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 DistanceCommand::DistanceCommand(string option) {
  20         try {
  21                 abort = false;
  22                 Estimators.clear();
  23
  24                 //allow user to run help
  25                 if(option == "help") { help(); abort = true; }
  26
  27                 else {
  28                         //valid paramters for this command
  29                         string Array[] =  {"fasta", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir"};
  30                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  31
  32                         OptionParser parser(option);
  33                         map<string, string> parameters = parser.getParameters();
  34
  35                         ValidParameters validParameter;
  36                         map<string, string>::iterator it2;
  37
  38                         //check to make sure all parameters are valid for command
  39                         for (it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  40                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  41                         }
  42
  43                         //if the user changes the input directory command factory will send this info to us in the output parameter
  44                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  45                         if (inputDir == "not found"){   inputDir = "";          }
  46                         else {
  47                                 string path;
  48                                 it2 = parameters.find("fasta");
  49                                 //user has given a template file
  50                                 if(it2 != parameters.end()){
  51                                         path = hasPath(it2->second);
  52                                         //if the user has not given a path then, add inputdir. else leave path alone.
  53                                         if (path == "") {       parameters["fasta"] = inputDir + it2->second;           }
  54                                 }
  55                         }
  56
  57                         //check for required parameters
  58                         fastafile = validParameter.validFile(parameters, "fasta", true);
  59                         if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the dist.seqs command."); m->mothurOutEndLine(); abort = true; }
  60                         else if (fastafile == "not open") { abort = true; }
  61                         else{
  62                                 ifstream inFASTA;
  63                                 openInputFile(fastafile, inFASTA);
  64                                 alignDB = SequenceDB(inFASTA);
  65                                 inFASTA.close();
  66                         }
  67
  68                         //if the user changes the output directory command factory will send this info to us in the output parameter
  69                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  70                                 outputDir = "";
  71                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
  72                         }
  73
  74                         //check for optional parameter and set defaults
  75                         // ...at some point should added some additional type checking...
  76                         calc = validParameter.validFile(parameters, "calc", false);
  77                         if (calc == "not found") { calc = "onegap";  }
  78                         else {
  79                                  if (calc == "default")  {  calc = "onegap";  }
  80                         }
  81                         splitAtDash(calc, Estimators);
  82
  83                         string temp;
  84                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
  85                         convert(temp, countends);
  86
  87                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
  88                         convert(temp, cutoff);
  89
  90                         temp = validParameter.validFile(parameters, "processors", false);       if(temp == "not found"){        temp = "1"; }
  91                         convert(temp, processors);
  92
  93                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "column"; }
  94
  95                         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"; }
  96
  97                         ValidCalculators validCalculator;
  98
  99                         if (isTrue(countends) == true) {
 100                                 for (int i=0; i<Estimators.size(); i++) {
 101                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 102                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
 103                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
 104                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
 105                                         }
 106                                 }
 107                         }else {
 108                                 for (int i=0; i<Estimators.size(); i++) {
 109                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 110                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
 111                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
 112                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
 113                                         }
 114                                 }
 115                         }
 116
 117                 }
 118
 119         }
 120         catch(exception& e) {
 121                 m->errorOut(e, "DistanceCommand", "DistanceCommand");
 122                 exit(1);
 123         }
 124 }
 125
 126 //**********************************************************************************************************************
 127
 128 DistanceCommand::~DistanceCommand(){
 129
 130         for(int i=0;i<lines.size();i++){
 131                 delete lines[i];
 132         }
 133
 134 }
 135
 136 //**********************************************************************************************************************
 137
 138 void DistanceCommand::help(){
 139         try {
 140                 m->mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
 141                 m->mothurOut("The dist.seqs command parameters are fasta, calc, countends, output, cutoff and processors.  \n");
 142                 m->mothurOut("The fasta parameter is required.\n");
 143                 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");
 144                 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");
 145                 m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 146                 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");
 147                 m->mothurOut("The processors parameter allows you to specify number of processors to use.  The default is 1.\n");
 148                 m->mothurOut("The dist.seqs command should be in the following format: \n");
 149                 m->mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
 150                 m->mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
 151                 m->mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
 152         }
 153         catch(exception& e) {
 154                 m->errorOut(e, "DistanceCommand", "help");
 155                 exit(1);
 156         }
 157 }
 158 //**********************************************************************************************************************
 159
 160 int DistanceCommand::execute(){
 161         try {
 162
 163                 if (abort == true) { return 0; }
 164
 165                 int startTime = time(NULL);
 166
 167                 int numSeqs = alignDB.getNumSeqs();
 168                 cutoff += 0.005;
 169
 170                 string outputFile;
 171
 172                 if (output == "lt") { //does the user want lower triangle phylip formatted file
 173                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "phylip.dist";
 174                         remove(outputFile.c_str());
 175
 176                         //output numSeqs to phylip formatted dist file
 177                 }else if (output == "column") { //user wants column format
 178                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "dist";
 179                         remove(outputFile.c_str());
 180                 }else { //assume square
 181                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "square.dist";
 182                         remove(outputFile.c_str());
 183                 }
 184
 185
 186 #ifdef USE_MPI
 187
 188                 int pid, start, end;
 189                 int tag = 2001;
 190
 191                 MPI_Status status;
 192                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 193                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 194
 195                 //each process gets where it should start and stop in the file
 196                 start = int (sqrt(float(pid)/float(processors)) * numSeqs);
 197                 end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
 198
 199                 if (output != "lt") {
 200                         MPI_File outMPI;
 201                         int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 202
 203                         char filename[outputFile.length()];
 204                         strcpy(filename, outputFile.c_str());
 205
 206                         MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
 207
 208                         if (pid == 0) { //you are the root process
 209
 210                                 //do your part
 211                                 string outputMyPart;
 212                                 driverMPI(start, end, outMPI, cutoff);
 213
 214                                 if (m->control_pressed) { MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 215
 216                                 //wait on chidren
 217                                 for(int i = 1; i < processors; i++) {
 218                                         if (m->control_pressed) { MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 219
 220                                         char buf[4];
 221                                         MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 222                                 }
 223                         }else { //you are a child process
 224                                 //do your part
 225                                 driverMPI(start, end, outMPI, cutoff);
 226
 227                                 if (m->control_pressed) { MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 228
 229                                 char buf[4];
 230                                 strcpy(buf, "done");
 231                                 //tell parent you are done.
 232                                 MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 233                         }
 234
 235                         MPI_File_close(&outMPI);
 236
 237                 }else { //lower triangle format
 238                         if (pid == 0) { //you are the root process
 239
 240                                 //do your part
 241                                 string outputMyPart;
 242                                 long mySize;
 243                                 driverMPI(start, end, outputFile, mySize);
 244
 245                                 if (m->control_pressed) {  delete distCalculator;  return 0; }
 246
 247                                 int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 248                                 MPI_File outMPI;
 249                                 MPI_File inMPI;
 250
 251                                 char filename[outputFile.length()];
 252                                 strcpy(filename, outputFile.c_str());
 253
 254                                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 255
 256                                 //wait on chidren
 257                                 for(int b = 1; b < processors; b++) {
 258                                         long fileSize;
 259
 260                                         if (m->control_pressed) { MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 261
 262                                         MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
 263
 264                                         string outTemp = outputFile + toString(b) + ".temp";
 265                                         char buf[outTemp.length()];
 266                                         strcpy(buf, outTemp.c_str());
 267
 268                                         MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 269
 270                                         int count = 0;
 271                                         while (count < fileSize) { //read 1000 characters at a time
 272                                                 //send freqs
 273                                                 char buf2[1];
 274                                                 MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
 275                                                 MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
 276                                                 count += 1;
 277                                         }
 278
 279                                         MPI_File_close(&inMPI); //deleted on close
 280                                 }
 281
 282                                 MPI_File_close(&outMPI);
 283                         }else { //you are a child process
 284                                 //do your part
 285                                 long size;
 286                                 driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size);
 287
 288                                 if (m->control_pressed) { delete distCalculator;  return 0; }
 289
 290                                 //tell parent you are done.
 291                                 MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
 292                         }
 293                 }
 294 #else
 295
 296         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 297                 //if you don't need to fork anything
 298                 if(processors == 1){
 299                         driver(0, numSeqs, outputFile, cutoff);
 300                 }else{ //you have multiple processors
 301
 302                         for (int i = 0; i < processors; i++) {
 303                                 lines.push_back(new linePair());
 304                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 305                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 306                         }
 307
 308                         createProcesses(outputFile);
 309
 310                         map<int, int>::iterator it = processIDS.begin();
 311                         rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 312                         it++;
 313
 314                         //append and remove temp files
 315                         for (; it != processIDS.end(); it++) {
 316                                 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 317                                 remove((outputFile + toString(it->second) + ".temp").c_str());
 318                         }
 319                 }
 320         #else
 321                 ifstream inFASTA;
 322                 driver(0, numSeqs, outputFile, cutoff);
 323         #endif
 324
 325 #endif
 326                 if (m->control_pressed) { delete distCalculator; remove(outputFile.c_str()); return 0; }
 327
 328                 #ifdef USE_MPI
 329                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 330
 331                         if (pid == 0) { //only one process should output to screen
 332                 #endif
 333
 334                 if (output == "square") {  convertMatrix(outputFile); }
 335
 336                 #ifdef USE_MPI
 337                         }
 338                 #endif
 339
 340                 if (m->control_pressed) { delete distCalculator; remove(outputFile.c_str()); return 0; }
 341
 342                 delete distCalculator;
 343
 344                 m->mothurOutEndLine();
 345                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 346                 m->mothurOut(outputFile); m->mothurOutEndLine();
 347                 m->mothurOutEndLine();
 348                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 349                 return 0;
 350
 351         }
 352         catch(exception& e) {
 353                 m->errorOut(e, "DistanceCommand", "execute");
 354                 exit(1);
 355         }
 356 }
 357 /**************************************************************************************************/
 358 void DistanceCommand::createProcesses(string filename) {
 359         try {
 360 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 361                 int process = 0;
 362                 processIDS.clear();
 363
 364                 //loop through and create all the processes you want
 365                 while (process != processors) {
 366                         int pid = fork();
 367
 368                         if (pid > 0) {
 369                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 370                                 process++;
 371                         }else if (pid == 0){
 372                                 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
 373                                 exit(0);
 374                         }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
 375                 }
 376
 377                 //force parent to wait until all the processes are done
 378                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 379                         int temp = it->second;
 380                         wait(&temp);
 381                 }
 382 #endif
 383         }
 384         catch(exception& e) {
 385                 m->errorOut(e, "DistanceCommand", "createProcesses");
 386                 exit(1);
 387         }
 388 }
 389
 390 /**************************************************************************************************/
 391 /////// need to fix to work with calcs and sequencedb
 392 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 393         try {
 394
 395                 int startTime = time(NULL);
 396
 397                 //column file
 398                 ofstream outFile(dFileName.c_str(), ios::trunc);
 399                 outFile.setf(ios::fixed, ios::showpoint);
 400                 outFile << setprecision(4);
 401
 402                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 403
 404                 for(int i=startLine;i<endLine;i++){
 405                         if(output == "lt")      {
 406                                 string name = alignDB.get(i).getName();
 407                                 if (name.length() < 10) { //pad with spaces to make compatible
 408                                         while (name.length() < 10) {  name += " ";  }
 409                                 }
 410                                 outFile << name << '\t';
 411                         }
 412                         for(int j=0;j<i;j++){
 413
 414                                 if (m->control_pressed) { outFile.close(); return 0;  }
 415
 416                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 417                                 double dist = distCalculator->getDist();
 418
 419                                 if(dist <= cutoff){
 420                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 421                                 }
 422                                 if (output == "lt") {  outFile << dist << '\t'; }
 423
 424                                 if (output == "square") { //make a square column you can convert to square phylip
 425                                         outFile << alignDB.get(i).getName() << '\t' << alignDB.get(j).getName() << '\t' << dist << endl;
 426                                         outFile << alignDB.get(j).getName() << '\t' << alignDB.get(i).getName() << '\t' << dist << endl;
 427                                 }
 428
 429                         }
 430
 431                         if (output == "lt") { outFile << endl; }
 432
 433                         if(i % 100 == 0){
 434                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 435                         }
 436
 437                 }
 438                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 439
 440                 outFile.close();
 441
 442                 return 1;
 443         }
 444         catch(exception& e) {
 445                 m->errorOut(e, "DistanceCommand", "driver");
 446                 exit(1);
 447         }
 448 }
 449 #ifdef USE_MPI
 450 /**************************************************************************************************/
 451 /////// need to fix to work with calcs and sequencedb
 452 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 453         try {
 454                 MPI_Status status;
 455                 int startTime = time(NULL);
 456
 457                 string outputString = "";
 458
 459                 for(int i=startLine;i<endLine;i++){
 460
 461                         for(int j=0;j<i;j++){
 462
 463                                 if (m->control_pressed) {  return 0;  }
 464
 465                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 466                                 double dist = distCalculator->getDist();
 467
 468                                 if(dist <= cutoff){
 469                                         if (output == "column") { outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n'); }
 470                                 }
 471
 472                                 if (output == "square") { //make a square column you can convert to square phylip
 473                                         outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 474                                         outputString += (alignDB.get(j).getName() + ' ' + alignDB.get(i).getName() + ' ' + toString(dist) + '\n');
 475                                 }
 476                         }
 477
 478                         if(i % 100 == 0){
 479                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 480                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 481                         }
 482
 483
 484                         //send results to parent
 485                         int length = outputString.length();
 486                         char buf[length];
 487                         strcpy(buf, outputString.c_str());
 488
 489                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 490                         outputString = "";
 491
 492                 }
 493
 494                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 495                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 496                 return 1;
 497         }
 498         catch(exception& e) {
 499                 m->errorOut(e, "DistanceCommand", "driverMPI");
 500                 exit(1);
 501         }
 502 }
 503 /**************************************************************************************************/
 504 /////// need to fix to work with calcs and sequencedb
 505 int DistanceCommand::driverMPI(int startLine, int endLine, string file, long& size){
 506         try {
 507                 MPI_Status status;
 508
 509                 MPI_File outMPI;
 510                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 511
 512                 char filename[file.length()];
 513                 strcpy(filename, file.c_str());
 514
 515                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 516
 517                 int startTime = time(NULL);
 518
 519                 string outputString = "";
 520                 size = 0;
 521
 522                 if((output == "lt") && startLine == 0){ outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 523
 524                 for(int i=startLine;i<endLine;i++){
 525                         if(output == "lt")      {
 526                                 string name = alignDB.get(i).getName();
 527                                 if (name.length() < 10) { //pad with spaces to make compatible
 528                                         while (name.length() < 10) {  name += " ";  }
 529                                 }
 530                                 outputString += name + "\t";
 531                         }
 532                         for(int j=0;j<i;j++){
 533
 534                                 if (m->control_pressed) {  return 0;  }
 535
 536                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 537                                 double dist = distCalculator->getDist();
 538
 539                                 if (output == "lt") {  outputString += toString(dist) + "\t"; }
 540                         }
 541
 542                         if (output == "lt") { outputString += "\n"; }
 543
 544
 545                         if(i % 100 == 0){
 546                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 547                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 548                         }
 549
 550
 551                         //send results to parent
 552                         int length = outputString.length();
 553                         char buf[length];
 554                         strcpy(buf, outputString.c_str());
 555
 556                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 557                         size += outputString.length();
 558                         outputString = "";
 559
 560
 561                 }
 562
 563                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 564                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 565                 MPI_File_close(&outMPI);
 566
 567                 return 1;
 568         }
 569         catch(exception& e) {
 570                 m->errorOut(e, "DistanceCommand", "driverMPI");
 571                 exit(1);
 572         }
 573 }
 574 #endif
 575 /**************************************************************************************************/
 576 int DistanceCommand::convertMatrix(string outputFile) {
 577         try{
 578
 579                 //sort file by first column so the distances for each row are together
 580                 string outfile = getRootName(outputFile) + "sorted.dist.temp";
 581
 582                 //use the unix sort
 583                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 584                         string command = "sort -n " + outputFile + " -o " + outfile;
 585                         system(command.c_str());
 586                 #else //sort using windows sort
 587                         string command = "sort " + outputFile + " /O " + outfile;
 588                         system(command.c_str());
 589                 #endif
 590
 591
 592                 //output to new file distance for each row and save positions in file where new row begins
 593                 ifstream in;
 594                 openInputFile(outfile, in);
 595
 596                 ofstream out;
 597                 openOutputFile(outputFile, out);
 598
 599                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 600
 601                 out << alignDB.getNumSeqs() << endl;
 602
 603                 //get first currentRow
 604                 string first, currentRow, second;
 605                 float dist;
 606                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 607                 map<string, float>::iterator it;
 608
 609                 in >> first;
 610                 currentRow = first;
 611
 612                 rowDists[first] = 0.00; //distance to yourself is 0.0
 613
 614                 in.seekg(0);
 615                 //openInputFile(outfile, in);
 616
 617                 while(!in.eof()) {
 618                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 619
 620                         in >> first >> second >> dist; gobble(in);
 621
 622                         if (first != currentRow) {
 623                                 //print out last row
 624                                 out << currentRow << '\t'; //print name
 625
 626                                 //print dists
 627                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 628                                         out << it->second << '\t';
 629                                 }
 630                                 out << endl;
 631
 632                                 //start new row
 633                                 currentRow = first;
 634                                 rowDists.clear();
 635                                 rowDists[first] = 0.00;
 636                                 rowDists[second] = dist;
 637                         }else{
 638                                 rowDists[second] = dist;
 639                         }
 640                 }
 641                 //print out last row
 642                 out << currentRow << '\t'; //print name
 643
 644                 //print dists
 645                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 646                         out << it->second << '\t';
 647                 }
 648                 out << endl;
 649
 650                 in.close();
 651                 out.close();
 652
 653                 remove(outfile.c_str());
 654
 655                 return 1;
 656
 657         }
 658         catch(exception& e) {
 659                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 660                 exit(1);
 661         }
 662 }
 663 /**************************************************************************************************/
 664 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 665         try{
 666
 667                 //sort file by first column so the distances for each row are together
 668                 string outfile = getRootName(outputFile) + "sorted.dist.temp";
 669
 670                 //use the unix sort
 671                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 672                         string command = "sort -n " + outputFile + " -o " + outfile;
 673                         system(command.c_str());
 674                 #else //sort using windows sort
 675                         string command = "sort " + outputFile + " /O " + outfile;
 676                         system(command.c_str());
 677                 #endif
 678
 679
 680                 //output to new file distance for each row and save positions in file where new row begins
 681                 ifstream in;
 682                 openInputFile(outfile, in);
 683
 684                 ofstream out;
 685                 openOutputFile(outputFile, out);
 686
 687                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 688
 689                 out << alignDB.getNumSeqs() << endl;
 690
 691                 //get first currentRow
 692                 string first, currentRow, second;
 693                 float dist;
 694                 int i, j;
 695                 i = 0; j = 0;
 696                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 697                 map<string, float>::iterator it;
 698
 699                 in >> first;
 700                 currentRow = first;
 701
 702                 rowDists[first] = 0.00; //distance to yourself is 0.0
 703
 704                 in.seekg(0);
 705                 //openInputFile(outfile, in);
 706
 707                 while(!in.eof()) {
 708                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 709
 710                         in >> first >> second >> dist; gobble(in);
 711
 712                         if (first != currentRow) {
 713                                 //print out last row
 714                                 out << currentRow << '\t'; //print name
 715
 716                                 //print dists
 717                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 718                                         if (j >= i) { break; }
 719                                         out << it->second << '\t';
 720                                         j++;
 721                                 }
 722                                 out << endl;
 723
 724                                 //start new row
 725                                 currentRow = first;
 726                                 rowDists.clear();
 727                                 rowDists[first] = 0.00;
 728                                 rowDists[second] = dist;
 729                                 j = 0;
 730                                 i++;
 731                         }else{
 732                                 rowDists[second] = dist;
 733                         }
 734                 }
 735                 //print out last row
 736                 out << currentRow << '\t'; //print name
 737
 738                 //print dists
 739                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 740                         out << it->second << '\t';
 741                 }
 742                 out << endl;
 743
 744                 in.close();
 745                 out.close();
 746
 747                 remove(outfile.c_str());
 748
 749                 return 1;
 750
 751         }
 752         catch(exception& e) {
 753                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
 754                 exit(1);
 755         }
 756 }
 757 /**************************************************************************************************
 758 void DistanceCommand::appendFiles(string temp, string filename) {
 759         try{
 760                 ofstream output;
 761                 ifstream input;
 762
 763                 //open output file in append mode
 764                 openOutputFileAppend(filename, output);
 765                 openInputFile(temp, input);
 766
 767                 while(char c = input.get()){
 768                         if(input.eof())         {       break;                  }
 769                         else                            {       output << c;    }
 770                 }
 771
 772                 input.close();
 773                 output.close();
 774         }
 775         catch(exception& e) {
 776                 m->errorOut(e, "DistanceCommand", "appendFiles");
 777                 exit(1);
 778         }
 779 }
 780 /**************************************************************************************************/
 781
 782