pairwiseseqscommand.cpp

   1 /*
   2  *  pairwiseseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/20/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "pairwiseseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 #include "gotohoverlap.hpp"
  14 #include "needlemanoverlap.hpp"
  15 #include "blastalign.hpp"
  16 #include "noalign.hpp"
  17 #include "nast.hpp"
  18
  19 #include "ignoregaps.h"
  20 #include "eachgapdist.h"
  21 #include "eachgapignore.h"
  22 #include "onegapdist.h"
  23 #include "onegapignore.h"
  24
  25
  26 //**********************************************************************************************************************
  27 vector<string> PairwiseSeqsCommand::getValidParameters(){
  28         try {
  29                 string AlignArray[] =  {"fasta","align","match","mismatch","gapopen","gapextend", "processors","calc","compress","cutoff","countends","output","outputdir","inputdir"};
  30                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  31                 return myArray;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "PairwiseSeqsCommand", "getValidParameters");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> PairwiseSeqsCommand::getRequiredParameters(){
  40         try {
  41                 string AlignArray[] =  {"fasta"};
  42                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "PairwiseSeqsCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> PairwiseSeqsCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "PairwiseSeqsCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 PairwiseSeqsCommand::PairwiseSeqsCommand(){
  63         try {
  64                 //initialize outputTypes
  65                 vector<string> tempOutNames;
  66                 outputTypes["phylip"] = tempOutNames;
  67                 outputTypes["column"] = tempOutNames;
  68         }
  69         catch(exception& e) {
  70                 m->errorOut(e, "PairwiseSeqsCommand", "PairwiseSeqsCommand");
  71                 exit(1);
  72         }
  73 }
  74 //**********************************************************************************************************************
  75 PairwiseSeqsCommand::PairwiseSeqsCommand(string option)  {
  76         try {
  77                 abort = false;
  78
  79                 //allow user to run help
  80                 if(option == "help") { help(); abort = true; }
  81
  82                 else {
  83
  84                         //valid paramters for this command
  85                         string AlignArray[] =  {"fasta","align","match","mismatch","gapopen","gapextend", "processors","cutoff","compress","calc","countends","output","outputdir","inputdir"};
  86                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  87
  88                         OptionParser parser(option);
  89                         map<string, string> parameters = parser.getParameters();
  90
  91                         ValidParameters validParameter("pairwise.seqs");
  92                         map<string, string>::iterator it;
  93
  94                         //check to make sure all parameters are valid for command
  95                         for (it = parameters.begin(); it != parameters.end(); it++) {
  96                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  97                         }
  98
  99                         //initialize outputTypes
 100                         vector<string> tempOutNames;
 101                         outputTypes["phylip"] = tempOutNames;
 102                         outputTypes["column"] = tempOutNames;
 103
 104                         //if the user changes the output directory command factory will send this info to us in the output parameter
 105                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 106
 107
 108                         //if the user changes the input directory command factory will send this info to us in the output parameter
 109                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 110
 111                         if (inputDir == "not found"){   inputDir = "";          }
 112
 113                         fastaFileName = validParameter.validFile(parameters, "fasta", false);
 114                         if (fastaFileName == "not found") { m->mothurOut("fasta is a required parameter for the pairwise.seqs command."); m->mothurOutEndLine(); abort = true;  }
 115                         else {
 116                                 m->splitAtDash(fastaFileName, fastaFileNames);
 117
 118                                 //go through files and make sure they are good, if not, then disregard them
 119                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 120
 121                                         if (inputDir != "") {
 122                                                 string path = m->hasPath(fastaFileNames[i]);
 123                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 124                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 125                                         }
 126
 127                                         int ableToOpen;
 128                                         ifstream in;
 129
 130                                         ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 131
 132                                         //if you can't open it, try default location
 133                                         if (ableToOpen == 1) {
 134                                                 if (m->getDefaultPath() != "") { //default path is set
 135                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 136                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 137                                                         ifstream in2;
 138                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 139                                                         in2.close();
 140                                                         fastaFileNames[i] = tryPath;
 141                                                 }
 142                                         }
 143
 144                                         //if you can't open it, try output location
 145                                         if (ableToOpen == 1) {
 146                                                 if (m->getOutputDir() != "") { //default path is set
 147                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 148                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 149                                                         ifstream in2;
 150                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 151                                                         in2.close();
 152                                                         fastaFileNames[i] = tryPath;
 153                                                 }
 154                                         }
 155
 156                                         in.close();
 157
 158                                         if (ableToOpen == 1) {
 159                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 160                                                 //erase from file list
 161                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 162                                                 i--;
 163                                         }
 164                                 }
 165
 166                                 //make sure there is at least one valid file left
 167                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 168                         }
 169
 170                         //check for optional parameter and set defaults
 171                         // ...at some point should added some additional type checking...
 172                         string temp;
 173                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 174                         convert(temp, match);
 175
 176                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 177                         convert(temp, misMatch);
 178
 179                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 180                         convert(temp, gapOpen);
 181
 182                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 183                         convert(temp, gapExtend);
 184
 185                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 186                         convert(temp, processors);
 187
 188                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
 189                         convert(temp, cutoff);
 190
 191                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
 192                         countends = m->isTrue(temp);
 193
 194                         temp = validParameter.validFile(parameters, "compress", false);         if(temp == "not found"){  temp = "F"; }
 195                         compress = m->isTrue(temp);
 196
 197                         align = validParameter.validFile(parameters, "align", false);           if (align == "not found"){      align = "needleman";    }
 198
 199                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "column"; }
 200                         if ((output != "column") && (output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are column, lt and square. I will use column."); m->mothurOutEndLine(); output = "column"; }
 201
 202                         calc = validParameter.validFile(parameters, "calc", false);
 203                         if (calc == "not found") { calc = "onegap";  }
 204                         else {
 205                                  if (calc == "default")  {  calc = "onegap";  }
 206                         }
 207                         m->splitAtDash(calc, Estimators);
 208
 209                         ValidCalculators validCalculator;
 210                         if (countends) {
 211                                 for (int i=0; i<Estimators.size(); i++) {
 212                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 213                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
 214                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
 215                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
 216                                         }
 217                                 }
 218                         }else {
 219                                 for (int i=0; i<Estimators.size(); i++) {
 220                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 221                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
 222                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
 223                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
 224                                         }
 225                                 }
 226                         }
 227                 }
 228
 229         }
 230         catch(exception& e) {
 231                 m->errorOut(e, "PairwiseSeqsCommand", "PairwiseSeqsCommand");
 232                 exit(1);
 233         }
 234 }
 235 //**********************************************************************************************************************
 236 PairwiseSeqsCommand::~PairwiseSeqsCommand(){}
 237 //**********************************************************************************************************************
 238
 239 void PairwiseSeqsCommand::help(){
 240         try {
 241                 m->mothurOut("The pairwise.seqs command reads a fasta file and creates distance matrix.\n");
 242                 m->mothurOut("The pairwise.seqs command parameters are fasta, align, match, mismatch, gapopen, gapextend, calc, output, cutoff and processors.\n");
 243                 m->mothurOut("The fasta parameter is required. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 244                 m->mothurOut("The align parameter allows you to specify the alignment method to use.  Your options are: gotoh, needleman, blast and noalign. The default is needleman.\n");
 245                 m->mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n");
 246                 m->mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n");
 247                 m->mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n");
 248                 m->mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n");
 249                 m->mothurOut("The calc parameter allows you to specify the method of calculating the distances.  Your options are: nogaps, onegap or eachgap. The default is onegap.\n");
 250                 m->mothurOut("The countends parameter allows you to specify whether to include terminal gaps in distance.  Your options are: T or F. The default is T.\n");
 251                 m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 252                 m->mothurOut("The output parameter allows you to specify format of your distance matrix. Options are column, lt, and square. The default is column.\n");
 253                 m->mothurOut("The compress parameter allows you to indicate that you want the resulting distance file compressed.  The default is false.\n");
 254                 m->mothurOut("The pairwise.seqs command should be in the following format: \n");
 255                 m->mothurOut("pairwise.seqs(fasta=yourfastaFile, align=yourAlignmentMethod) \n");
 256                 m->mothurOut("Example pairwise.seqs(fasta=candidate.fasta, align=blast)\n");
 257                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "PairwiseSeqsCommand", "help");
 261                 exit(1);
 262         }
 263 }
 264
 265
 266 //**********************************************************************************************************************
 267
 268 int PairwiseSeqsCommand::execute(){
 269         try {
 270                 if (abort == true) {    return 0;       }
 271
 272                 int longestBase = 2000; //will need to update this in driver if we find sequences with more bases.  hardcoded so we don't have the pre-read user fasta file.
 273
 274                 if(align == "gotoh")                    {       alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase);                 }
 275                 else if(align == "needleman")   {       alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);                                }
 276                 else if(align == "blast")               {       alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch);            }
 277                 else if(align == "noalign")             {       alignment = new NoAlign();                                                                                                      }
 278                 else {
 279                         m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
 280                         m->mothurOutEndLine();
 281                         alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
 282                 }
 283
 284                 cutoff += 0.005;
 285
 286                 for (int s = 0; s < fastaFileNames.size(); s++) {
 287                         if (m->control_pressed) { outputTypes.clear(); return 0; }
 288
 289                         m->mothurOut("Processing sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 290
 291                         if (outputDir == "") {  outputDir += m->hasPath(fastaFileNames[s]); }
 292
 293                         ifstream inFASTA;
 294                         m->openInputFile(fastaFileNames[s], inFASTA);
 295                         alignDB = SequenceDB(inFASTA);
 296                         inFASTA.close();
 297
 298                         int numSeqs = alignDB.getNumSeqs();
 299                         int startTime = time(NULL);
 300                         string outputFile = "";
 301
 302                         if (output == "lt") { //does the user want lower triangle phylip formatted file
 303                                 outputFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "phylip.dist";
 304                                 remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
 305                         }else if (output == "column") { //user wants column format
 306                                 outputFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "dist";
 307                                 outputTypes["column"].push_back(outputFile);
 308                                 remove(outputFile.c_str());
 309                         }else { //assume square
 310                                 outputFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "square.dist";
 311                                 remove(outputFile.c_str());
 312                                 outputTypes["phylip"].push_back(outputFile);
 313                         }
 314
 315                         #ifdef USE_MPI
 316
 317                         int pid, start, end;
 318                         int tag = 2001;
 319
 320                         MPI_Status status;
 321                         MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 322                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 323
 324                         //each process gets where it should start and stop in the file
 325                         if (output != "square") {
 326                                 start = int (sqrt(float(pid)/float(processors)) * numSeqs);
 327                                 end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
 328                         }else{
 329                                 start = int ((float(pid)/float(processors)) * numSeqs);
 330                                 end = int ((float(pid+1)/float(processors)) * numSeqs);
 331                         }
 332
 333                         if (output == "column") {
 334                                 MPI_File outMPI;
 335                                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 336
 337                         char filename[1024];
 338                                 strcpy(filename, outputFile.c_str());
 339
 340                                 MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
 341
 342                                 if (pid == 0) { //you are the root process
 343
 344                                         //do your part
 345                                         string outputMyPart;
 346
 347                                         driverMPI(start, end, outMPI, cutoff);
 348
 349                                         if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI);  remove(outputFile.c_str()); delete distCalculator;  return 0; }
 350
 351                                         //wait on chidren
 352                                         for(int i = 1; i < processors; i++) {
 353                                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);   remove(outputFile.c_str()); delete distCalculator;  return 0; }
 354
 355                                                 char buf[5];
 356                                                 MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 357                                         }
 358                                 }else { //you are a child process
 359                                         //do your part
 360                                         driverMPI(start, end, outMPI, cutoff);
 361
 362                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  remove(outputFile.c_str()); delete distCalculator;  return 0; }
 363
 364                                         char buf[5];
 365                                         strcpy(buf, "done");
 366                                         //tell parent you are done.
 367                                         MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 368                                 }
 369
 370                                 MPI_File_close(&outMPI);
 371
 372                         }else { //lower triangle format
 373                                 if (pid == 0) { //you are the root process
 374
 375                                         //do your part
 376                                         string outputMyPart;
 377                                         unsigned long int mySize;
 378
 379                                         if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
 380                                         else { driverMPI(start, end, outputFile, mySize, output); }
 381
 382                                         if (m->control_pressed) {  outputTypes.clear();   remove(outputFile.c_str()); delete distCalculator;  return 0; }
 383
 384                                         int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 385                                         MPI_File outMPI;
 386                                         MPI_File inMPI;
 387
 388                                         char filename[1024];
 389                                         strcpy(filename, outputFile.c_str());
 390
 391                                         MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 392
 393                                         //wait on chidren
 394                                         for(int b = 1; b < processors; b++) {
 395                                                 unsigned long int fileSize;
 396
 397                                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  remove(outputFile.c_str());  delete distCalculator;  return 0; }
 398
 399                                                 MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
 400
 401                                                 string outTemp = outputFile + toString(b) + ".temp";
 402
 403                                                 char* buf = new char[outTemp.length()];
 404                                                 memcpy(buf, outTemp.c_str(), outTemp.length());
 405
 406                                                 MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 407                                                 delete buf;
 408
 409                                                 int count = 0;
 410                                                 while (count < fileSize) {
 411                                                         char buf2[1];
 412                                                         MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
 413                                                         MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
 414                                                         count += 1;
 415                                                 }
 416
 417                                                 MPI_File_close(&inMPI); //deleted on close
 418                                         }
 419
 420                                         MPI_File_close(&outMPI);
 421                                 }else { //you are a child process
 422                                         //do your part
 423                                         unsigned long int size;
 424                                         if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
 425                                         else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
 426
 427                                         if (m->control_pressed) { delete distCalculator;  return 0; }
 428
 429                                         //tell parent you are done.
 430                                         MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
 431                                 }
 432                         }
 433                         MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 434         #else
 435
 436                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 437                         //if you don't need to fork anything
 438                         if(processors == 1){
 439                                 if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 440                                 else { driver(0, numSeqs, outputFile, "square");  }
 441                         }else{ //you have multiple processors
 442
 443                                 for (int i = 0; i < processors; i++) {
 444                                         lines.push_back(new linePair());
 445                                         if (output != "square") {
 446                                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 447                                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 448                                         }else{
 449                                                 lines[i]->start = int ((float(i)/float(processors)) * numSeqs);
 450                                                 lines[i]->end = int ((float(i+1)/float(processors)) * numSeqs);
 451                                         }
 452                                 }
 453
 454                                 createProcesses(outputFile);
 455
 456                                 map<int, int>::iterator it = processIDS.begin();
 457                                 rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 458                                 it++;
 459
 460                                 //append and remove temp files
 461                                 for (; it != processIDS.end(); it++) {
 462                                         m->appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 463                                         remove((outputFile + toString(it->second) + ".temp").c_str());
 464                                 }
 465                         }
 466                 #else
 467                         //ifstream inFASTA;
 468                         if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 469                         else { driver(0, numSeqs, outputFile, "square");  }
 470                 #endif
 471
 472         #endif
 473                         if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 474
 475                         #ifdef USE_MPI
 476                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 477
 478                                 if (pid == 0) { //only one process should output to screen
 479                         #endif
 480
 481                         ifstream fileHandle;
 482                         fileHandle.open(outputFile.c_str());
 483                         if(fileHandle) {
 484                                 m->gobble(fileHandle);
 485                                 if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff.");  m->mothurOutEndLine(); }
 486                         }
 487
 488                         if (compress) {
 489                                 m->mothurOut("Compressing..."); m->mothurOutEndLine();
 490                                 m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 491                                 system(("gzip -v " + outputFile).c_str());
 492                                 outputNames.push_back(outputFile + ".gz");
 493                         }else { outputNames.push_back(outputFile); }
 494
 495                         #ifdef USE_MPI
 496                                 }
 497                         #endif
 498
 499                         m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 500
 501                         if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 502                 }
 503
 504                 delete distCalculator;
 505
 506                 m->mothurOutEndLine();
 507                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 508                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 509                 m->mothurOutEndLine();
 510
 511
 512                 return 0;
 513         }
 514         catch(exception& e) {
 515                 m->errorOut(e, "PairwiseSeqsCommand", "execute");
 516                 exit(1);
 517         }
 518 }
 519
 520 /**************************************************************************************************/
 521 void PairwiseSeqsCommand::createProcesses(string filename) {
 522         try {
 523 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 524                 int process = 0;
 525                 processIDS.clear();
 526
 527                 //loop through and create all the processes you want
 528                 while (process != processors) {
 529                         int pid = fork();
 530
 531                         if (pid > 0) {
 532                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 533                                 process++;
 534                         }else if (pid == 0){
 535                                 if (output != "square") {  driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff); }
 536                                 else { driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", "square"); }
 537                                 exit(0);
 538                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 539                 }
 540
 541                 //force parent to wait until all the processes are done
 542                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 543                         int temp = it->second;
 544                         wait(&temp);
 545                 }
 546 #endif
 547         }
 548         catch(exception& e) {
 549                 m->errorOut(e, "PairwiseSeqsCommand", "createProcesses");
 550                 exit(1);
 551         }
 552 }
 553
 554 /**************************************************************************************************/
 555 /////// need to fix to work with calcs and sequencedb
 556 int PairwiseSeqsCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 557         try {
 558
 559                 int startTime = time(NULL);
 560
 561                 //column file
 562                 ofstream outFile(dFileName.c_str(), ios::trunc);
 563                 outFile.setf(ios::fixed, ios::showpoint);
 564                 outFile << setprecision(4);
 565
 566                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 567
 568                 for(int i=startLine;i<endLine;i++){
 569                         if(output == "lt")      {
 570                                 string name = alignDB.get(i).getName();
 571                                 if (name.length() < 10) { //pad with spaces to make compatible
 572                                         while (name.length() < 10) {  name += " ";  }
 573                                 }
 574                                 outFile << name << '\t';
 575                         }
 576
 577                         for(int j=0;j<i;j++){
 578
 579                                 if (m->control_pressed) { outFile.close(); return 0;  }
 580
 581                                 if (alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
 582                                         alignment->resize(alignDB.get(i).getUnaligned().length()+1);
 583                                 }
 584
 585                                 if (alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
 586                                         alignment->resize(alignDB.get(j).getUnaligned().length()+1);
 587                                 }
 588
 589                                 Sequence* seqI = new Sequence(alignDB.get(i).getName(), alignDB.get(i).getAligned());
 590                                 Sequence* seqJ = new Sequence(alignDB.get(j).getName(), alignDB.get(j).getAligned());
 591
 592                                 Nast(alignment, seqI, seqJ);
 593
 594                                 distCalculator->calcDist(*seqI, *seqJ);
 595                                 double dist = distCalculator->getDist();
 596
 597                                 delete seqI; delete seqJ;
 598
 599                                 if(dist <= cutoff){
 600                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 601                                 }
 602                                 if (output == "lt") {  outFile << dist << '\t'; }
 603                         }
 604
 605                         if (output == "lt") { outFile << endl; }
 606
 607                         if(i % 100 == 0){
 608                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 609                         }
 610
 611                 }
 612                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 613
 614                 outFile.close();
 615
 616                 return 1;
 617         }
 618         catch(exception& e) {
 619                 m->errorOut(e, "PairwiseSeqsCommand", "driver");
 620                 exit(1);
 621         }
 622 }
 623 /**************************************************************************************************/
 624 /////// need to fix to work with calcs and sequencedb
 625 int PairwiseSeqsCommand::driver(int startLine, int endLine, string dFileName, string square){
 626         try {
 627
 628                 int startTime = time(NULL);
 629
 630                 //column file
 631                 ofstream outFile(dFileName.c_str(), ios::trunc);
 632                 outFile.setf(ios::fixed, ios::showpoint);
 633                 outFile << setprecision(4);
 634
 635                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 636
 637                 for(int i=startLine;i<endLine;i++){
 638
 639                         string name = alignDB.get(i).getName();
 640                         //pad with spaces to make compatible
 641                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 642
 643                         outFile << name << '\t';
 644
 645                         for(int j=0;j<alignDB.getNumSeqs();j++){
 646
 647                                 if (m->control_pressed) { outFile.close(); return 0;  }
 648
 649                                 if (alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
 650                                         alignment->resize(alignDB.get(i).getUnaligned().length()+1);
 651                                 }
 652
 653                                 if (alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
 654                                         alignment->resize(alignDB.get(j).getUnaligned().length()+1);
 655                                 }
 656
 657                                 Sequence* seqI = new Sequence(alignDB.get(i).getName(), alignDB.get(i).getAligned());
 658                                 Sequence* seqJ = new Sequence(alignDB.get(j).getName(), alignDB.get(j).getAligned());
 659
 660                                 Nast(alignment, seqI, seqJ);
 661
 662                                 distCalculator->calcDist(*seqI, *seqJ);
 663                                 double dist = distCalculator->getDist();
 664
 665                                 delete seqI; delete seqJ;
 666
 667                                 outFile << dist << '\t';
 668                         }
 669
 670                         outFile << endl;
 671
 672                         if(i % 100 == 0){
 673                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 674                         }
 675
 676                 }
 677                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 678
 679                 outFile.close();
 680
 681                 return 1;
 682         }
 683         catch(exception& e) {
 684                 m->errorOut(e, "PairwiseSeqsCommand", "driver");
 685                 exit(1);
 686         }
 687 }
 688 #ifdef USE_MPI
 689 /**************************************************************************************************/
 690 /////// need to fix to work with calcs and sequencedb
 691 int PairwiseSeqsCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 692         try {
 693                 MPI_Status status;
 694                 int startTime = time(NULL);
 695
 696                 string outputString = "";
 697
 698                 for(int i=startLine;i<endLine;i++){
 699
 700                         for(int j=0;j<i;j++){
 701
 702                                 if (m->control_pressed) {  return 0;  }
 703
 704                                 if (alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
 705                                         alignment->resize(alignDB.get(i).getUnaligned().length()+1);
 706                                 }
 707
 708                                 if (alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
 709                                         alignment->resize(alignDB.get(j).getUnaligned().length()+1);
 710                                 }
 711
 712                                 Sequence* seqI = new Sequence(alignDB.get(i).getName(), alignDB.get(i).getAligned());
 713                                 Sequence* seqJ = new Sequence(alignDB.get(j).getName(), alignDB.get(j).getAligned());
 714
 715                                 Nast(alignment, seqI, seqJ);
 716
 717                                 distCalculator->calcDist(*seqI, *seqJ);
 718                                 double dist = distCalculator->getDist();
 719
 720                                 delete seqI; delete seqJ;
 721
 722                                 if(dist <= cutoff){
 723                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 724                                 }
 725                         }
 726
 727                         if(i % 100 == 0){
 728                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 729                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 730                         }
 731
 732
 733                         //send results to parent
 734                         int length = outputString.length();
 735
 736                         char* buf = new char[length];
 737                         memcpy(buf, outputString.c_str(), length);
 738
 739                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 740                         outputString = "";
 741                         delete buf;
 742
 743                 }
 744
 745                 return 1;
 746         }
 747         catch(exception& e) {
 748                 m->errorOut(e, "PairwiseSeqsCommand", "driverMPI");
 749                 exit(1);
 750         }
 751 }
 752 /**************************************************************************************************/
 753 /////// need to fix to work with calcs and sequencedb
 754 int PairwiseSeqsCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 755         try {
 756                 MPI_Status status;
 757
 758                 MPI_File outMPI;
 759                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 760
 761                 char filename[1024];
 762                 strcpy(filename, file.c_str());
 763
 764                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 765
 766                 int startTime = time(NULL);
 767
 768                 string outputString = "";
 769                 size = 0;
 770
 771                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 772
 773                 for(int i=startLine;i<endLine;i++){
 774
 775                         string name = alignDB.get(i).getName();
 776                         if (name.length() < 10) { //pad with spaces to make compatible
 777                                 while (name.length() < 10) {  name += " ";  }
 778                         }
 779                         outputString += name + "\t";
 780
 781                         for(int j=0;j<i;j++){
 782
 783                                 if (m->control_pressed) {  return 0;  }
 784
 785                                 if (alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
 786                                         alignment->resize(alignDB.get(i).getUnaligned().length()+1);
 787                                 }
 788
 789                                 if (alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
 790                                         alignment->resize(alignDB.get(j).getUnaligned().length()+1);
 791                                 }
 792
 793                                 Sequence* seqI = new Sequence(alignDB.get(i).getName(), alignDB.get(i).getAligned());
 794                                 Sequence* seqJ = new Sequence(alignDB.get(j).getName(), alignDB.get(j).getAligned());
 795
 796                                 Nast(alignment, seqI, seqJ);
 797
 798                                 distCalculator->calcDist(*seqI, *seqJ);
 799                                 double dist = distCalculator->getDist();
 800
 801                                 delete seqI; delete seqJ;
 802
 803                                 outputString += toString(dist) + "\t";
 804                         }
 805
 806                         outputString += "\n";
 807
 808                         //send results to parent
 809                         int length = outputString.length();
 810                         char* buf = new char[length];
 811                         memcpy(buf, outputString.c_str(), length);
 812
 813                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 814                         size += outputString.length();
 815                         outputString = "";
 816                         delete buf;
 817                 }
 818
 819                 MPI_File_close(&outMPI);
 820
 821                 return 1;
 822         }
 823         catch(exception& e) {
 824                 m->errorOut(e, "PairwiseSeqsCommand", "driverMPI");
 825                 exit(1);
 826         }
 827 }
 828 /**************************************************************************************************/
 829 /////// need to fix to work with calcs and sequencedb
 830 int PairwiseSeqsCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 831         try {
 832                 MPI_Status status;
 833
 834                 MPI_File outMPI;
 835                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 836
 837                 char filename[1024];
 838                 strcpy(filename, file.c_str());
 839
 840                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 841
 842                 int startTime = time(NULL);
 843
 844                 string outputString = "";
 845                 size = 0;
 846
 847                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 848
 849                 for(int i=startLine;i<endLine;i++){
 850
 851                         string name = alignDB.get(i).getName();
 852                         if (name.length() < 10) { //pad with spaces to make compatible
 853                                 while (name.length() < 10) {  name += " ";  }
 854                         }
 855                         outputString += name + "\t";
 856
 857                         for(int j=0;j<alignDB.getNumSeqs();j++){
 858
 859                                 if (m->control_pressed) {  return 0;  }
 860
 861                                 if (alignDB.get(i).getUnaligned().length() > alignment->getnRows()) {
 862                                         alignment->resize(alignDB.get(i).getUnaligned().length()+1);
 863                                 }
 864
 865                                 if (alignDB.get(j).getUnaligned().length() > alignment->getnRows()) {
 866                                         alignment->resize(alignDB.get(j).getUnaligned().length()+1);
 867                                 }
 868
 869                                 Sequence* seqI = new Sequence(alignDB.get(i).getName(), alignDB.get(i).getAligned());
 870                                 Sequence* seqJ = new Sequence(alignDB.get(j).getName(), alignDB.get(j).getAligned());
 871
 872                                 Nast(alignment, seqI, seqJ);
 873
 874                                 distCalculator->calcDist(*seqI, *seqJ);
 875                                 double dist = distCalculator->getDist();
 876
 877                                 delete seqI; delete seqJ;
 878
 879                                 outputString += toString(dist) + "\t";
 880                         }
 881
 882                         outputString += "\n";
 883
 884                         //send results to parent
 885                         int length = outputString.length();
 886                         char* buf = new char[length];
 887                         memcpy(buf, outputString.c_str(), length);
 888
 889                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 890                         size += outputString.length();
 891                         outputString = "";
 892                         delete buf;
 893                 }
 894
 895                 MPI_File_close(&outMPI);
 896
 897                 return 1;
 898         }
 899         catch(exception& e) {
 900                 m->errorOut(e, "PairwiseSeqsCommand", "driverMPI");
 901                 exit(1);
 902         }
 903 }
 904 #endif
 905 /**************************************************************************************************/
 906