distancecommand.cpp

   1 /*
   2  *  distancecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/7/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "distancecommand.h"
  11 #include "ignoregaps.h"
  12 #include "eachgapdist.h"
  13 #include "eachgapignore.h"
  14 #include "onegapdist.h"
  15 #include "onegapignore.h"
  16
  17
  18 //**********************************************************************************************************************
  19 vector<string> DistanceCommand::getValidParameters(){
  20         try {
  21                 string Array[] =  {"fasta","oldfasta","column", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir","compress"};
  22                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  23                 return myArray;
  24         }
  25         catch(exception& e) {
  26                 m->errorOut(e, "DistanceCommand", "getValidParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 DistanceCommand::DistanceCommand(){
  32         try {
  33                 abort = true;
  34                 //initialize outputTypes
  35                 vector<string> tempOutNames;
  36                 outputTypes["phylip"] = tempOutNames;
  37                 outputTypes["column"] = tempOutNames;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "DistanceCommand", "DistanceCommand");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 vector<string> DistanceCommand::getRequiredParameters(){
  46         try {
  47                 string Array[] =  {"fasta"};
  48                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  49                 return myArray;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "DistanceCommand", "getRequiredParameters");
  53                 exit(1);
  54         }
  55 }
  56 //**********************************************************************************************************************
  57 vector<string> DistanceCommand::getRequiredFiles(){
  58         try {
  59                 vector<string> myArray;
  60                 return myArray;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "DistanceCommand", "getRequiredFiles");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 DistanceCommand::DistanceCommand(string option) {
  69         try {
  70                 abort = false;
  71                 Estimators.clear();
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; }
  75
  76                 else {
  77                         //valid paramters for this command
  78                         string Array[] =  {"fasta","oldfasta","column", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir","compress"};
  79
  80                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  81
  82                         OptionParser parser(option);
  83                         map<string, string> parameters = parser.getParameters();
  84
  85                         ValidParameters validParameter("dist.seqs");
  86                         map<string, string>::iterator it2;
  87
  88                         //check to make sure all parameters are valid for command
  89                         for (it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  90                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  91                         }
  92
  93                         //initialize outputTypes
  94                         vector<string> tempOutNames;
  95                         outputTypes["phylip"] = tempOutNames;
  96                         outputTypes["column"] = tempOutNames;
  97
  98                         //if the user changes the input directory command factory will send this info to us in the output parameter
  99                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 100                         if (inputDir == "not found"){   inputDir = "";          }
 101                         else {
 102                                 string path;
 103                                 it2 = parameters.find("fasta");
 104                                 //user has given a template file
 105                                 if(it2 != parameters.end()){
 106                                         path = m->hasPath(it2->second);
 107                                         //if the user has not given a path then, add inputdir. else leave path alone.
 108                                         if (path == "") {       parameters["fasta"] = inputDir + it2->second;           }
 109                                 }
 110
 111                                 it2 = parameters.find("oldfasta");
 112                                 //user has given a template file
 113                                 if(it2 != parameters.end()){
 114                                         path = m->hasPath(it2->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["oldfasta"] = inputDir + it2->second;                }
 117                                 }
 118
 119                                 it2 = parameters.find("column");
 120                                 //user has given a template file
 121                                 if(it2 != parameters.end()){
 122                                         path = m->hasPath(it2->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["column"] = inputDir + it2->second;          }
 125                                 }
 126                         }
 127
 128                         //check for required parameters
 129                         fastafile = validParameter.validFile(parameters, "fasta", true);
 130                         if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the dist.seqs command."); m->mothurOutEndLine(); abort = true; }
 131                         else if (fastafile == "not open") { abort = true; }
 132                         else{
 133                                 ifstream inFASTA;
 134                                 m->openInputFile(fastafile, inFASTA);
 135                                 alignDB = SequenceDB(inFASTA);
 136                                 inFASTA.close();
 137                         }
 138
 139                         oldfastafile = validParameter.validFile(parameters, "oldfasta", true);
 140                         if (oldfastafile == "not found") { oldfastafile = ""; }
 141                         else if (oldfastafile == "not open") { abort = true; }
 142
 143                         column = validParameter.validFile(parameters, "column", true);
 144                         if (column == "not found") { column = ""; }
 145                         else if (column == "not open") { abort = true; }
 146
 147                         //if the user changes the output directory command factory will send this info to us in the output parameter
 148                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 149                                 outputDir = "";
 150                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 151                         }
 152
 153                         //check for optional parameter and set defaults
 154                         // ...at some point should added some additional type checking...
 155                         calc = validParameter.validFile(parameters, "calc", false);
 156                         if (calc == "not found") { calc = "onegap";  }
 157                         else {
 158                                  if (calc == "default")  {  calc = "onegap";  }
 159                         }
 160                         m->splitAtDash(calc, Estimators);
 161
 162                         string temp;
 163                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
 164                         convert(temp, countends);
 165
 166                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
 167                         convert(temp, cutoff);
 168
 169                         temp = validParameter.validFile(parameters, "processors", false);       if(temp == "not found"){        temp = "1"; }
 170                         convert(temp, processors);
 171
 172                         temp = validParameter.validFile(parameters, "compress", false);         if(temp == "not found"){  temp = "F"; }
 173                         convert(temp, compress);
 174
 175                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "column"; }
 176
 177                         if (((column != "") && (oldfastafile == "")) || ((column == "") && (oldfastafile != ""))) { m->mothurOut("If you provide column or oldfasta, you must provide both."); m->mothurOutEndLine(); abort=true; }
 178
 179                         if ((column != "") && (oldfastafile != "") && (output != "column")) { m->mothurOut("You have provided column and oldfasta, indicating you want to append distances to your column file. Your output must be in column format to do so."); m->mothurOutEndLine(); abort=true; }
 180
 181                         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"; }
 182
 183                         ValidCalculators validCalculator;
 184
 185                         if (m->isTrue(countends) == true) {
 186                                 for (int i=0; i<Estimators.size(); i++) {
 187                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 188                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
 189                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
 190                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
 191                                         }
 192                                 }
 193                         }else {
 194                                 for (int i=0; i<Estimators.size(); i++) {
 195                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 196                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
 197                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
 198                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
 199                                         }
 200                                 }
 201                         }
 202
 203                 }
 204
 205         }
 206         catch(exception& e) {
 207                 m->errorOut(e, "DistanceCommand", "DistanceCommand");
 208                 exit(1);
 209         }
 210 }
 211
 212 //**********************************************************************************************************************
 213
 214 DistanceCommand::~DistanceCommand(){}
 215
 216 //**********************************************************************************************************************
 217
 218 void DistanceCommand::help(){
 219         try {
 220                 m->mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
 221                 m->mothurOut("The dist.seqs command parameters are fasta, oldfasta, column, calc, countends, output, compress, cutoff and processors.  \n");
 222                 m->mothurOut("The fasta parameter is required.\n");
 223                 m->mothurOut("The oldfasta and column parameters allow you to append the distances calculated to the column file.\n");
 224                 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");
 225                 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");
 226                 m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 227                 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");
 228                 m->mothurOut("The processors parameter allows you to specify number of processors to use.  The default is 1.\n");
 229                 m->mothurOut("The compress parameter allows you to indicate that you want the resulting distance file compressed.  The default is false.\n");
 230                 m->mothurOut("The dist.seqs command should be in the following format: \n");
 231                 m->mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
 232                 m->mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
 233                 m->mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
 234         }
 235         catch(exception& e) {
 236                 m->errorOut(e, "DistanceCommand", "help");
 237                 exit(1);
 238         }
 239 }
 240 //**********************************************************************************************************************
 241
 242 int DistanceCommand::execute(){
 243         try {
 244
 245                 if (abort == true) { return 0; }
 246
 247                 int startTime = time(NULL);
 248
 249                 //save number of new sequence
 250                 numNewFasta = alignDB.getNumSeqs();
 251
 252                 //sanity check the oldfasta and column file as well as add oldfasta sequences to alignDB
 253                 if ((oldfastafile != "") && (column != ""))  {  if (!(sanityCheck())) { return 0; }  }
 254
 255                 if (m->control_pressed) { return 0; }
 256
 257                 int numSeqs = alignDB.getNumSeqs();
 258                 cutoff += 0.005;
 259
 260                 string outputFile;
 261
 262                 if (output == "lt") { //does the user want lower triangle phylip formatted file
 263                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
 264                         remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
 265
 266                         //output numSeqs to phylip formatted dist file
 267                 }else if (output == "column") { //user wants column format
 268                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
 269                         outputTypes["column"].push_back(outputFile);
 270
 271                         //so we don't accidentally overwrite
 272                         if (outputFile == column) {
 273                                 string tempcolumn = column + ".old";
 274                                 rename(column.c_str(), tempcolumn.c_str());
 275                         }
 276
 277                         remove(outputFile.c_str());
 278                 }else { //assume square
 279                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
 280                         remove(outputFile.c_str());
 281                         outputTypes["phylip"].push_back(outputFile);
 282                 }
 283
 284
 285 #ifdef USE_MPI
 286
 287                 int pid, start, end;
 288                 int tag = 2001;
 289
 290                 MPI_Status status;
 291                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 292                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 293
 294                 //each process gets where it should start and stop in the file
 295                 if (output != "square") {
 296                         start = int (sqrt(float(pid)/float(processors)) * numSeqs);
 297                         end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
 298                 }else{
 299                         start = int ((float(pid)/float(processors)) * numSeqs);
 300                         end = int ((float(pid+1)/float(processors)) * numSeqs);
 301                 }
 302
 303                 if (output == "column") {
 304                         MPI_File outMPI;
 305                         int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 306
 307                         //char* filename = new char[outputFile.length()];
 308                         //memcpy(filename, outputFile.c_str(), outputFile.length());
 309
 310                         char filename[1024];
 311                         strcpy(filename, outputFile.c_str());
 312
 313                         MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
 314                         //delete filename;
 315
 316                         if (pid == 0) { //you are the root process
 317
 318                                 //do your part
 319                                 string outputMyPart;
 320
 321                                 driverMPI(start, end, outMPI, cutoff);
 322
 323                                 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator;  return 0; }
 324
 325                                 //wait on chidren
 326                                 for(int i = 1; i < processors; i++) {
 327                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 328
 329                                         char buf[5];
 330                                         MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 331                                 }
 332                         }else { //you are a child process
 333                                 //do your part
 334                                 driverMPI(start, end, outMPI, cutoff);
 335
 336                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 337
 338                                 char buf[5];
 339                                 strcpy(buf, "done");
 340                                 //tell parent you are done.
 341                                 MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 342                         }
 343
 344                         MPI_File_close(&outMPI);
 345
 346                 }else { //lower triangle format
 347                         if (pid == 0) { //you are the root process
 348
 349                                 //do your part
 350                                 string outputMyPart;
 351                                 unsigned long int mySize;
 352
 353                                 if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
 354                                 else { driverMPI(start, end, outputFile, mySize, output); }
 355
 356                                 if (m->control_pressed) {  outputTypes.clear();  delete distCalculator;  return 0; }
 357
 358                                 int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 359                                 MPI_File outMPI;
 360                                 MPI_File inMPI;
 361
 362                                 //char* filename = new char[outputFile.length()];
 363                                 //memcpy(filename, outputFile.c_str(), outputFile.length());
 364
 365                                 char filename[1024];
 366                                 strcpy(filename, outputFile.c_str());
 367
 368                                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 369                                 //delete filename;
 370
 371                                 //wait on chidren
 372                                 for(int b = 1; b < processors; b++) {
 373                                         unsigned long int fileSize;
 374
 375                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 376
 377                                         MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
 378
 379                                         string outTemp = outputFile + toString(b) + ".temp";
 380
 381                                         char* buf = new char[outTemp.length()];
 382                                         memcpy(buf, outTemp.c_str(), outTemp.length());
 383
 384                                         MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 385                                         delete buf;
 386
 387                                         int count = 0;
 388                                         while (count < fileSize) {
 389                                                 char buf2[1];
 390                                                 MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
 391                                                 MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
 392                                                 count += 1;
 393                                         }
 394
 395                                         MPI_File_close(&inMPI); //deleted on close
 396                                 }
 397
 398                                 MPI_File_close(&outMPI);
 399                         }else { //you are a child process
 400                                 //do your part
 401                                 unsigned long int size;
 402                                 if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
 403                                 else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
 404
 405                                 if (m->control_pressed) { delete distCalculator;  return 0; }
 406
 407                                 //tell parent you are done.
 408                                 MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
 409                         }
 410                 }
 411                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 412 #else
 413
 414         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 415                 //if you don't need to fork anything
 416                 if(processors == 1){
 417                         if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 418                         else { driver(0, numSeqs, outputFile, "square");  }
 419                 }else{ //you have multiple processors
 420
 421                         unsigned long int numDists = 0;
 422
 423                         if (output == "square") {
 424                                  numDists = numSeqs * numSeqs;
 425                         }else {
 426                                 for(int i=0;i<numSeqs;i++){
 427                                         for(int j=0;j<i;j++){
 428                                                 numDists++;
 429                                                 if (numDists > processors) { break; }
 430                                         }
 431                                 }
 432                         }
 433
 434                         if (numDists < processors) { processors = numDists; }
 435
 436                         for (int i = 0; i < processors; i++) {
 437                                 distlinePair tempLine;
 438                                 lines.push_back(tempLine);
 439                                 if (output != "square") {
 440                                         lines[i].start = int (sqrt(float(i)/float(processors)) * numSeqs);
 441                                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 442                                 }else{
 443                                         lines[i].start = int ((float(i)/float(processors)) * numSeqs);
 444                                         lines[i].end = int ((float(i+1)/float(processors)) * numSeqs);
 445                                 }
 446
 447                         }
 448
 449                         createProcesses(outputFile);
 450                 }
 451         #else
 452                 //ifstream inFASTA;
 453                 if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 454                 else { driver(0, numSeqs, outputFile, "square");  }
 455         #endif
 456
 457 #endif
 458                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 459
 460                 #ifdef USE_MPI
 461                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 462
 463                         if (pid == 0) { //only one process should output to screen
 464                 #endif
 465
 466                 //if (output == "square") {  convertMatrix(outputFile); }
 467
 468                 ifstream fileHandle;
 469                 fileHandle.open(outputFile.c_str());
 470                 if(fileHandle) {
 471                         m->gobble(fileHandle);
 472                         if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff.");  m->mothurOutEndLine(); }
 473                 }
 474
 475                 //append the old column file to the new one
 476                 if ((oldfastafile != "") && (column != ""))  {
 477                         //we had to rename the column file so we didnt overwrite above, but we want to keep old name
 478                         if (outputFile == column) {
 479                                 string tempcolumn = column + ".old";
 480                                 m->appendFiles(tempcolumn, outputFile);
 481                                 remove(tempcolumn.c_str());
 482                         }else{
 483                                 m->appendFiles(outputFile, column);
 484                                 remove(outputFile.c_str());
 485                                 outputFile = column;
 486                         }
 487
 488                         if (outputDir != "") {
 489                                 string newOutputName = outputDir + m->getSimpleName(outputFile);
 490                                 rename(outputFile.c_str(), newOutputName.c_str());
 491                                 remove(outputFile.c_str());
 492                                 outputFile = newOutputName;
 493                         }
 494                 }
 495
 496
 497                 #ifdef USE_MPI
 498                         }
 499                 #endif
 500
 501                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 502
 503                 delete distCalculator;
 504
 505                 m->mothurOutEndLine();
 506                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 507                 m->mothurOut(outputFile); m->mothurOutEndLine();
 508                 m->mothurOutEndLine();
 509                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 510
 511
 512                 if (m->isTrue(compress)) {
 513                         m->mothurOut("Compressing..."); m->mothurOutEndLine();
 514                         m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 515                         system(("gzip -v " + outputFile).c_str());
 516                         outputNames.push_back(outputFile + ".gz");
 517                 }else { outputNames.push_back(outputFile); }
 518
 519                 return 0;
 520
 521         }
 522         catch(exception& e) {
 523                 m->errorOut(e, "DistanceCommand", "execute");
 524                 exit(1);
 525         }
 526 }
 527 /**************************************************************************************************/
 528 void DistanceCommand::createProcesses(string filename) {
 529         try {
 530 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 531                 int process = 1;
 532                 processIDS.clear();
 533
 534                 //loop through and create all the processes you want
 535                 while (process != processors) {
 536                         int pid = fork();
 537
 538                         if (pid > 0) {
 539                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 540                                 process++;
 541                         }else if (pid == 0){
 542                                 if (output != "square") {  driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", cutoff); }
 543                                 else { driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", "square"); }
 544                                 exit(0);
 545                         }else {
 546                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
 547                                 perror(" : ");
 548                                 for (int i=0;i<processIDS.size();i++) {  int temp = processIDS[i]; kill (temp, SIGINT); }
 549                                 exit(0);
 550                         }
 551                 }
 552
 553                 //parent does its part
 554                 if (output != "square") {  driver(lines[0].start, lines[0].end, filename, cutoff); }
 555                 else { driver(lines[0].start, lines[0].end, filename, "square"); }
 556
 557
 558                 //force parent to wait until all the processes are done
 559                 for (int i=0;i<processIDS.size();i++) {
 560                         int temp = processIDS[i];
 561                         wait(&temp);
 562                 }
 563
 564                 //append and remove temp files
 565                 for (int i=0;i<processIDS.size();i++) {
 566                         m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
 567                         remove((filename + toString(processIDS[i]) + ".temp").c_str());
 568                 }
 569 #endif
 570         }
 571         catch(exception& e) {
 572                 m->errorOut(e, "DistanceCommand", "createProcesses");
 573                 exit(1);
 574         }
 575 }
 576
 577 /**************************************************************************************************/
 578 /////// need to fix to work with calcs and sequencedb
 579 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 580         try {
 581
 582                 int startTime = time(NULL);
 583
 584                 //column file
 585                 ofstream outFile(dFileName.c_str(), ios::trunc);
 586                 outFile.setf(ios::fixed, ios::showpoint);
 587                 outFile << setprecision(4);
 588
 589                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 590
 591                 for(int i=startLine;i<endLine;i++){
 592                         if(output == "lt")      {
 593                                 string name = alignDB.get(i).getName();
 594                                 if (name.length() < 10) { //pad with spaces to make compatible
 595                                         while (name.length() < 10) {  name += " ";  }
 596                                 }
 597                                 outFile << name << '\t';
 598                         }
 599                         for(int j=0;j<i;j++){
 600
 601                                 if (m->control_pressed) { outFile.close(); return 0;  }
 602
 603                                 //if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
 604                                 //the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
 605                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 606
 607                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 608                                 double dist = distCalculator->getDist();
 609
 610                                 if(dist <= cutoff){
 611                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 612                                 }
 613                                 if (output == "lt") {  outFile << dist << '\t'; }
 614                         }
 615
 616                         if (output == "lt") { outFile << endl; }
 617
 618                         if(i % 100 == 0){
 619                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 620                         }
 621
 622                 }
 623                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 624
 625                 outFile.close();
 626
 627                 return 1;
 628         }
 629         catch(exception& e) {
 630                 m->errorOut(e, "DistanceCommand", "driver");
 631                 exit(1);
 632         }
 633 }
 634 /**************************************************************************************************/
 635 /////// need to fix to work with calcs and sequencedb
 636 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 637         try {
 638
 639                 int startTime = time(NULL);
 640
 641                 //column file
 642                 ofstream outFile(dFileName.c_str(), ios::trunc);
 643                 outFile.setf(ios::fixed, ios::showpoint);
 644                 outFile << setprecision(4);
 645
 646                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 647
 648                 for(int i=startLine;i<endLine;i++){
 649
 650                         string name = alignDB.get(i).getName();
 651                         //pad with spaces to make compatible
 652                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 653
 654                         outFile << name << '\t';
 655
 656                         for(int j=0;j<alignDB.getNumSeqs();j++){
 657
 658                                 if (m->control_pressed) { outFile.close(); return 0;  }
 659
 660                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 661                                 double dist = distCalculator->getDist();
 662
 663                                 outFile << dist << '\t';
 664                         }
 665
 666                         outFile << endl;
 667
 668                         if(i % 100 == 0){
 669                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 670                         }
 671
 672                 }
 673                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 674
 675                 outFile.close();
 676
 677                 return 1;
 678         }
 679         catch(exception& e) {
 680                 m->errorOut(e, "DistanceCommand", "driver");
 681                 exit(1);
 682         }
 683 }
 684 #ifdef USE_MPI
 685 /**************************************************************************************************/
 686 /////// need to fix to work with calcs and sequencedb
 687 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 688         try {
 689                 MPI_Status status;
 690                 int startTime = time(NULL);
 691
 692                 string outputString = "";
 693
 694                 for(int i=startLine;i<endLine;i++){
 695
 696                         for(int j=0;j<i;j++){
 697
 698                                 if (m->control_pressed) {  return 0;  }
 699
 700                                 //if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
 701                                 //the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
 702                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 703
 704                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 705                                 double dist = distCalculator->getDist();
 706
 707                                 if(dist <= cutoff){
 708                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 709                                 }
 710                         }
 711
 712                         if(i % 100 == 0){
 713                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 714                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 715                         }
 716
 717
 718                         //send results to parent
 719                         int length = outputString.length();
 720
 721                         char* buf = new char[length];
 722                         memcpy(buf, outputString.c_str(), length);
 723
 724                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 725                         outputString = "";
 726                         delete buf;
 727
 728                 }
 729
 730                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 731                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 732                 return 1;
 733         }
 734         catch(exception& e) {
 735                 m->errorOut(e, "DistanceCommand", "driverMPI");
 736                 exit(1);
 737         }
 738 }
 739 /**************************************************************************************************/
 740 /////// need to fix to work with calcs and sequencedb
 741 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 742         try {
 743                 MPI_Status status;
 744
 745                 MPI_File outMPI;
 746                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 747
 748                 //char* filename = new char[file.length()];
 749                 //memcpy(filename, file.c_str(), file.length());
 750
 751                 char filename[1024];
 752                 strcpy(filename, file.c_str());
 753
 754                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 755                 //delete filename;
 756
 757                 int startTime = time(NULL);
 758
 759                 string outputString = "";
 760                 size = 0;
 761
 762                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 763
 764                 for(int i=startLine;i<endLine;i++){
 765
 766                         string name = alignDB.get(i).getName();
 767                         if (name.length() < 10) { //pad with spaces to make compatible
 768                                 while (name.length() < 10) {  name += " ";  }
 769                         }
 770                         outputString += name + "\t";
 771
 772                         for(int j=0;j<i;j++){
 773
 774                                 if (m->control_pressed) {  return 0;  }
 775
 776                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 777                                 double dist = distCalculator->getDist();
 778
 779                                 outputString += toString(dist) + "\t";
 780                         }
 781
 782                         outputString += "\n";
 783
 784
 785                         if(i % 100 == 0){
 786                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 787                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 788                         }
 789
 790
 791                         //send results to parent
 792                         int length = outputString.length();
 793                         char* buf = new char[length];
 794                         memcpy(buf, outputString.c_str(), length);
 795
 796                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 797                         size += outputString.length();
 798                         outputString = "";
 799                         delete buf;
 800                 }
 801
 802                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 803                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 804                 MPI_File_close(&outMPI);
 805
 806                 return 1;
 807         }
 808         catch(exception& e) {
 809                 m->errorOut(e, "DistanceCommand", "driverMPI");
 810                 exit(1);
 811         }
 812 }
 813 /**************************************************************************************************/
 814 /////// need to fix to work with calcs and sequencedb
 815 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 816         try {
 817                 MPI_Status status;
 818
 819                 MPI_File outMPI;
 820                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 821
 822                 //char* filename = new char[file.length()];
 823                 //memcpy(filename, file.c_str(), file.length());
 824
 825                 char filename[1024];
 826                 strcpy(filename, file.c_str());
 827
 828                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 829                 //delete filename;
 830
 831                 int startTime = time(NULL);
 832
 833                 string outputString = "";
 834                 size = 0;
 835
 836                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 837
 838                 for(int i=startLine;i<endLine;i++){
 839
 840                         string name = alignDB.get(i).getName();
 841                         if (name.length() < 10) { //pad with spaces to make compatible
 842                                 while (name.length() < 10) {  name += " ";  }
 843                         }
 844                         outputString += name + "\t";
 845
 846                         for(int j=0;j<alignDB.getNumSeqs();j++){
 847
 848                                 if (m->control_pressed) {  return 0;  }
 849
 850                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 851                                 double dist = distCalculator->getDist();
 852
 853                                 outputString += toString(dist) + "\t";
 854                         }
 855
 856                         outputString += "\n";
 857
 858
 859                         if(i % 100 == 0){
 860                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 861                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 862                         }
 863
 864
 865                         //send results to parent
 866                         int length = outputString.length();
 867                         char* buf = new char[length];
 868                         memcpy(buf, outputString.c_str(), length);
 869
 870                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 871                         size += outputString.length();
 872                         outputString = "";
 873                         delete buf;
 874                 }
 875
 876                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 877                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 878                 MPI_File_close(&outMPI);
 879
 880                 return 1;
 881         }
 882         catch(exception& e) {
 883                 m->errorOut(e, "DistanceCommand", "driverMPI");
 884                 exit(1);
 885         }
 886 }
 887 #endif
 888 /**************************************************************************************************
 889 int DistanceCommand::convertMatrix(string outputFile) {
 890         try{
 891
 892                 //sort file by first column so the distances for each row are together
 893                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 894
 895                 //use the unix sort
 896                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 897                         string command = "sort -n " + outputFile + " -o " + outfile;
 898                         system(command.c_str());
 899                 #else //sort using windows sort
 900                         string command = "sort " + outputFile + " /O " + outfile;
 901                         system(command.c_str());
 902                 #endif
 903
 904
 905                 //output to new file distance for each row and save positions in file where new row begins
 906                 ifstream in;
 907                 m->openInputFile(outfile, in);
 908
 909                 ofstream out;
 910                 m->openOutputFile(outputFile, out);
 911
 912                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 913
 914                 out << alignDB.getNumSeqs() << endl;
 915
 916                 //get first currentRow
 917                 string first, currentRow, second;
 918                 float dist;
 919                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 920                 map<string, float>::iterator it;
 921
 922                 in >> first;
 923                 currentRow = first;
 924
 925                 rowDists[first] = 0.00; //distance to yourself is 0.0
 926
 927                 in.seekg(0);
 928                 //m->openInputFile(outfile, in);
 929
 930                 while(!in.eof()) {
 931                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 932
 933                         in >> first >> second >> dist; m->gobble(in);
 934
 935                         if (first != currentRow) {
 936                                 //print out last row
 937                                 out << currentRow << '\t'; //print name
 938
 939                                 //print dists
 940                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 941                                         out << it->second << '\t';
 942                                 }
 943                                 out << endl;
 944
 945                                 //start new row
 946                                 currentRow = first;
 947                                 rowDists.clear();
 948                                 rowDists[first] = 0.00;
 949                                 rowDists[second] = dist;
 950                         }else{
 951                                 rowDists[second] = dist;
 952                         }
 953                 }
 954                 //print out last row
 955                 out << currentRow << '\t'; //print name
 956
 957                 //print dists
 958                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 959                         out << it->second << '\t';
 960                 }
 961                 out << endl;
 962
 963                 in.close();
 964                 out.close();
 965
 966                 remove(outfile.c_str());
 967
 968                 return 1;
 969
 970         }
 971         catch(exception& e) {
 972                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 973                 exit(1);
 974         }
 975 }
 976 /**************************************************************************************************
 977 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 978         try{
 979
 980                 //sort file by first column so the distances for each row are together
 981                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 982
 983                 //use the unix sort
 984                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 985                         string command = "sort -n " + outputFile + " -o " + outfile;
 986                         system(command.c_str());
 987                 #else //sort using windows sort
 988                         string command = "sort " + outputFile + " /O " + outfile;
 989                         system(command.c_str());
 990                 #endif
 991
 992
 993                 //output to new file distance for each row and save positions in file where new row begins
 994                 ifstream in;
 995                 m->openInputFile(outfile, in);
 996
 997                 ofstream out;
 998                 m->openOutputFile(outputFile, out);
 999
1000                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1001
1002                 out << alignDB.getNumSeqs() << endl;
1003
1004                 //get first currentRow
1005                 string first, currentRow, second;
1006                 float dist;
1007                 int i, j;
1008                 i = 0; j = 0;
1009                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
1010                 map<string, float>::iterator it;
1011
1012                 in >> first;
1013                 currentRow = first;
1014
1015                 rowDists[first] = 0.00; //distance to yourself is 0.0
1016
1017                 in.seekg(0);
1018                 //m->openInputFile(outfile, in);
1019
1020                 while(!in.eof()) {
1021                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
1022
1023                         in >> first >> second >> dist; m->gobble(in);
1024
1025                         if (first != currentRow) {
1026                                 //print out last row
1027                                 out << currentRow << '\t'; //print name
1028
1029                                 //print dists
1030                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1031                                         if (j >= i) { break; }
1032                                         out << it->second << '\t';
1033                                         j++;
1034                                 }
1035                                 out << endl;
1036
1037                                 //start new row
1038                                 currentRow = first;
1039                                 rowDists.clear();
1040                                 rowDists[first] = 0.00;
1041                                 rowDists[second] = dist;
1042                                 j = 0;
1043                                 i++;
1044                         }else{
1045                                 rowDists[second] = dist;
1046                         }
1047                 }
1048                 //print out last row
1049                 out << currentRow << '\t'; //print name
1050
1051                 //print dists
1052                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1053                         out << it->second << '\t';
1054                 }
1055                 out << endl;
1056
1057                 in.close();
1058                 out.close();
1059
1060                 remove(outfile.c_str());
1061
1062                 return 1;
1063
1064         }
1065         catch(exception& e) {
1066                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
1067                 exit(1);
1068         }
1069 }
1070 /**************************************************************************************************/
1071 //its okay if the column file does not contain all the names in the fasta file, since some distance may have been above a cutoff,
1072 //but no sequences can be in the column file that are not in oldfasta. also, if a distance is above the cutoff given then remove it.
1073 //also check to make sure the 2 files have the same alignment length.
1074 bool DistanceCommand::sanityCheck() {
1075         try{
1076                 bool good = true;
1077
1078                 //make sure the 2 fasta files have the same alignment length
1079                 ifstream in;
1080                 m->openInputFile(fastafile, in);
1081                 int fastaAlignLength = 0;
1082                 if (in) {
1083                         Sequence tempIn(in);
1084                         fastaAlignLength = tempIn.getAligned().length();
1085                 }
1086                 in.close();
1087
1088                 ifstream in2;
1089                 m->openInputFile(oldfastafile, in2);
1090                 int oldfastaAlignLength = 0;
1091                 if (in2) {
1092                         Sequence tempIn2(in2);
1093                         oldfastaAlignLength = tempIn2.getAligned().length();
1094                 }
1095                 in2.close();
1096
1097                 if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false;  }
1098
1099                 //read fasta file and save names as well as adding them to the alignDB
1100                 set<string> namesOldFasta;
1101
1102                 ifstream inFasta;
1103                 m->openInputFile(oldfastafile, inFasta);
1104
1105                 while (!inFasta.eof()) {
1106                         if (m->control_pressed) {  inFasta.close(); return good;  }
1107
1108                         Sequence temp(inFasta);
1109
1110                         if (temp.getName() != "") {
1111                                 namesOldFasta.insert(temp.getName());  //save name
1112                                 alignDB.push_back(temp);  //add to DB
1113                         }
1114
1115                         m->gobble(inFasta);
1116                 }
1117
1118                 inFasta.close();
1119
1120                 //read through the column file checking names and removing distances above the cutoff
1121                 ifstream inDist;
1122                 m->openInputFile(column, inDist);
1123
1124                 ofstream outDist;
1125                 string outputFile = column + ".temp";
1126                 m->openOutputFile(outputFile, outDist);
1127
1128                 string name1, name2;
1129                 float dist;
1130                 while (!inDist.eof()) {
1131                         if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
1132
1133                         inDist >> name1 >> name2 >> dist; m->gobble(inDist);
1134
1135                         //both names are in fasta file and distance is below cutoff
1136                         if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) {  good = false; break;  }
1137                         else{
1138                                 if (dist <= cutoff) {
1139                                         outDist << name1 << '\t' << name2 << '\t' << dist << endl;
1140                                 }
1141                         }
1142                 }
1143
1144                 inDist.close();
1145                 outDist.close();
1146
1147                 if (good) {
1148                         remove(column.c_str());
1149                         rename(outputFile.c_str(), column.c_str());
1150                 }else{
1151                         remove(outputFile.c_str()); //temp file is bad because file mismatch above
1152                 }
1153
1154         }
1155         catch(exception& e) {
1156                 m->errorOut(e, "DistanceCommand", "sanityCheck");
1157                 exit(1);
1158         }
1159 }
1160 /**************************************************************************************************/
1161
1162
1163
1164