seqsummarycommand.cpp

   1 /*
   2  *  seqcoordcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 5/30/09.
   6  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 #include "seqsummarycommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> SeqSummaryCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta","name","processors","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "SeqSummaryCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 SeqSummaryCommand::SeqSummaryCommand(){
  27         try {
  28                 abort = true; calledHelp = true;
  29                 vector<string> tempOutNames;
  30                 outputTypes["summary"] = tempOutNames;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "SeqSummaryCommand", "SeqSummaryCommand");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 vector<string> SeqSummaryCommand::getRequiredParameters(){
  39         try {
  40                 string Array[] =  {"fasta"};
  41                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  42                 return myArray;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "SeqSummaryCommand", "getRequiredParameters");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 vector<string> SeqSummaryCommand::getRequiredFiles(){
  51         try {
  52                 vector<string> myArray;
  53                 return myArray;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "SeqSummaryCommand", "getRequiredFiles");
  57                 exit(1);
  58         }
  59 }
  60 //***************************************************************************************************************
  61
  62 SeqSummaryCommand::SeqSummaryCommand(string option)  {
  63         try {
  64                 abort = false; calledHelp = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; calledHelp = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fasta","name","processors","outputdir","inputdir"};
  72                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  73
  74                         OptionParser parser(option);
  75                         map<string,string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter("summary.seqs");
  78                         map<string,string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         //if the user changes the input directory command factory will send this info to us in the output parameter
  86                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  87                         if (inputDir == "not found"){   inputDir = "";          }
  88                         else {
  89                                 string path;
  90                                 it = parameters.find("fasta");
  91                                 //user has given a template file
  92                                 if(it != parameters.end()){
  93                                         path = m->hasPath(it->second);
  94                                         //if the user has not given a path then, add inputdir. else leave path alone.
  95                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  96                                 }
  97
  98                                 it = parameters.find("name");
  99                                 //user has given a template file
 100                                 if(it != parameters.end()){
 101                                         path = m->hasPath(it->second);
 102                                         //if the user has not given a path then, add inputdir. else leave path alone.
 103                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 104                                 }
 105                         }
 106
 107                         //initialize outputTypes
 108                         vector<string> tempOutNames;
 109                         outputTypes["summary"] = tempOutNames;
 110
 111                         //check for required parameters
 112                         fastafile = validParameter.validFile(parameters, "fasta", true);
 113                         if (fastafile == "not open") { abort = true; }
 114                         else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the summary.seqs command."); m->mothurOutEndLine(); abort = true;  }
 115
 116                         namefile = validParameter.validFile(parameters, "name", true);
 117                         if (namefile == "not open") { namefile = ""; abort = true; }
 118                         else if (namefile == "not found") { namefile = "";  }
 119
 120                         //if the user changes the output directory command factory will send this info to us in the output parameter
 121                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 122                                 outputDir = "";
 123                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 124                         }
 125
 126                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = "1";                             }
 127                         convert(temp, processors);
 128
 129
 130                 }
 131         }
 132         catch(exception& e) {
 133                 m->errorOut(e, "SeqSummaryCommand", "SeqSummaryCommand");
 134                 exit(1);
 135         }
 136 }
 137 //**********************************************************************************************************************
 138
 139 void SeqSummaryCommand::help(){
 140         try {
 141                 m->mothurOut("The summary.seqs command reads a fastafile and summarizes the sequences.\n");
 142                 m->mothurOut("The summary.seqs command parameters are fasta, name and processors, fasta is required.\n");
 143                 m->mothurOut("The name parameter allows you to enter a name file associated with your fasta file. \n");
 144                 m->mothurOut("The summary.seqs command should be in the following format: \n");
 145                 m->mothurOut("summary.seqs(fasta=yourFastaFile, processors=2) \n");
 146                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 147         }
 148         catch(exception& e) {
 149                 m->errorOut(e, "SeqSummaryCommand", "help");
 150                 exit(1);
 151         }
 152 }
 153
 154 //***************************************************************************************************************
 155
 156 SeqSummaryCommand::~SeqSummaryCommand(){        /*      do nothing      */      }
 157
 158 //***************************************************************************************************************
 159
 160 int SeqSummaryCommand::execute(){
 161         try{
 162
 163                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 164
 165                 string summaryFile = outputDir + m->getSimpleName(fastafile) + ".summary";
 166
 167                 int numSeqs = 0;
 168
 169                 vector<int> startPosition;
 170                 vector<int> endPosition;
 171                 vector<int> seqLength;
 172                 vector<int> ambigBases;
 173                 vector<int> longHomoPolymer;
 174
 175                 if (namefile != "") { readNames(); }
 176
 177                 if (m->control_pressed) { return 0; }
 178
 179 #ifdef USE_MPI
 180                                 int pid, numSeqsPerProcessor;
 181                                 int tag = 2001;
 182                                 int startTag = 1; int endTag = 2; int lengthTag = 3; int baseTag = 4; int lhomoTag = 5;
 183                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 184                                 vector<unsigned long int> MPIPos;
 185
 186                                 MPI_Status status;
 187                                 MPI_Status statusOut;
 188                                 MPI_File inMPI;
 189                                 MPI_File outMPI;
 190                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 191                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 192
 193                                 char tempFileName[1024];
 194                                 strcpy(tempFileName, fastafile.c_str());
 195
 196                                 char sumFileName[1024];
 197                                 strcpy(sumFileName, summaryFile.c_str());
 198
 199                                 MPI_File_open(MPI_COMM_WORLD, tempFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 200                                 MPI_File_open(MPI_COMM_WORLD, sumFileName, outMode, MPI_INFO_NULL, &outMPI);
 201
 202                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); return 0;  }
 203
 204                                 if (pid == 0) { //you are the root process
 205                                                 //print header
 206                                                 string outputString = "seqname\tstart\tend\tnbases\tambigs\tpolymer\tnumSeqs\n";
 207                                                 int length = outputString.length();
 208                                                 char* buf2 = new char[length];
 209                                                 memcpy(buf2, outputString.c_str(), length);
 210
 211                                                 MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &statusOut);
 212                                                 delete buf2;
 213
 214                                                 MPIPos = m->setFilePosFasta(fastafile, numSeqs); //fills MPIPos, returns numSeqs
 215
 216                                                 for(int i = 1; i < processors; i++) {
 217                                                         MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 218                                                         MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 219                                                 }
 220
 221                                                 //figure out how many sequences you have to do
 222                                                 numSeqsPerProcessor = numSeqs / processors;
 223                                                 int startIndex =  pid * numSeqsPerProcessor;
 224                                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 225
 226                                                 //do your part
 227                                                 MPICreateSummary(startIndex, numSeqsPerProcessor, startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, inMPI, outMPI, MPIPos);
 228
 229                                 }else { //i am the child process
 230
 231                                         MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 232                                         MPIPos.resize(numSeqs+1);
 233                                         MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 234
 235                                         //figure out how many sequences you have to align
 236                                         numSeqsPerProcessor = numSeqs / processors;
 237                                         int startIndex =  pid * numSeqsPerProcessor;
 238                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 239
 240                                         //do your part
 241                                         MPICreateSummary(startIndex, numSeqsPerProcessor, startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, inMPI, outMPI, MPIPos);
 242                                 }
 243
 244                                 MPI_File_close(&inMPI);
 245                                 MPI_File_close(&outMPI);
 246                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 247
 248                                 if (pid == 0) {
 249                                         //get the info from the child processes
 250                                         for(int i = 1; i < processors; i++) {
 251                                                 int size;
 252                                                 MPI_Recv(&size, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
 253
 254                                                 vector<int> temp; temp.resize(size+1);
 255
 256                                                 for(int j = 0; j < 5; j++) {
 257
 258                                                         MPI_Recv(&temp[0], (size+1), MPI_INT, i, 2001, MPI_COMM_WORLD, &status);
 259                                                         int receiveTag = temp[temp.size()-1];  //child process added a int to the end to indicate what count this is for
 260
 261                                                         if (receiveTag == startTag) {
 262                                                                 for (int k = 0; k < size; k++) {                startPosition.push_back(temp[k]);       }
 263                                                         }else if (receiveTag == endTag) {
 264                                                                 for (int k = 0; k < size; k++) {                endPosition.push_back(temp[k]); }
 265                                                         }else if (receiveTag == lengthTag) {
 266                                                                 for (int k = 0; k < size; k++) {                seqLength.push_back(temp[k]);   }
 267                                                         }else if (receiveTag == baseTag) {
 268                                                                 for (int k = 0; k < size; k++) {                ambigBases.push_back(temp[k]);  }
 269                                                         }else if (receiveTag == lhomoTag) {
 270                                                                 for (int k = 0; k < size; k++) {                longHomoPolymer.push_back(temp[k]);     }
 271                                                         }
 272                                                 }
 273                                         }
 274
 275                                 }else{
 276
 277                                         //send my counts
 278                                         int size = startPosition.size();
 279                                         MPI_Send(&size, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 280
 281                                         startPosition.push_back(startTag);
 282                                         int ierr = MPI_Send(&(startPosition[0]), (size+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 283                                         endPosition.push_back(endTag);
 284                                         ierr = MPI_Send (&(endPosition[0]), (size+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 285                                         seqLength.push_back(lengthTag);
 286                                         ierr = MPI_Send(&(seqLength[0]), (size+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 287                                         ambigBases.push_back(baseTag);
 288                                         ierr = MPI_Send(&(ambigBases[0]), (size+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 289                                         longHomoPolymer.push_back(lhomoTag);
 290                                         ierr = MPI_Send(&(longHomoPolymer[0]), (size+1), MPI_INT, 0, 2001, MPI_COMM_WORLD);
 291                                 }
 292
 293                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 294 #else
 295                         vector<unsigned long int> positions = m->divideFile(fastafile, processors);
 296
 297                         for (int i = 0; i < (positions.size()-1); i++) {
 298                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 299                         }
 300
 301                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 302                                 if(processors == 1){
 303                                         numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile, lines[0]);
 304                                 }else{
 305                                         numSeqs = createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile);
 306
 307                                         rename((summaryFile + toString(processIDS[0]) + ".temp").c_str(), summaryFile.c_str());
 308                                         //append files
 309                                         for(int i=1;i<processors;i++){
 310                                                 m->appendFiles((summaryFile + toString(processIDS[i]) + ".temp"), summaryFile);
 311                                                 remove((summaryFile + toString(processIDS[i]) + ".temp").c_str());
 312                                         }
 313                                 }
 314
 315                                 if (m->control_pressed) {  return 0; }
 316                 #else
 317                                 numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile, lines[0]);
 318                                 if (m->control_pressed) {  return 0; }
 319                 #endif
 320 #endif
 321
 322                 #ifdef USE_MPI
 323                         if (pid == 0) {
 324                 #endif
 325
 326                 sort(startPosition.begin(), startPosition.end());
 327                 sort(endPosition.begin(), endPosition.end());
 328                 sort(seqLength.begin(), seqLength.end());
 329                 sort(ambigBases.begin(), ambigBases.end());
 330                 sort(longHomoPolymer.begin(), longHomoPolymer.end());
 331
 332                 int ptile0_25   = int(numSeqs * 0.025);
 333                 int ptile25             = int(numSeqs * 0.250);
 334                 int ptile50             = int(numSeqs * 0.500);
 335                 int ptile75             = int(numSeqs * 0.750);
 336                 int ptile97_5   = int(numSeqs * 0.975);
 337                 int ptile100    = numSeqs - 1;
 338
 339                 //to compensate for blank sequences that would result in startPosition and endPostion equalling -1
 340                 if (startPosition[0] == -1) {  startPosition[0] = 0;    }
 341                 if (endPosition[0] == -1)       {  endPosition[0] = 0;          }
 342
 343                 if (m->control_pressed) {  remove(summaryFile.c_str()); return 0; }
 344
 345                 m->mothurOutEndLine();
 346                 m->mothurOut("\t\tStart\tEnd\tNBases\tAmbigs\tPolymer"); m->mothurOutEndLine();
 347                 m->mothurOut("Minimum:\t" + toString(startPosition[0]) + "\t" + toString(endPosition[0]) + "\t" + toString(seqLength[0]) + "\t" + toString(ambigBases[0]) + "\t" + toString(longHomoPolymer[0])); m->mothurOutEndLine();
 348                 m->mothurOut("2.5%-tile:\t" + toString(startPosition[ptile0_25]) + "\t" + toString(endPosition[ptile0_25]) + "\t" + toString(seqLength[ptile0_25]) + "\t" + toString(ambigBases[ptile0_25]) + "\t"+ toString(longHomoPolymer[ptile0_25])); m->mothurOutEndLine();
 349                 m->mothurOut("25%-tile:\t" + toString(startPosition[ptile25]) + "\t" + toString(endPosition[ptile25]) + "\t" + toString(seqLength[ptile25]) + "\t" + toString(ambigBases[ptile25]) + "\t" + toString(longHomoPolymer[ptile25])); m->mothurOutEndLine();
 350                 m->mothurOut("Median: \t" + toString(startPosition[ptile50]) + "\t" + toString(endPosition[ptile50]) + "\t" + toString(seqLength[ptile50]) + "\t" + toString(ambigBases[ptile50]) + "\t" + toString(longHomoPolymer[ptile50])); m->mothurOutEndLine();
 351                 m->mothurOut("75%-tile:\t" + toString(startPosition[ptile75]) + "\t" + toString(endPosition[ptile75]) + "\t" + toString(seqLength[ptile75]) + "\t" + toString(ambigBases[ptile75]) + "\t" + toString(longHomoPolymer[ptile75])); m->mothurOutEndLine();
 352                 m->mothurOut("97.5%-tile:\t" + toString(startPosition[ptile97_5]) + "\t" + toString(endPosition[ptile97_5]) + "\t" + toString(seqLength[ptile97_5]) + "\t" + toString(ambigBases[ptile97_5]) + "\t" + toString(longHomoPolymer[ptile97_5])); m->mothurOutEndLine();
 353                 m->mothurOut("Maximum:\t" + toString(startPosition[ptile100]) + "\t" + toString(endPosition[ptile100]) + "\t" + toString(seqLength[ptile100]) + "\t" + toString(ambigBases[ptile100]) + "\t" + toString(longHomoPolymer[ptile100])); m->mothurOutEndLine();
 354                 if (namefile == "") {  m->mothurOut("# of Seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); }
 355                 else { m->mothurOut("# of unique seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(startPosition.size())); m->mothurOutEndLine(); }
 356
 357                 if (m->control_pressed) {  remove(summaryFile.c_str()); return 0; }
 358
 359                 m->mothurOutEndLine();
 360                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 361                 m->mothurOut(summaryFile); m->mothurOutEndLine();       outputNames.push_back(summaryFile); outputTypes["summary"].push_back(summaryFile);
 362                 m->mothurOutEndLine();
 363
 364                 #ifdef USE_MPI
 365                         }
 366                 #endif
 367
 368                 return 0;
 369         }
 370         catch(exception& e) {
 371                 m->errorOut(e, "SeqSummaryCommand", "execute");
 372                 exit(1);
 373         }
 374 }
 375 /**************************************************************************************/
 376 int SeqSummaryCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, string sumFile, linePair* filePos) {
 377         try {
 378
 379                 ofstream outSummary;
 380                 m->openOutputFile(sumFile, outSummary);
 381
 382                 //print header if you are process 0
 383                 if (filePos->start == 0) {
 384                         outSummary << "seqname\tstart\tend\tnbases\tambigs\tpolymer\tnumSeqs" << endl;
 385                 }
 386
 387                 ifstream in;
 388                 m->openInputFile(filename, in);
 389
 390                 in.seekg(filePos->start);
 391
 392                 bool done = false;
 393                 int count = 0;
 394
 395                 while (!done) {
 396
 397                         if (m->control_pressed) { in.close(); outSummary.close(); return 1; }
 398
 399                         Sequence current(in); m->gobble(in);
 400
 401                         if (current.getName() != "") {
 402
 403                                 int num = 1;
 404                                 if (namefile != "") {
 405                                         //make sure this sequence is in the namefile, else error
 406                                         map<string, int>::iterator it = nameMap.find(current.getName());
 407
 408                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 409                                         else { num = it->second; }
 410                                 }
 411
 412                                 //for each sequence this sequence represents
 413                                 for (int i = 0; i < num; i++) {
 414                                         startPosition.push_back(current.getStartPos());
 415                                         endPosition.push_back(current.getEndPos());
 416                                         seqLength.push_back(current.getNumBases());
 417                                         ambigBases.push_back(current.getAmbigBases());
 418                                         longHomoPolymer.push_back(current.getLongHomoPolymer());
 419                                 }
 420
 421                                 count++;
 422                                 outSummary << current.getName() << '\t';
 423                                 outSummary << current.getStartPos() << '\t' << current.getEndPos() << '\t';
 424                                 outSummary << current.getNumBases() << '\t' << current.getAmbigBases() << '\t';
 425                                 outSummary << current.getLongHomoPolymer() << '\t' << num << endl;
 426                         }
 427
 428                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 429                                 unsigned long int pos = in.tellg();
 430                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 431                         #else
 432                                 if (in.eof()) { break; }
 433                         #endif
 434
 435                         //report progress
 436                         //if((count) % 100 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 437                 }
 438                 //report progress
 439                 //if((count) % 100 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 440
 441                 in.close();
 442
 443                 return count;
 444         }
 445         catch(exception& e) {
 446                 m->errorOut(e, "SeqSummaryCommand", "driverCreateSummary");
 447                 exit(1);
 448         }
 449 }
 450 #ifdef USE_MPI
 451 /**************************************************************************************/
 452 int SeqSummaryCommand::MPICreateSummary(int start, int num, vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
 453         try {
 454
 455                 int pid;
 456                 MPI_Status status;
 457                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 458
 459                 for(int i=0;i<num;i++){
 460
 461                         if (m->control_pressed) { return 0; }
 462
 463                         //read next sequence
 464                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 465
 466                         char* buf4 = new char[length];
 467                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 468
 469                         string tempBuf = buf4;
 470                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 471                         istringstream iss (tempBuf,istringstream::in);
 472                         delete buf4;
 473
 474                         Sequence current(iss);
 475
 476                         if (current.getName() != "") {
 477
 478                                 int num = 1;
 479                                 if (namefile != "") {
 480                                         //make sure this sequence is in the namefile, else error
 481                                         map<string, int>::iterator it = nameMap.find(current.getName());
 482
 483                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 484                                         else { num = it->second; }
 485                                 }
 486
 487                                 //for each sequence this sequence represents
 488                                 for (int i = 0; i < num; i++) {
 489                                         startPosition.push_back(current.getStartPos());
 490                                         endPosition.push_back(current.getEndPos());
 491                                         seqLength.push_back(current.getNumBases());
 492                                         ambigBases.push_back(current.getAmbigBases());
 493                                         longHomoPolymer.push_back(current.getLongHomoPolymer());
 494                                 }
 495
 496                                 string outputString = current.getName() + "\t" + toString(current.getStartPos()) + "\t" + toString(current.getEndPos()) + "\t";
 497                                 outputString += toString(current.getNumBases()) + "\t" + toString(current.getAmbigBases()) + "\t" + toString(current.getLongHomoPolymer()) + "\t" + toString(num) + "\n";
 498
 499                                 //output to file
 500                                 length = outputString.length();
 501                                 char* buf3 = new char[length];
 502                                 memcpy(buf3, outputString.c_str(), length);
 503
 504                                 MPI_File_write_shared(outMPI, buf3, length, MPI_CHAR, &status);
 505                                 delete buf3;
 506                         }
 507                 }
 508
 509                 return 0;
 510         }
 511         catch(exception& e) {
 512                 m->errorOut(e, "SeqSummaryCommand", "MPICreateSummary");
 513                 exit(1);
 514         }
 515 }
 516 #endif
 517 /**************************************************************************************************/
 518 int SeqSummaryCommand::createProcessesCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, string sumFile) {
 519         try {
 520 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 521                 int process = 0;
 522                 int num = 0;
 523                 processIDS.clear();
 524
 525                 //loop through and create all the processes you want
 526                 while (process != processors) {
 527                         int pid = fork();
 528
 529                         if (pid > 0) {
 530                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 531                                 process++;
 532                         }else if (pid == 0){
 533                                 num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, sumFile + toString(getpid()) + ".temp", lines[process]);
 534
 535                                 //pass numSeqs to parent
 536                                 ofstream out;
 537                                 string tempFile = fastafile + toString(getpid()) + ".num.temp";
 538                                 m->openOutputFile(tempFile, out);
 539
 540                                 out << num << endl;
 541                                 out << startPosition.size() << endl;
 542                                 for (int k = 0; k < startPosition.size(); k++)          {               out << startPosition[k] << '\t'; }  out << endl;
 543                                 for (int k = 0; k < endPosition.size(); k++)            {               out << endPosition[k] << '\t'; }  out << endl;
 544                                 for (int k = 0; k < seqLength.size(); k++)                      {               out << seqLength[k] << '\t'; }  out << endl;
 545                                 for (int k = 0; k < ambigBases.size(); k++)                     {               out << ambigBases[k] << '\t'; }  out << endl;
 546                                 for (int k = 0; k < longHomoPolymer.size(); k++)        {               out << longHomoPolymer[k] << '\t'; }  out << endl;
 547
 548                                 out.close();
 549
 550                                 exit(0);
 551                         }else {
 552                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 553                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 554                                 exit(0);
 555                         }
 556                 }
 557
 558                 //force parent to wait until all the processes are done
 559                 for (int i=0;i<processors;i++) {
 560                         int temp = processIDS[i];
 561                         wait(&temp);
 562                 }
 563
 564                 //parent reads in and combine Filter info
 565                 for (int i = 0; i < processIDS.size(); i++) {
 566                         string tempFilename = fastafile + toString(processIDS[i]) + ".num.temp";
 567                         ifstream in;
 568                         m->openInputFile(tempFilename, in);
 569
 570                         int temp, tempNum;
 571                         in >> tempNum; m->gobble(in); num += tempNum;
 572                         in >> tempNum; m->gobble(in);
 573                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; startPosition.push_back(temp);              }               m->gobble(in);
 574                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; endPosition.push_back(temp);                }               m->gobble(in);
 575                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; seqLength.push_back(temp);                  }               m->gobble(in);
 576                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; ambigBases.push_back(temp);                 }               m->gobble(in);
 577                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; longHomoPolymer.push_back(temp);    }               m->gobble(in);
 578
 579                         in.close();
 580                         remove(tempFilename.c_str());
 581                 }
 582
 583                 return num;
 584 #endif
 585         }
 586         catch(exception& e) {
 587                 m->errorOut(e, "SeqSummaryCommand", "createProcessesCreateSummary");
 588                 exit(1);
 589         }
 590 }
 591 /**********************************************************************************************************************/
 592 int SeqSummaryCommand::readNames() {
 593         try {
 594                 //open input file
 595                 ifstream in;
 596                 m->openInputFile(namefile, in);
 597
 598                 while (!in.eof()) {
 599                         if (m->control_pressed) { break; }
 600
 601                         string firstCol, secondCol;
 602                         in >> firstCol >> secondCol; m->gobble(in);
 603
 604                         int num = m->getNumNames(secondCol);
 605
 606                         nameMap[firstCol] = num;
 607                 }
 608                 in.close();
 609
 610                 return 0;
 611
 612         }
 613         catch(exception& e) {
 614                 m->errorOut(e, "SeqSummaryCommand", "readNames");
 615                 exit(1);
 616         }
 617 }
 618
 619 /**********************************************************************************************************************/
 620
 621
 622