distancecommand.cpp

   1 /*
   2  *  distancecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/7/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "distancecommand.h"
  11 #include "ignoregaps.h"
  12 #include "eachgapdist.h"
  13 #include "eachgapignore.h"
  14 #include "onegapdist.h"
  15 #include "onegapignore.h"
  16
  17
  18 //**********************************************************************************************************************
  19 vector<string> DistanceCommand::getValidParameters(){
  20         try {
  21                 string Array[] =  {"fasta","oldfasta","column", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir","compress"};
  22                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  23                 return myArray;
  24         }
  25         catch(exception& e) {
  26                 m->errorOut(e, "DistanceCommand", "getValidParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 DistanceCommand::DistanceCommand(){
  32         try {
  33                 abort = true; calledHelp = true;
  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, "DistanceCommand", "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; calledHelp = false;
  70                 Estimators.clear();
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; calledHelp = 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 //**********************************************************************************************************************
 216
 217 void DistanceCommand::help(){
 218         try {
 219                 m->mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
 220                 m->mothurOut("The dist.seqs command parameters are fasta, oldfasta, column, calc, countends, output, compress, cutoff and processors.  \n");
 221                 m->mothurOut("The fasta parameter is required.\n");
 222                 m->mothurOut("The oldfasta and column parameters allow you to append the distances calculated to the column file.\n");
 223                 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");
 224                 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");
 225                 m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 226                 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");
 227                 m->mothurOut("The processors parameter allows you to specify number of processors to use.  The default is 1.\n");
 228                 m->mothurOut("The compress parameter allows you to indicate that you want the resulting distance file compressed.  The default is false.\n");
 229                 m->mothurOut("The dist.seqs command should be in the following format: \n");
 230                 m->mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
 231                 m->mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
 232                 m->mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
 233         }
 234         catch(exception& e) {
 235                 m->errorOut(e, "DistanceCommand", "help");
 236                 exit(1);
 237         }
 238 }
 239 //**********************************************************************************************************************
 240
 241 int DistanceCommand::execute(){
 242         try {
 243
 244                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 245
 246                 int startTime = time(NULL);
 247
 248                 //save number of new sequence
 249                 numNewFasta = alignDB.getNumSeqs();
 250
 251                 //sanity check the oldfasta and column file as well as add oldfasta sequences to alignDB
 252                 if ((oldfastafile != "") && (column != ""))  {  if (!(sanityCheck())) { return 0; }  }
 253
 254                 if (m->control_pressed) { return 0; }
 255
 256                 int numSeqs = alignDB.getNumSeqs();
 257                 cutoff += 0.005;
 258
 259                 string outputFile;
 260
 261                 if (output == "lt") { //does the user want lower triangle phylip formatted file
 262                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
 263                         remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
 264
 265                         //output numSeqs to phylip formatted dist file
 266                 }else if (output == "column") { //user wants column format
 267                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
 268                         outputTypes["column"].push_back(outputFile);
 269
 270                         //so we don't accidentally overwrite
 271                         if (outputFile == column) {
 272                                 string tempcolumn = column + ".old";
 273                                 rename(column.c_str(), tempcolumn.c_str());
 274                         }
 275
 276                         remove(outputFile.c_str());
 277                 }else { //assume square
 278                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
 279                         remove(outputFile.c_str());
 280                         outputTypes["phylip"].push_back(outputFile);
 281                 }
 282
 283
 284 #ifdef USE_MPI
 285
 286                 int pid, start, end;
 287                 int tag = 2001;
 288
 289                 MPI_Status status;
 290                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 291                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 292
 293                 //each process gets where it should start and stop in the file
 294                 if (output != "square") {
 295                         start = int (sqrt(float(pid)/float(processors)) * numSeqs);
 296                         end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
 297                 }else{
 298                         start = int ((float(pid)/float(processors)) * numSeqs);
 299                         end = int ((float(pid+1)/float(processors)) * numSeqs);
 300                 }
 301
 302                 if (output == "column") {
 303                         MPI_File outMPI;
 304                         int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 305
 306                         //char* filename = new char[outputFile.length()];
 307                         //memcpy(filename, outputFile.c_str(), outputFile.length());
 308
 309                         char filename[1024];
 310                         strcpy(filename, outputFile.c_str());
 311
 312                         MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
 313                         //delete filename;
 314
 315                         if (pid == 0) { //you are the root process
 316
 317                                 //do your part
 318                                 string outputMyPart;
 319
 320                                 driverMPI(start, end, outMPI, cutoff);
 321
 322                                 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator;  return 0; }
 323
 324                                 //wait on chidren
 325                                 for(int i = 1; i < processors; i++) {
 326                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 327
 328                                         char buf[5];
 329                                         MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 330                                 }
 331                         }else { //you are a child process
 332                                 //do your part
 333                                 driverMPI(start, end, outMPI, cutoff);
 334
 335                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 336
 337                                 char buf[5];
 338                                 strcpy(buf, "done");
 339                                 //tell parent you are done.
 340                                 MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 341                         }
 342
 343                         MPI_File_close(&outMPI);
 344
 345                 }else { //lower triangle format
 346                         if (pid == 0) { //you are the root process
 347
 348                                 //do your part
 349                                 string outputMyPart;
 350                                 unsigned long int mySize;
 351
 352                                 if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
 353                                 else { driverMPI(start, end, outputFile, mySize, output); }
 354
 355                                 if (m->control_pressed) {  outputTypes.clear();  delete distCalculator;  return 0; }
 356
 357                                 int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 358                                 MPI_File outMPI;
 359                                 MPI_File inMPI;
 360
 361                                 //char* filename = new char[outputFile.length()];
 362                                 //memcpy(filename, outputFile.c_str(), outputFile.length());
 363
 364                                 char filename[1024];
 365                                 strcpy(filename, outputFile.c_str());
 366
 367                                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 368                                 //delete filename;
 369
 370                                 //wait on chidren
 371                                 for(int b = 1; b < processors; b++) {
 372                                         unsigned long int fileSize;
 373
 374                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 375
 376                                         MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
 377
 378                                         string outTemp = outputFile + toString(b) + ".temp";
 379
 380                                         char* buf = new char[outTemp.length()];
 381                                         memcpy(buf, outTemp.c_str(), outTemp.length());
 382
 383                                         MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 384                                         delete buf;
 385
 386                                         int count = 0;
 387                                         while (count < fileSize) {
 388                                                 char buf2[1];
 389                                                 MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
 390                                                 MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
 391                                                 count += 1;
 392                                         }
 393
 394                                         MPI_File_close(&inMPI); //deleted on close
 395                                 }
 396
 397                                 MPI_File_close(&outMPI);
 398                         }else { //you are a child process
 399                                 //do your part
 400                                 unsigned long int size;
 401                                 if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
 402                                 else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
 403
 404                                 if (m->control_pressed) { delete distCalculator;  return 0; }
 405
 406                                 //tell parent you are done.
 407                                 MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
 408                         }
 409                 }
 410                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 411 #else
 412
 413         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 414                 //if you don't need to fork anything
 415                 if(processors == 1){
 416                         if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 417                         else { driver(0, numSeqs, outputFile, "square");  }
 418                 }else{ //you have multiple processors
 419
 420                         unsigned long int numDists = 0;
 421
 422                         if (output == "square") {
 423                                  numDists = numSeqs * numSeqs;
 424                         }else {
 425                                 for(int i=0;i<numSeqs;i++){
 426                                         for(int j=0;j<i;j++){
 427                                                 numDists++;
 428                                                 if (numDists > processors) { break; }
 429                                         }
 430                                 }
 431                         }
 432
 433                         if (numDists < processors) { processors = numDists; }
 434
 435                         for (int i = 0; i < processors; i++) {
 436                                 distlinePair tempLine;
 437                                 lines.push_back(tempLine);
 438                                 if (output != "square") {
 439                                         lines[i].start = int (sqrt(float(i)/float(processors)) * numSeqs);
 440                                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 441                                 }else{
 442                                         lines[i].start = int ((float(i)/float(processors)) * numSeqs);
 443                                         lines[i].end = int ((float(i+1)/float(processors)) * numSeqs);
 444                                 }
 445
 446                         }
 447
 448                         createProcesses(outputFile);
 449                 }
 450         #else
 451                 //ifstream inFASTA;
 452                 if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 453                 else { driver(0, numSeqs, outputFile, "square");  }
 454         #endif
 455
 456 #endif
 457                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 458
 459                 #ifdef USE_MPI
 460                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 461
 462                         if (pid == 0) { //only one process should output to screen
 463                 #endif
 464
 465                 //if (output == "square") {  convertMatrix(outputFile); }
 466
 467                 ifstream fileHandle;
 468                 fileHandle.open(outputFile.c_str());
 469                 if(fileHandle) {
 470                         m->gobble(fileHandle);
 471                         if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff.");  m->mothurOutEndLine(); }
 472                 }
 473
 474                 //append the old column file to the new one
 475                 if ((oldfastafile != "") && (column != ""))  {
 476                         //we had to rename the column file so we didnt overwrite above, but we want to keep old name
 477                         if (outputFile == column) {
 478                                 string tempcolumn = column + ".old";
 479                                 m->appendFiles(tempcolumn, outputFile);
 480                                 remove(tempcolumn.c_str());
 481                         }else{
 482                                 m->appendFiles(outputFile, column);
 483                                 remove(outputFile.c_str());
 484                                 outputFile = column;
 485                         }
 486
 487                         if (outputDir != "") {
 488                                 string newOutputName = outputDir + m->getSimpleName(outputFile);
 489                                 rename(outputFile.c_str(), newOutputName.c_str());
 490                                 remove(outputFile.c_str());
 491                                 outputFile = newOutputName;
 492                         }
 493                 }
 494
 495
 496                 #ifdef USE_MPI
 497                         }
 498                 #endif
 499
 500                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 501
 502                 delete distCalculator;
 503
 504                 m->mothurOutEndLine();
 505                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 506                 m->mothurOut(outputFile); m->mothurOutEndLine();
 507                 m->mothurOutEndLine();
 508                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 509
 510
 511                 if (m->isTrue(compress)) {
 512                         m->mothurOut("Compressing..."); m->mothurOutEndLine();
 513                         m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 514                         system(("gzip -v " + outputFile).c_str());
 515                         outputNames.push_back(outputFile + ".gz");
 516                 }else { outputNames.push_back(outputFile); }
 517
 518                 return 0;
 519
 520         }
 521         catch(exception& e) {
 522                 m->errorOut(e, "DistanceCommand", "execute");
 523                 exit(1);
 524         }
 525 }
 526 /**************************************************************************************************/
 527 void DistanceCommand::createProcesses(string filename) {
 528         try {
 529 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 530                 int process = 1;
 531                 processIDS.clear();
 532
 533                 //loop through and create all the processes you want
 534                 while (process != processors) {
 535                         int pid = fork();
 536
 537                         if (pid > 0) {
 538                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 539                                 process++;
 540                         }else if (pid == 0){
 541                                 if (output != "square") {  driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", cutoff); }
 542                                 else { driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", "square"); }
 543                                 exit(0);
 544                         }else {
 545                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
 546                                 perror(" : ");
 547                                 for (int i=0;i<processIDS.size();i++) {  int temp = processIDS[i]; kill (temp, SIGINT); }
 548                                 exit(0);
 549                         }
 550                 }
 551
 552                 //parent does its part
 553                 if (output != "square") {  driver(lines[0].start, lines[0].end, filename, cutoff); }
 554                 else { driver(lines[0].start, lines[0].end, filename, "square"); }
 555
 556
 557                 //force parent to wait until all the processes are done
 558                 for (int i=0;i<processIDS.size();i++) {
 559                         int temp = processIDS[i];
 560                         wait(&temp);
 561                 }
 562
 563                 //append and remove temp files
 564                 for (int i=0;i<processIDS.size();i++) {
 565                         m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
 566                         remove((filename + toString(processIDS[i]) + ".temp").c_str());
 567                 }
 568 #endif
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "DistanceCommand", "createProcesses");
 572                 exit(1);
 573         }
 574 }
 575
 576 /**************************************************************************************************/
 577 /////// need to fix to work with calcs and sequencedb
 578 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 579         try {
 580
 581                 int startTime = time(NULL);
 582
 583                 //column file
 584                 ofstream outFile(dFileName.c_str(), ios::trunc);
 585                 outFile.setf(ios::fixed, ios::showpoint);
 586                 outFile << setprecision(4);
 587
 588                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 589
 590                 for(int i=startLine;i<endLine;i++){
 591                         if(output == "lt")      {
 592                                 string name = alignDB.get(i).getName();
 593                                 if (name.length() < 10) { //pad with spaces to make compatible
 594                                         while (name.length() < 10) {  name += " ";  }
 595                                 }
 596                                 outFile << name << '\t';
 597                         }
 598                         for(int j=0;j<i;j++){
 599
 600                                 if (m->control_pressed) { outFile.close(); return 0;  }
 601
 602                                 //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
 603                                 //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
 604                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 605
 606                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 607                                 double dist = distCalculator->getDist();
 608
 609                                 if(dist <= cutoff){
 610                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 611                                 }
 612                                 if (output == "lt") {  outFile << dist << '\t'; }
 613                         }
 614
 615                         if (output == "lt") { outFile << endl; }
 616
 617                         if(i % 100 == 0){
 618                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 619                         }
 620
 621                 }
 622                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 623
 624                 outFile.close();
 625
 626                 return 1;
 627         }
 628         catch(exception& e) {
 629                 m->errorOut(e, "DistanceCommand", "driver");
 630                 exit(1);
 631         }
 632 }
 633 /**************************************************************************************************/
 634 /////// need to fix to work with calcs and sequencedb
 635 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 636         try {
 637
 638                 int startTime = time(NULL);
 639
 640                 //column file
 641                 ofstream outFile(dFileName.c_str(), ios::trunc);
 642                 outFile.setf(ios::fixed, ios::showpoint);
 643                 outFile << setprecision(4);
 644
 645                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 646
 647                 for(int i=startLine;i<endLine;i++){
 648
 649                         string name = alignDB.get(i).getName();
 650                         //pad with spaces to make compatible
 651                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 652
 653                         outFile << name << '\t';
 654
 655                         for(int j=0;j<alignDB.getNumSeqs();j++){
 656
 657                                 if (m->control_pressed) { outFile.close(); return 0;  }
 658
 659                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 660                                 double dist = distCalculator->getDist();
 661
 662                                 outFile << dist << '\t';
 663                         }
 664
 665                         outFile << endl;
 666
 667                         if(i % 100 == 0){
 668                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 669                         }
 670
 671                 }
 672                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 673
 674                 outFile.close();
 675
 676                 return 1;
 677         }
 678         catch(exception& e) {
 679                 m->errorOut(e, "DistanceCommand", "driver");
 680                 exit(1);
 681         }
 682 }
 683 #ifdef USE_MPI
 684 /**************************************************************************************************/
 685 /////// need to fix to work with calcs and sequencedb
 686 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 687         try {
 688                 MPI_Status status;
 689                 int startTime = time(NULL);
 690
 691                 string outputString = "";
 692
 693                 for(int i=startLine;i<endLine;i++){
 694
 695                         for(int j=0;j<i;j++){
 696
 697                                 if (m->control_pressed) {  return 0;  }
 698
 699                                 //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
 700                                 //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
 701                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 702
 703                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 704                                 double dist = distCalculator->getDist();
 705
 706                                 if(dist <= cutoff){
 707                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 708                                 }
 709                         }
 710
 711                         if(i % 100 == 0){
 712                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 713                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 714                         }
 715
 716
 717                         //send results to parent
 718                         int length = outputString.length();
 719
 720                         char* buf = new char[length];
 721                         memcpy(buf, outputString.c_str(), length);
 722
 723                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 724                         outputString = "";
 725                         delete buf;
 726
 727                 }
 728
 729                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 730                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 731                 return 1;
 732         }
 733         catch(exception& e) {
 734                 m->errorOut(e, "DistanceCommand", "driverMPI");
 735                 exit(1);
 736         }
 737 }
 738 /**************************************************************************************************/
 739 /////// need to fix to work with calcs and sequencedb
 740 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 741         try {
 742                 MPI_Status status;
 743
 744                 MPI_File outMPI;
 745                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 746
 747                 //char* filename = new char[file.length()];
 748                 //memcpy(filename, file.c_str(), file.length());
 749
 750                 char filename[1024];
 751                 strcpy(filename, file.c_str());
 752
 753                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 754                 //delete filename;
 755
 756                 int startTime = time(NULL);
 757
 758                 string outputString = "";
 759                 size = 0;
 760
 761                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 762
 763                 for(int i=startLine;i<endLine;i++){
 764
 765                         string name = alignDB.get(i).getName();
 766                         if (name.length() < 10) { //pad with spaces to make compatible
 767                                 while (name.length() < 10) {  name += " ";  }
 768                         }
 769                         outputString += name + "\t";
 770
 771                         for(int j=0;j<i;j++){
 772
 773                                 if (m->control_pressed) {  return 0;  }
 774
 775                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 776                                 double dist = distCalculator->getDist();
 777
 778                                 outputString += toString(dist) + "\t";
 779                         }
 780
 781                         outputString += "\n";
 782
 783
 784                         if(i % 100 == 0){
 785                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 786                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 787                         }
 788
 789
 790                         //send results to parent
 791                         int length = outputString.length();
 792                         char* buf = new char[length];
 793                         memcpy(buf, outputString.c_str(), length);
 794
 795                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 796                         size += outputString.length();
 797                         outputString = "";
 798                         delete buf;
 799                 }
 800
 801                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 802                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 803                 MPI_File_close(&outMPI);
 804
 805                 return 1;
 806         }
 807         catch(exception& e) {
 808                 m->errorOut(e, "DistanceCommand", "driverMPI");
 809                 exit(1);
 810         }
 811 }
 812 /**************************************************************************************************/
 813 /////// need to fix to work with calcs and sequencedb
 814 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 815         try {
 816                 MPI_Status status;
 817
 818                 MPI_File outMPI;
 819                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 820
 821                 //char* filename = new char[file.length()];
 822                 //memcpy(filename, file.c_str(), file.length());
 823
 824                 char filename[1024];
 825                 strcpy(filename, file.c_str());
 826
 827                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 828                 //delete filename;
 829
 830                 int startTime = time(NULL);
 831
 832                 string outputString = "";
 833                 size = 0;
 834
 835                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 836
 837                 for(int i=startLine;i<endLine;i++){
 838
 839                         string name = alignDB.get(i).getName();
 840                         if (name.length() < 10) { //pad with spaces to make compatible
 841                                 while (name.length() < 10) {  name += " ";  }
 842                         }
 843                         outputString += name + "\t";
 844
 845                         for(int j=0;j<alignDB.getNumSeqs();j++){
 846
 847                                 if (m->control_pressed) {  return 0;  }
 848
 849                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 850                                 double dist = distCalculator->getDist();
 851
 852                                 outputString += toString(dist) + "\t";
 853                         }
 854
 855                         outputString += "\n";
 856
 857
 858                         if(i % 100 == 0){
 859                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 860                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 861                         }
 862
 863
 864                         //send results to parent
 865                         int length = outputString.length();
 866                         char* buf = new char[length];
 867                         memcpy(buf, outputString.c_str(), length);
 868
 869                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 870                         size += outputString.length();
 871                         outputString = "";
 872                         delete buf;
 873                 }
 874
 875                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 876                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 877                 MPI_File_close(&outMPI);
 878
 879                 return 1;
 880         }
 881         catch(exception& e) {
 882                 m->errorOut(e, "DistanceCommand", "driverMPI");
 883                 exit(1);
 884         }
 885 }
 886 #endif
 887 /**************************************************************************************************
 888 int DistanceCommand::convertMatrix(string outputFile) {
 889         try{
 890
 891                 //sort file by first column so the distances for each row are together
 892                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 893
 894                 //use the unix sort
 895                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 896                         string command = "sort -n " + outputFile + " -o " + outfile;
 897                         system(command.c_str());
 898                 #else //sort using windows sort
 899                         string command = "sort " + outputFile + " /O " + outfile;
 900                         system(command.c_str());
 901                 #endif
 902
 903
 904                 //output to new file distance for each row and save positions in file where new row begins
 905                 ifstream in;
 906                 m->openInputFile(outfile, in);
 907
 908                 ofstream out;
 909                 m->openOutputFile(outputFile, out);
 910
 911                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 912
 913                 out << alignDB.getNumSeqs() << endl;
 914
 915                 //get first currentRow
 916                 string first, currentRow, second;
 917                 float dist;
 918                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 919                 map<string, float>::iterator it;
 920
 921                 in >> first;
 922                 currentRow = first;
 923
 924                 rowDists[first] = 0.00; //distance to yourself is 0.0
 925
 926                 in.seekg(0);
 927                 //m->openInputFile(outfile, in);
 928
 929                 while(!in.eof()) {
 930                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 931
 932                         in >> first >> second >> dist; m->gobble(in);
 933
 934                         if (first != currentRow) {
 935                                 //print out last row
 936                                 out << currentRow << '\t'; //print name
 937
 938                                 //print dists
 939                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 940                                         out << it->second << '\t';
 941                                 }
 942                                 out << endl;
 943
 944                                 //start new row
 945                                 currentRow = first;
 946                                 rowDists.clear();
 947                                 rowDists[first] = 0.00;
 948                                 rowDists[second] = dist;
 949                         }else{
 950                                 rowDists[second] = dist;
 951                         }
 952                 }
 953                 //print out last row
 954                 out << currentRow << '\t'; //print name
 955
 956                 //print dists
 957                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 958                         out << it->second << '\t';
 959                 }
 960                 out << endl;
 961
 962                 in.close();
 963                 out.close();
 964
 965                 remove(outfile.c_str());
 966
 967                 return 1;
 968
 969         }
 970         catch(exception& e) {
 971                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 972                 exit(1);
 973         }
 974 }
 975 /**************************************************************************************************
 976 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 977         try{
 978
 979                 //sort file by first column so the distances for each row are together
 980                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 981
 982                 //use the unix sort
 983                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 984                         string command = "sort -n " + outputFile + " -o " + outfile;
 985                         system(command.c_str());
 986                 #else //sort using windows sort
 987                         string command = "sort " + outputFile + " /O " + outfile;
 988                         system(command.c_str());
 989                 #endif
 990
 991
 992                 //output to new file distance for each row and save positions in file where new row begins
 993                 ifstream in;
 994                 m->openInputFile(outfile, in);
 995
 996                 ofstream out;
 997                 m->openOutputFile(outputFile, out);
 998
 999                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1000
1001                 out << alignDB.getNumSeqs() << endl;
1002
1003                 //get first currentRow
1004                 string first, currentRow, second;
1005                 float dist;
1006                 int i, j;
1007                 i = 0; j = 0;
1008                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
1009                 map<string, float>::iterator it;
1010
1011                 in >> first;
1012                 currentRow = first;
1013
1014                 rowDists[first] = 0.00; //distance to yourself is 0.0
1015
1016                 in.seekg(0);
1017                 //m->openInputFile(outfile, in);
1018
1019                 while(!in.eof()) {
1020                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
1021
1022                         in >> first >> second >> dist; m->gobble(in);
1023
1024                         if (first != currentRow) {
1025                                 //print out last row
1026                                 out << currentRow << '\t'; //print name
1027
1028                                 //print dists
1029                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1030                                         if (j >= i) { break; }
1031                                         out << it->second << '\t';
1032                                         j++;
1033                                 }
1034                                 out << endl;
1035
1036                                 //start new row
1037                                 currentRow = first;
1038                                 rowDists.clear();
1039                                 rowDists[first] = 0.00;
1040                                 rowDists[second] = dist;
1041                                 j = 0;
1042                                 i++;
1043                         }else{
1044                                 rowDists[second] = dist;
1045                         }
1046                 }
1047                 //print out last row
1048                 out << currentRow << '\t'; //print name
1049
1050                 //print dists
1051                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1052                         out << it->second << '\t';
1053                 }
1054                 out << endl;
1055
1056                 in.close();
1057                 out.close();
1058
1059                 remove(outfile.c_str());
1060
1061                 return 1;
1062
1063         }
1064         catch(exception& e) {
1065                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
1066                 exit(1);
1067         }
1068 }
1069 /**************************************************************************************************/
1070 //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,
1071 //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.
1072 //also check to make sure the 2 files have the same alignment length.
1073 bool DistanceCommand::sanityCheck() {
1074         try{
1075                 bool good = true;
1076
1077                 //make sure the 2 fasta files have the same alignment length
1078                 ifstream in;
1079                 m->openInputFile(fastafile, in);
1080                 int fastaAlignLength = 0;
1081                 if (in) {
1082                         Sequence tempIn(in);
1083                         fastaAlignLength = tempIn.getAligned().length();
1084                 }
1085                 in.close();
1086
1087                 ifstream in2;
1088                 m->openInputFile(oldfastafile, in2);
1089                 int oldfastaAlignLength = 0;
1090                 if (in2) {
1091                         Sequence tempIn2(in2);
1092                         oldfastaAlignLength = tempIn2.getAligned().length();
1093                 }
1094                 in2.close();
1095
1096                 if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false;  }
1097
1098                 //read fasta file and save names as well as adding them to the alignDB
1099                 set<string> namesOldFasta;
1100
1101                 ifstream inFasta;
1102                 m->openInputFile(oldfastafile, inFasta);
1103
1104                 while (!inFasta.eof()) {
1105                         if (m->control_pressed) {  inFasta.close(); return good;  }
1106
1107                         Sequence temp(inFasta);
1108
1109                         if (temp.getName() != "") {
1110                                 namesOldFasta.insert(temp.getName());  //save name
1111                                 alignDB.push_back(temp);  //add to DB
1112                         }
1113
1114                         m->gobble(inFasta);
1115                 }
1116
1117                 inFasta.close();
1118
1119                 //read through the column file checking names and removing distances above the cutoff
1120                 ifstream inDist;
1121                 m->openInputFile(column, inDist);
1122
1123                 ofstream outDist;
1124                 string outputFile = column + ".temp";
1125                 m->openOutputFile(outputFile, outDist);
1126
1127                 string name1, name2;
1128                 float dist;
1129                 while (!inDist.eof()) {
1130                         if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
1131
1132                         inDist >> name1 >> name2 >> dist; m->gobble(inDist);
1133
1134                         //both names are in fasta file and distance is below cutoff
1135                         if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) {  good = false; break;  }
1136                         else{
1137                                 if (dist <= cutoff) {
1138                                         outDist << name1 << '\t' << name2 << '\t' << dist << endl;
1139                                 }
1140                         }
1141                 }
1142
1143                 inDist.close();
1144                 outDist.close();
1145
1146                 if (good) {
1147                         remove(column.c_str());
1148                         rename(outputFile.c_str(), column.c_str());
1149                 }else{
1150                         remove(outputFile.c_str()); //temp file is bad because file mismatch above
1151                 }
1152
1153                 return good;
1154
1155         }
1156         catch(exception& e) {
1157                 m->errorOut(e, "DistanceCommand", "sanityCheck");
1158                 exit(1);
1159         }
1160 }
1161 /**************************************************************************************************/
1162
1163
1164
1165