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