter.close();
 282                 outputNames.push_back(filterFile); outputTypes["filter"].push_back(filterFile);
 283
 284                 #ifdef USE_MPI
 285                         }
 286                 #endif
 287
 288                 ////////////run filter/////////////////
 289
 290                 m->mothurOut("Running Filter... "); m->mothurOutEndLine();
 291
 292                 filterSequences();
 293
 294                 m->mothurOutEndLine();  m->mothurOutEndLine();
 295
 296                 int filteredLength = 0;
 297                 for(int i=0;i<alignmentLength;i++){
 298                         if(filter[i] == '1'){   filteredLength++;       }
 299                 }
 300
 301                 if (m->control_pressed) {  outputTypes.clear(); for(int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }  return 0; }
 302
 303
 304                 m->mothurOutEndLine();
 305                 m->mothurOut("Length of filtered alignment: " + toString(filteredLength)); m->mothurOutEndLine();
 306                 m->mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); m->mothurOutEndLine();
 307                 m->mothurOut("Length of the original alignment: " + toString(alignmentLength)); m->mothurOutEndLine();
 308                 m->mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); m->mothurOutEndLine();
 309
 310                 //set fasta file as new current fastafile
 311                 string current = "";
 312                 itTypes = outputTypes.find("fasta");
 313                 if (itTypes != outputTypes.end()) {
 314                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 315                 }
 316
 317                 m->mothurOutEndLine();
 318                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 319                 for(int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();      }
 320                 m->mothurOutEndLine();
 321
 322                 return 0;
 323
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "FilterSeqsCommand", "execute");
 327                 exit(1);
 328         }
 329 }
 330 /**************************************************************************************/
 331 int FilterSeqsCommand::filterSequences() {
 332         try {
 333
 334                 numSeqs = 0;
 335
 336                 for (int s = 0; s < fastafileNames.size(); s++) {
 337
 338                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 339
 340                                 string filteredFasta = outputDir + m->getRootName(m->getSimpleName(fastafileNames[s])) + "filter.fasta";
 341 #ifdef USE_MPI
 342                                 int pid, numSeqsPerProcessor, num;
 343                                 int tag = 2001;
 344                                 vector<unsigned long int>MPIPos;
 345
 346                                 MPI_Status status;
 347                                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 348                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 349
 350                                 MPI_File outMPI;
 351                                 MPI_File inMPI;
 352                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 353                                 int inMode=MPI_MODE_RDONLY;
 354
 355                                 char outFilename[1024];
 356                                 strcpy(outFilename, filteredFasta.c_str());
 357
 358                                 char inFileName[1024];
 359                                 strcpy(inFileName, fastafileNames[s].c_str());
 360
 361                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 362                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 363
 364                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 365
 366                                 if (pid == 0) { //you are the root process
 367
 368                                         MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
 369                                         numSeqs += num;
 370
 371                                         //send file positions to all processes
 372                                         for(int i = 1; i < processors; i++) {
 373                                                 MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 374                                                 MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 375                                         }
 376
 377                                         //figure out how many sequences you have to do
 378                                         numSeqsPerProcessor = num / processors;
 379                                         int startIndex =  pid * numSeqsPerProcessor;
 380                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 381
 382
 383                                         //do your part
 384                                         driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 385
 386                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 387
 388                                         //wait on chidren
 389                                         for(int i = 1; i < processors; i++) {
 390                                                 char buf[5];
 391                                                 MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 392                                         }
 393
 394                                 }else { //you are a child process
 395                                         MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 396                                         MPIPos.resize(num+1);
 397                                         numSeqs += num;
 398                                         MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 399
 400                                         //figure out how many sequences you have to align
 401                                         numSeqsPerProcessor = num / processors;
 402                                         int startIndex =  pid * numSeqsPerProcessor;
 403                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 404
 405
 406                                         //align your part
 407                                         driverMPIRun(startIndex, numSeqsPerProcessor, inMPI, outMPI, MPIPos);
 408
 409                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI);  return 0;  }
 410
 411                                         char buf[5];
 412                                         strcpy(buf, "done");
 413
 414                                         //tell parent you are done.
 415                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 416                                 }
 417
 418                                 MPI_File_close(&outMPI);
 419                                 MPI_File_close(&inMPI);
 420                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 421
 422 #else
 423                         vector<unsigned long int> positions = m->divideFile(fastafileNames[s], processors);
 424
 425                         for (int i = 0; i < (positions.size()-1); i++) {
 426                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 427                         }
 428                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 429                                 if(processors == 1){
 430                                         int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
 431                                         numSeqs += numFastaSeqs;
 432                                 }else{
 433                                         int numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s]);
 434                                         numSeqs += numFastaSeqs;
 435
 436                                         rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
 437
 438                                         //append fasta files
 439                                         for(int i=1;i<processors;i++){
 440                                                 m->appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
 441                                                 remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
 442                                         }
 443                                 }
 444
 445                                 if (m->control_pressed) {  return 1; }
 446                 #else
 447                                 int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
 448                                 numSeqs += numFastaSeqs;
 449
 450                                 if (m->control_pressed) {  return 1; }
 451                 #endif
 452 #endif
 453                         outputNames.push_back(filteredFasta); outputTypes["fasta"].push_back(filteredFasta);
 454                 }
 455
 456                 return 0;
 457         }
 458         catch(exception& e) {
 459                 m->errorOut(e, "FilterSeqsCommand", "filterSequences");
 460                 exit(1);
 461         }
 462 }
 463 #ifdef USE_MPI
 464 /**************************************************************************************/
 465 int FilterSeqsCommand::driverMPIRun(int start, int num, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
 466         try {
 467                 string outputString = "";
 468                 int count = 0;
 469                 MPI_Status status;
 470
 471                 for(int i=0;i<num;i++){
 472
 473                         if (m->control_pressed) { return 0; }
 474
 475                         //read next sequence
 476                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 477                         char* buf4 = new char[length];
 478                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 479
 480                         string tempBuf = buf4;
 481                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 482                         istringstream iss (tempBuf,istringstream::in);
 483                         delete buf4;
 484
 485                         Sequence seq(iss);  m->gobble(iss);
 486
 487                         if (seq.getName() != "") {
 488                                 string align = seq.getAligned();
 489                                 string filterSeq = "";
 490
 491                                 for(int j=0;j<alignmentLength;j++){
 492                                         if(filter[j] == '1'){
 493                                                 filterSeq += align[j];
 494                                         }
 495                                 }
 496
 497                                 count++;
 498                                 outputString += ">" + seq.getName() + "\n" + filterSeq + "\n";
 499
 500                                 if(count % 10 == 0){ //output to file
 501                                         //send results to parent
 502                                         int length = outputString.length();
 503                                         char* buf = new char[length];
 504                                         memcpy(buf, outputString.c_str(), length);
 505
 506                                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 507                                         outputString = "";
 508                                         delete buf;
 509                                 }
 510
 511                         }
 512
 513                         if((i+1) % 100 == 0){   cout << (i+1) << endl;   m->mothurOutJustToLog(toString(i+1) + "\n");   }
 514                 }
 515
 516                 if(outputString != ""){ //output to file
 517                         //send results to parent
 518                         int length = outputString.length();
 519                         char* buf = new char[length];
 520                         memcpy(buf, outputString.c_str(), length);
 521
 522                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 523                         outputString = "";
 524                         delete buf;
 525                 }
 526
 527                 if((num) % 100 != 0){   cout << (num) << endl;   m->mothurOutJustToLog(toString(num) + "\n");   }
 528
 529                 return 0;
 530         }
 531         catch(exception& e) {
 532                 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
 533                 exit(1);
 534         }
 535 }
 536 #endif
 537 /**************************************************************************************/
 538 int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* filePos) {
 539         try {
 540                 ofstream out;
 541                 m->openOutputFile(outputFilename, out);
 542
 543                 ifstream in;
 544                 m->openInputFile(inputFilename, in);
 545
 546                 in.seekg(filePos->start);
 547
 548                 bool done = false;
 549                 int count = 0;
 550
 551                 while (!done) {
 552
 553                                 if (m->control_pressed) { in.close(); out.close(); return 0; }
 554
 555                                 Sequence seq(in); m->gobble(in);
 556                                 if (seq.getName() != "") {
 557                                         string align = seq.getAligned();
 558                                         string filterSeq = "";
 559
 560                                         for(int j=0;j<alignmentLength;j++){
 561                                                 if(filter[j] == '1'){
 562                                                         filterSeq += align[j];
 563                                                 }
 564                                         }
 565
 566                                         out << '>' << seq.getName() << endl << filterSeq << endl;
 567                                 count++;
 568                         }
 569
 570                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 571                                 unsigned long int pos = in.tellg();
 572                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 573                         #else
 574                                 if (in.eof()) { break; }
 575                         #endif
 576
 577                         //report progress
 578                         if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 579                 }
 580                 //report progress
 581                 if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 582
 583
 584                 out.close();
 585                 in.close();
 586
 587                 return count;
 588         }
 589         catch(exception& e) {
 590                 m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
 591                 exit(1);
 592         }
 593 }
 594 /**************************************************************************************************/
 595
 596 int FilterSeqsCommand::createProcessesRunFilter(string F, string filename) {
 597         try {
 598 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 599                 int process = 0;
 600                 int num = 0;
 601                 processIDS.clear();
 602
 603                 //loop through and create all the processes you want
 604                 while (process != processors) {
 605                         int pid = fork();
 606
 607                         if (pid > 0) {
 608                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 609                                 process++;
 610                         }else if (pid == 0){
 611                                 string filteredFasta = filename + toString(getpid()) + ".temp";
 612                                 num = driverRunFilter(F, filteredFasta, filename, lines[process]);
 613
 614                                 //pass numSeqs to parent
 615                                 ofstream out;
 616                                 string tempFile = filename +  toString(getpid()) + ".num.temp";
 617                                 m->openOutputFile(tempFile, out);
 618                                 out << num << endl;
 619                                 out.close();
 620
 621                                 exit(0);
 622                         }else {
 623                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 624                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 625                                 exit(0);
 626                         }
 627                 }
 628
 629                 //force parent to wait until all the processes are done
 630                 for (int i=0;i<processors;i++) {
 631                         int temp = processIDS[i];
 632                         wait(&temp);
 633                 }
 634
 635                 for (int i = 0; i < processIDS.size(); i++) {
 636                         ifstream in;
 637                         string tempFile =  filename + toString(processIDS[i]) + ".num.temp";
 638                         m->openInputFile(tempFile, in);
 639                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 640                         in.close(); remove(tempFile.c_str());
 641                 }
 642
 643
 644                 return num;
 645 #endif
 646         }
 647         catch(exception& e) {
 648                 m->errorOut(e, "FilterSeqsCommand", "createProcessesRunFilter");
 649                 exit(1);
 650         }
 651 }
 652 /**************************************************************************************/
 653 string FilterSeqsCommand::createFilter() {
 654         try {
 655                 string filterString = "";
 656                 Filters F;
 657
 658                 if (soft != 0)                  {  F.setSoft(soft);             }
 659                 if (trump != '*')               {  F.setTrump(trump);   }
 660
 661                 F.setLength(alignmentLength);
 662
 663                 if(trump != '*' || m->isTrue(vertical) || soft != 0){
 664                         F.initialize();
 665                 }
 666
 667                 if(hard.compare("") != 0)       {       F.doHard(hard);         }
 668                 else                                            {       F.setFilter(string(alignmentLength, '1'));      }
 669
 670                 numSeqs = 0;
 671                 if(trump != '*' || m->isTrue(vertical) || soft != 0){
 672                         for (int s = 0; s < fastafileNames.size(); s++) {
 673
 674                                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 675
 676 #ifdef USE_MPI
 677                                 int pid, numSeqsPerProcessor, num;
 678                                 int tag = 2001;
 679                                 vector<unsigned long int> MPIPos;
 680
 681                                 MPI_Status status;
 682                                 MPI_File inMPI;
 683                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 684                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 685
 686                                 //char* tempFileName = new char(fastafileNames[s].length());
 687                                 //tempFileName = &(fastafileNames[s][0]);
 688
 689                                 char tempFileName[1024];
 690                                 strcpy(tempFileName, fastafileNames[s].c_str());
 691
 692                                 MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 693
 694                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 695
 696                                 if (pid == 0) { //you are the root process
 697                                                 MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
 698                                                 numSeqs += num;
 699
 700                                                 //send file positions to all processes
 701                                                 for(int i = 1; i < processors; i++) {
 702                                                         MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 703                                                         MPI_Send(&MPIPos[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 704                                                 }
 705
 706                                                 //figure out how many sequences you have to do
 707                                                 numSeqsPerProcessor = num / processors;
 708                                                 int startIndex =  pid * numSeqsPerProcessor;
 709                                                 if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 710
 711
 712                                                 //do your part
 713                                                 MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI, MPIPos);
 714
 715                                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 716
 717                                 }else { //i am the child process
 718                                         MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 719                                         MPIPos.resize(num+1);
 720                                         numSeqs += num;
 721                                         MPI_Recv(&MPIPos[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 722
 723                                         //figure out how many sequences you have to align
 724                                         numSeqsPerProcessor = num / processors;
 725                                         int startIndex =  pid * numSeqsPerProcessor;
 726                                         if(pid == (processors - 1)){    numSeqsPerProcessor = num - pid * numSeqsPerProcessor;  }
 727
 728
 729                                         //do your part
 730                                         MPICreateFilter(startIndex, numSeqsPerProcessor, F, inMPI,  MPIPos);
 731
 732                                         if (m->control_pressed) {  MPI_File_close(&inMPI);  return 0;  }
 733                                 }
 734
 735                                 MPI_File_close(&inMPI);
 736                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 737
 738 #else
 739                 vector<unsigned long int> positions = m->divideFile(fastafileNames[s], processors);
 740
 741                 for (int i = 0; i < (positions.size()-1); i++) {
 742                         lines.push_back(new linePair(positions[i], positions[(i+1)]));
 743                 }
 744                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 745                                 if(processors == 1){
 746                                         int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
 747                                         numSeqs += numFastaSeqs;
 748                                 }else{
 749                                         int numFastaSeqs = createProcessesCreateFilter(F, fastafileNames[s]);
 750                                         numSeqs += numFastaSeqs;
 751                                 }
 752
 753                                 if (m->control_pressed) {  return filterString; }
 754                 #else
 755                                 int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
 756                                 numSeqs += numFastaSeqs;
 757                                 if (m->control_pressed) {  return filterString; }
 758                 #endif
 759 #endif
 760
 761                         }
 762                 }
 763
 764
 765 #ifdef USE_MPI
 766                 int pid;
 767                 int Atag = 1; int Ttag = 2; int Ctag = 3; int Gtag = 4; int Gaptag = 5;
 768                 MPI_Status status;
 769
 770                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 771
 772                 if(trump != '*' || m->isTrue(vertical) || soft != 0){
 773
 774                         if (pid == 0) { //only one process should output the filter
 775
 776                                 vector<int> temp; temp.resize(alignmentLength+1);
 777
 778                                 //get the frequencies from the child processes
 779                                 for(int i = 1; i < processors; i++) {
 780
 781                                         for (int j = 0; j < 5; j++) {
 782
 783                                                 MPI_Recv(&temp[0], (alignmentLength+1), MPI_INT, i, 2001, MPI_COMM_WORLD, &status);
 784                                                 int receiveTag = temp[temp.size()-1];  //child process added a int to the end to indicate what letter count this is for
 785
 786                                                 if (receiveTag == Atag) { //you are recieveing the A frequencies
 787                                                         for (int k = 0; k < alignmentLength; k++) {             F.a[k] += temp[k];      }
 788                                                 }else if (receiveTag == Ttag) { //you are recieveing the T frequencies
 789                                                         for (int k = 0; k < alignmentLength; k++) {             F.t[k] += temp[k];      }
 790                                                 }else if (receiveTag == Ctag) { //you are recieveing the C frequencies
 791                                                         for (int k = 0; k < alignmentLength; k++) {             F.c[k] += temp[k];      }
 792                                                 }else if (receiveTag == Gtag) { //you are recieveing the G frequencies
 793                                                         for (int k = 0; k < alignmentLength; k++) {             F.g[k] += temp[k];      }
 794                                                 }else if (receiveTag == Gaptag) { //you are recieveing the gap frequencies
 795                                                         for (int k = 0; k < alignmentLength; k++) {             F.gap[k] += temp[k];    }
 796                                                 }
 797                                         }
 798                                 }
 799                         }else{
 800
 801                                 //send my fequency counts
 802                                 F.a.push_back(Atag);
 803                                 int ierr = MPI_Send(&(F.a[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 804                                 F.t.push_back(Ttag);
 805                                 ierr = MPI_Send (&(F.t[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 806                                 F.c.push_back(Ctag);
 807                                 ierr = MPI_Send(&(F.c[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 808                                 F.g.push_back(Gtag);
 809                                 ierr = MPI_Send(&(F.g[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 810                                 F.gap.push_back(Gaptag);
 811                                 ierr = MPI_Send(&(F.gap[0]), (alignmentLength+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 812                         }
 813
 814                 }
 815
 816                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 817
 818                 if (pid == 0) { //only one process should output the filter
 819 #endif
 820
 821                 F.setNumSeqs(numSeqs);
 822                 if(m->isTrue(vertical) == 1)    {       F.doVertical(); }
 823                 if(soft != 0)                           {       F.doSoft();             }
 824                 filterString = F.getFilter();
 825
 826 #ifdef USE_MPI
 827                 //send filter string to kids
 828                 //for(int i = 1; i < processors; i++) {
 829                 //      MPI_Send(&filterString[0], alignmentLength, MPI_CHAR, i, 2001, MPI_COMM_WORLD);
 830                 //}
 831                 MPI_Bcast(&filterString[0], alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
 832         }else{
 833                 //recieve filterString
 834                 char* tempBuf = new char[alignmentLength];
 835                 //MPI_Recv(&tempBuf[0], alignmentLength, MPI_CHAR, 0, 2001, MPI_COMM_WORLD, &status);
 836                 MPI_Bcast(tempBuf, alignmentLength, MPI_CHAR, 0, MPI_COMM_WORLD);
 837
 838                 filterString = tempBuf;
 839                 if (filterString.length() > alignmentLength) { filterString = filterString.substr(0, alignmentLength);  }
 840                 delete tempBuf;
 841         }
 842
 843         MPI_Barrier(MPI_COMM_WORLD);
 844 #endif
 845
 846                 return filterString;
 847         }
 848         catch(exception& e) {
 849                 m->errorOut(e, "FilterSeqsCommand", "createFilter");
 850                 exit(1);
 851         }
 852 }
 853 /**************************************************************************************/
 854 int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* filePos) {
 855         try {
 856
 857                 ifstream in;
 858                 m->openInputFile(filename, in);
 859
 860                 in.seekg(filePos->start);
 861
 862                 bool done = false;
 863                 int count = 0;
 864
 865                 while (!done) {
 866
 867                         if (m->control_pressed) { in.close(); return 1; }
 868
 869                         Sequence seq(in); m->gobble(in);
 870                         if (seq.getName() != "") {
 871                                         if (seq.getAligned().length() != alignmentLength) { m->mothurOut("Sequences are not all the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true;  }
 872
 873                                         if(trump != '*')                        {       F.doTrump(seq);         }
 874                                         if(m->isTrue(vertical) || soft != 0)    {       F.getFreqs(seq);        }
 875                                         cout.flush();
 876                                         count++;
 877                         }
 878
 879                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 880                                 unsigned long int pos = in.tellg();
 881                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 882                         #else
 883                                 if (in.eof()) { break; }
 884                         #endif
 885
 886                         //report progress
 887                         if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 888                 }
 889                 //report progress
 890                 if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 891                 in.close();
 892
 893                 return count;
 894         }
 895         catch(exception& e) {
 896                 m->errorOut(e, "FilterSeqsCommand", "driverCreateFilter");
 897                 exit(1);
 898         }
 899 }
 900 #ifdef USE_MPI
 901 /**************************************************************************************/
 902 int FilterSeqsCommand::MPICreateFilter(int start, int num, Filters& F, MPI_File& inMPI, vector<unsigned long int>& MPIPos) {
 903         try {
 904
 905                 MPI_Status status;
 906                 int pid;
 907                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 908
 909                 for(int i=0;i<num;i++){
 910
 911                         if (m->control_pressed) { return 0; }
 912
 913                         //read next sequence
 914                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 915
 916                         char* buf4 = new char[length];
 917                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 918
 919                         string tempBuf = buf4;
 920                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 921                         istringstream iss (tempBuf,istringstream::in);
 922                         delete buf4;
 923
 924                         Sequence seq(iss);
 925
 926                         if (seq.getAligned().length() != alignmentLength) {  cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1);  }
 927
 928                         if(trump != '*'){       F.doTrump(seq); }
 929                         if(m->isTrue(vertical) || soft != 0){   F.getFreqs(seq);        }
 930                         cout.flush();
 931
 932                         //report progress
 933                         if((i+1) % 100 == 0){   cout << (i+1) << endl;   m->mothurOutJustToLog(toString(i+1) + "\n");   }
 934                 }
 935
 936                 //report progress
 937                 if((num) % 100 != 0){   cout << num << endl; m->mothurOutJustToLog(toString(num) + "\n");       }
 938
 939                 return 0;
 940         }
 941         catch(exception& e) {
 942                 m->errorOut(e, "FilterSeqsCommand", "MPICreateFilter");
 943                 exit(1);
 944         }
 945 }
 946 #endif
 947 /**************************************************************************************************/
 948
 949 int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
 950         try {
 951 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 952                 int process = 1;
 953                 int num = 0;
 954                 processIDS.clear();
 955
 956                 //loop through and create all the processes you want
 957                 while (process != processors) {
 958                         int pid = fork();
 959
 960                         if (pid > 0) {
 961                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 962                                 process++;
 963                         }else if (pid == 0){
 964                                 //reset child's filter counts to 0;
 965                                 F.a.clear(); F.a.resize(alignmentLength, 0);
 966                                 F.t.clear(); F.t.resize(alignmentLength, 0);
 967                                 F.g.clear(); F.g.resize(alignmentLength, 0);
 968                                 F.c.clear(); F.c.resize(alignmentLength, 0);
 969                                 F.gap.clear(); F.gap.resize(alignmentLength, 0);
 970
 971                                 num = driverCreateFilter(F, filename, lines[process]);
 972
 973                                 //write out filter counts to file
 974                                 filename += toString(getpid()) + "filterValues.temp";
 975                                 ofstream out;
 976                                 m->openOutputFile(filename, out);
 977
 978                                 out << num << endl;
 979                                 out << F.getFilter() << endl;
 980                                 for (int k = 0; k < alignmentLength; k++) {             out << F.a[k] << '\t'; }  out << endl;
 981                                 for (int k = 0; k < alignmentLength; k++) {             out << F.t[k] << '\t'; }  out << endl;
 982                                 for (int k = 0; k < alignmentLength; k++) {             out << F.g[k] << '\t'; }  out << endl;
 983                                 for (int k = 0; k < alignmentLength; k++) {             out << F.c[k] << '\t'; }  out << endl;
 984                                 for (int k = 0; k < alignmentLength; k++) {             out << F.gap[k] << '\t'; }  out << endl;
 985
 986                                 //cout << F.getFilter() << endl;
 987                                 out.close();
 988
 989                                 exit(0);
 990                         }else {
 991                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 992                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 993                                 exit(0);
 994                         }
 995                 }
 996
 997                 //parent do your part
 998                 num = driverCreateFilter(F, filename, lines[0]);
 999
1000                 //force parent to wait until all the processes are done
1001                 for (int i=0;i<(processors-1);i++) {
1002                         int temp = processIDS[i];
1003                         wait(&temp);
1004                 }
1005
1006                 //parent reads in and combines Filter info
1007                 for (int i = 0; i < processIDS.size(); i++) {
1008                         string tempFilename = filename + toString(processIDS[i]) + "filterValues.temp";
1009                         ifstream in;
1010                         m->openInputFile(tempFilename, in);
1011
1012                         int temp, tempNum;
1013                         string tempFilterString;
1014
1015                         in >> tempNum; m->gobble(in); num += tempNum;
1016
1017                         in >> tempFilterString;
1018                         F.mergeFilter(tempFilterString);
1019
1020                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.a[k] += temp; }           m->gobble(in);
1021                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.t[k] += temp; }           m->gobble(in);
1022                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.g[k] += temp; }           m->gobble(in);
1023                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.c[k] += temp; }           m->gobble(in);
1024                         for (int k = 0; k < alignmentLength; k++) {             in >> temp; F.gap[k] += temp; } m->gobble(in);
1025
1026                         in.close();
1027                         remove(tempFilename.c_str());
1028                 }
1029
1030                 return num;
1031 #endif
1032         }
1033         catch(exception& e) {
1034                 m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
1035                 exit(1);
1036         }
1037 }
1038 /**************************************************************************************/