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                         unsigned long int numDists = 0;
 427
 428                         if (output == "square") {
 429                                  numDists = numSeqs * numSeqs;
 430                         }else {
 431                                 for(int i=0;i<numSeqs;i++){
 432                                         for(int j=0;j<i;j++){
 433                                                 numDists++;
 434                                                 if (numDists > processors) { break; }
 435                                         }
 436                                 }
 437                         }
 438
 439                         if (numDists < processors) { processors = numDists; }
 440
 441                         for (int i = 0; i < processors; i++) {
 442                                 lines.push_back(new linePair());
 443                                 if (output != "square") {
 444                                         lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 445                                         lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 446                                 }else{
 447                                         lines[i]->start = int ((float(i)/float(processors)) * numSeqs);
 448                                         lines[i]->end = int ((float(i+1)/float(processors)) * numSeqs);
 449                                 }
 450                                 //cout << i << '\t' << lines[i]->start << '\t' << lines[i]->end << endl;
 451                         }
 452
 453                         createProcesses(outputFile);
 454                 }
 455         #else
 456                 //ifstream inFASTA;
 457                 if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 458                 else { driver(0, numSeqs, outputFile, "square");  }
 459         #endif
 460
 461 #endif
 462                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 463
 464                 #ifdef USE_MPI
 465                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 466
 467                         if (pid == 0) { //only one process should output to screen
 468                 #endif
 469
 470                 //if (output == "square") {  convertMatrix(outputFile); }
 471
 472                 ifstream fileHandle;
 473                 fileHandle.open(outputFile.c_str());
 474                 if(fileHandle) {
 475                         m->gobble(fileHandle);
 476                         if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff.");  m->mothurOutEndLine(); }
 477                 }
 478
 479                 //append the old column file to the new one
 480                 if ((oldfastafile != "") && (column != ""))  {
 481                         //we had to rename the column file so we didnt overwrite above, but we want to keep old name
 482                         if (outputFile == column) {
 483                                 string tempcolumn = column + ".old";
 484                                 m->appendFiles(tempcolumn, outputFile);
 485                                 remove(tempcolumn.c_str());
 486                         }else{
 487                                 m->appendFiles(outputFile, column);
 488                                 remove(outputFile.c_str());
 489                                 outputFile = column;
 490                         }
 491
 492                         if (outputDir != "") {
 493                                 string newOutputName = outputDir + m->getSimpleName(outputFile);
 494                                 rename(outputFile.c_str(), newOutputName.c_str());
 495                                 remove(outputFile.c_str());
 496                                 outputFile = newOutputName;
 497                         }
 498                 }
 499
 500
 501                 #ifdef USE_MPI
 502                         }
 503                 #endif
 504
 505                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 506
 507                 delete distCalculator;
 508
 509                 m->mothurOutEndLine();
 510                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 511                 m->mothurOut(outputFile); m->mothurOutEndLine();
 512                 m->mothurOutEndLine();
 513                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 514
 515
 516                 if (m->isTrue(compress)) {
 517                         m->mothurOut("Compressing..."); m->mothurOutEndLine();
 518                         m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 519                         system(("gzip -v " + outputFile).c_str());
 520                         outputNames.push_back(outputFile + ".gz");
 521                 }else { outputNames.push_back(outputFile); }
 522
 523                 return 0;
 524
 525         }
 526         catch(exception& e) {
 527                 m->errorOut(e, "DistanceCommand", "execute");
 528                 exit(1);
 529         }
 530 }
 531 /**************************************************************************************************/
 532 void DistanceCommand::createProcesses(string filename) {
 533         try {
 534 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 535                 int process = 1;
 536                 processIDS.clear();
 537
 538                 //loop through and create all the processes you want
 539                 while (process != processors) {
 540                         int pid = fork();
 541
 542                         if (pid > 0) {
 543                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 544                                 process++;
 545                         }else if (pid == 0){
 546                                 if (output != "square") {  driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff); }
 547                                 else { driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", "square"); }
 548                                 exit(0);
 549                         }else {
 550                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
 551                                 perror(" : ");
 552                                 for (int i=0;i<processIDS.size();i++) {  int temp = processIDS[i]; kill (temp, SIGINT); }
 553                                 exit(0);
 554                         }
 555                 }
 556
 557                 //parent does its part
 558                 if (output != "square") {  driver(lines[0]->start, lines[0]->end, filename, cutoff); }
 559                 else { driver(lines[0]->start, lines[0]->end, filename, "square"); }
 560
 561
 562                 //force parent to wait until all the processes are done
 563                 for (int i=0;i<processIDS.size();i++) {
 564                         int temp = processIDS[i];
 565                         wait(&temp);
 566                 }
 567
 568                 //append and remove temp files
 569                 for (int i=0;i<processIDS.size();i++) {
 570                         m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
 571                         remove((filename + toString(processIDS[i]) + ".temp").c_str());
 572                 }
 573 #endif
 574         }
 575         catch(exception& e) {
 576                 m->errorOut(e, "DistanceCommand", "createProcesses");
 577                 exit(1);
 578         }
 579 }
 580
 581 /**************************************************************************************************/
 582 /////// need to fix to work with calcs and sequencedb
 583 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 584         try {
 585
 586                 int startTime = time(NULL);
 587
 588                 //column file
 589                 ofstream outFile(dFileName.c_str(), ios::trunc);
 590                 outFile.setf(ios::fixed, ios::showpoint);
 591                 outFile << setprecision(4);
 592
 593                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 594
 595                 for(int i=startLine;i<endLine;i++){
 596                         if(output == "lt")      {
 597                                 string name = alignDB.get(i).getName();
 598                                 if (name.length() < 10) { //pad with spaces to make compatible
 599                                         while (name.length() < 10) {  name += " ";  }
 600                                 }
 601                                 outFile << name << '\t';
 602                         }
 603                         for(int j=0;j<i;j++){
 604
 605                                 if (m->control_pressed) { outFile.close(); return 0;  }
 606
 607                                 //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
 608                                 //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
 609                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 610
 611                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 612                                 double dist = distCalculator->getDist();
 613
 614                                 if(dist <= cutoff){
 615                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 616                                 }
 617                                 if (output == "lt") {  outFile << dist << '\t'; }
 618                         }
 619
 620                         if (output == "lt") { outFile << endl; }
 621
 622                         if(i % 100 == 0){
 623                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 624                         }
 625
 626                 }
 627                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 628
 629                 outFile.close();
 630
 631                 return 1;
 632         }
 633         catch(exception& e) {
 634                 m->errorOut(e, "DistanceCommand", "driver");
 635                 exit(1);
 636         }
 637 }
 638 /**************************************************************************************************/
 639 /////// need to fix to work with calcs and sequencedb
 640 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 641         try {
 642
 643                 int startTime = time(NULL);
 644
 645                 //column file
 646                 ofstream outFile(dFileName.c_str(), ios::trunc);
 647                 outFile.setf(ios::fixed, ios::showpoint);
 648                 outFile << setprecision(4);
 649
 650                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 651
 652                 for(int i=startLine;i<endLine;i++){
 653
 654                         string name = alignDB.get(i).getName();
 655                         //pad with spaces to make compatible
 656                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 657
 658                         outFile << name << '\t';
 659
 660                         for(int j=0;j<alignDB.getNumSeqs();j++){
 661
 662                                 if (m->control_pressed) { outFile.close(); return 0;  }
 663
 664                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 665                                 double dist = distCalculator->getDist();
 666
 667                                 outFile << dist << '\t';
 668                         }
 669
 670                         outFile << endl;
 671
 672                         if(i % 100 == 0){
 673                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 674                         }
 675
 676                 }
 677                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 678
 679                 outFile.close();
 680
 681                 return 1;
 682         }
 683         catch(exception& e) {
 684                 m->errorOut(e, "DistanceCommand", "driver");
 685                 exit(1);
 686         }
 687 }
 688 #ifdef USE_MPI
 689 /**************************************************************************************************/
 690 /////// need to fix to work with calcs and sequencedb
 691 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 692         try {
 693                 MPI_Status status;
 694                 int startTime = time(NULL);
 695
 696                 string outputString = "";
 697
 698                 for(int i=startLine;i<endLine;i++){
 699
 700                         for(int j=0;j<i;j++){
 701
 702                                 if (m->control_pressed) {  return 0;  }
 703
 704                                 //if 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
 705                                 //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
 706                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 707
 708                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 709                                 double dist = distCalculator->getDist();
 710
 711                                 if(dist <= cutoff){
 712                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 713                                 }
 714                         }
 715
 716                         if(i % 100 == 0){
 717                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 718                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 719                         }
 720
 721
 722                         //send results to parent
 723                         int length = outputString.length();
 724
 725                         char* buf = new char[length];
 726                         memcpy(buf, outputString.c_str(), length);
 727
 728                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 729                         outputString = "";
 730                         delete buf;
 731
 732                 }
 733
 734                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 735                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 736                 return 1;
 737         }
 738         catch(exception& e) {
 739                 m->errorOut(e, "DistanceCommand", "driverMPI");
 740                 exit(1);
 741         }
 742 }
 743 /**************************************************************************************************/
 744 /////// need to fix to work with calcs and sequencedb
 745 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 746         try {
 747                 MPI_Status status;
 748
 749                 MPI_File outMPI;
 750                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 751
 752                 //char* filename = new char[file.length()];
 753                 //memcpy(filename, file.c_str(), file.length());
 754
 755                 char filename[1024];
 756                 strcpy(filename, file.c_str());
 757
 758                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 759                 //delete filename;
 760
 761                 int startTime = time(NULL);
 762
 763                 string outputString = "";
 764                 size = 0;
 765
 766                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 767
 768                 for(int i=startLine;i<endLine;i++){
 769
 770                         string name = alignDB.get(i).getName();
 771                         if (name.length() < 10) { //pad with spaces to make compatible
 772                                 while (name.length() < 10) {  name += " ";  }
 773                         }
 774                         outputString += name + "\t";
 775
 776                         for(int j=0;j<i;j++){
 777
 778                                 if (m->control_pressed) {  return 0;  }
 779
 780                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 781                                 double dist = distCalculator->getDist();
 782
 783                                 outputString += toString(dist) + "\t";
 784                         }
 785
 786                         outputString += "\n";
 787
 788
 789                         if(i % 100 == 0){
 790                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 791                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 792                         }
 793
 794
 795                         //send results to parent
 796                         int length = outputString.length();
 797                         char* buf = new char[length];
 798                         memcpy(buf, outputString.c_str(), length);
 799
 800                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 801                         size += outputString.length();
 802                         outputString = "";
 803                         delete buf;
 804                 }
 805
 806                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 807                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 808                 MPI_File_close(&outMPI);
 809
 810                 return 1;
 811         }
 812         catch(exception& e) {
 813                 m->errorOut(e, "DistanceCommand", "driverMPI");
 814                 exit(1);
 815         }
 816 }
 817 /**************************************************************************************************/
 818 /////// need to fix to work with calcs and sequencedb
 819 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 820         try {
 821                 MPI_Status status;
 822
 823                 MPI_File outMPI;
 824                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 825
 826                 //char* filename = new char[file.length()];
 827                 //memcpy(filename, file.c_str(), file.length());
 828
 829                 char filename[1024];
 830                 strcpy(filename, file.c_str());
 831
 832                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 833                 //delete filename;
 834
 835                 int startTime = time(NULL);
 836
 837                 string outputString = "";
 838                 size = 0;
 839
 840                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 841
 842                 for(int i=startLine;i<endLine;i++){
 843
 844                         string name = alignDB.get(i).getName();
 845                         if (name.length() < 10) { //pad with spaces to make compatible
 846                                 while (name.length() < 10) {  name += " ";  }
 847                         }
 848                         outputString += name + "\t";
 849
 850                         for(int j=0;j<alignDB.getNumSeqs();j++){
 851
 852                                 if (m->control_pressed) {  return 0;  }
 853
 854                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 855                                 double dist = distCalculator->getDist();
 856
 857                                 outputString += toString(dist) + "\t";
 858                         }
 859
 860                         outputString += "\n";
 861
 862
 863                         if(i % 100 == 0){
 864                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 865                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 866                         }
 867
 868
 869                         //send results to parent
 870                         int length = outputString.length();
 871                         char* buf = new char[length];
 872                         memcpy(buf, outputString.c_str(), length);
 873
 874                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 875                         size += outputString.length();
 876                         outputString = "";
 877                         delete buf;
 878                 }
 879
 880                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 881                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 882                 MPI_File_close(&outMPI);
 883
 884                 return 1;
 885         }
 886         catch(exception& e) {
 887                 m->errorOut(e, "DistanceCommand", "driverMPI");
 888                 exit(1);
 889         }
 890 }
 891 #endif
 892 /**************************************************************************************************
 893 int DistanceCommand::convertMatrix(string outputFile) {
 894         try{
 895
 896                 //sort file by first column so the distances for each row are together
 897                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 898
 899                 //use the unix sort
 900                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 901                         string command = "sort -n " + outputFile + " -o " + outfile;
 902                         system(command.c_str());
 903                 #else //sort using windows sort
 904                         string command = "sort " + outputFile + " /O " + outfile;
 905                         system(command.c_str());
 906                 #endif
 907
 908
 909                 //output to new file distance for each row and save positions in file where new row begins
 910                 ifstream in;
 911                 m->openInputFile(outfile, in);
 912
 913                 ofstream out;
 914                 m->openOutputFile(outputFile, out);
 915
 916                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 917
 918                 out << alignDB.getNumSeqs() << endl;
 919
 920                 //get first currentRow
 921                 string first, currentRow, second;
 922                 float dist;
 923                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 924                 map<string, float>::iterator it;
 925
 926                 in >> first;
 927                 currentRow = first;
 928
 929                 rowDists[first] = 0.00; //distance to yourself is 0.0
 930
 931                 in.seekg(0);
 932                 //m->openInputFile(outfile, in);
 933
 934                 while(!in.eof()) {
 935                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 936
 937                         in >> first >> second >> dist; m->gobble(in);
 938
 939                         if (first != currentRow) {
 940                                 //print out last row
 941                                 out << currentRow << '\t'; //print name
 942
 943                                 //print dists
 944                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 945                                         out << it->second << '\t';
 946                                 }
 947                                 out << endl;
 948
 949                                 //start new row
 950                                 currentRow = first;
 951                                 rowDists.clear();
 952                                 rowDists[first] = 0.00;
 953                                 rowDists[second] = dist;
 954                         }else{
 955                                 rowDists[second] = dist;
 956                         }
 957                 }
 958                 //print out last row
 959                 out << currentRow << '\t'; //print name
 960
 961                 //print dists
 962                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 963                         out << it->second << '\t';
 964                 }
 965                 out << endl;
 966
 967                 in.close();
 968                 out.close();
 969
 970                 remove(outfile.c_str());
 971
 972                 return 1;
 973
 974         }
 975         catch(exception& e) {
 976                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 977                 exit(1);
 978         }
 979 }
 980 /**************************************************************************************************
 981 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 982         try{
 983
 984                 //sort file by first column so the distances for each row are together
 985                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 986
 987                 //use the unix sort
 988                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 989                         string command = "sort -n " + outputFile + " -o " + outfile;
 990                         system(command.c_str());
 991                 #else //sort using windows sort
 992                         string command = "sort " + outputFile + " /O " + outfile;
 993                         system(command.c_str());
 994                 #endif
 995
 996
 997                 //output to new file distance for each row and save positions in file where new row begins
 998                 ifstream in;
 999                 m->openInputFile(outfile, in);
1000
1001                 ofstream out;
1002                 m->openOutputFile(outputFile, out);
1003
1004                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1005
1006                 out << alignDB.getNumSeqs() << endl;
1007
1008                 //get first currentRow
1009                 string first, currentRow, second;
1010                 float dist;
1011                 int i, j;
1012                 i = 0; j = 0;
1013                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
1014                 map<string, float>::iterator it;
1015
1016                 in >> first;
1017                 currentRow = first;
1018
1019                 rowDists[first] = 0.00; //distance to yourself is 0.0
1020
1021                 in.seekg(0);
1022                 //m->openInputFile(outfile, in);
1023
1024                 while(!in.eof()) {
1025                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
1026
1027                         in >> first >> second >> dist; m->gobble(in);
1028
1029                         if (first != currentRow) {
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                                         if (j >= i) { break; }
1036                                         out << it->second << '\t';
1037                                         j++;
1038                                 }
1039                                 out << endl;
1040
1041                                 //start new row
1042                                 currentRow = first;
1043                                 rowDists.clear();
1044                                 rowDists[first] = 0.00;
1045                                 rowDists[second] = dist;
1046                                 j = 0;
1047                                 i++;
1048                         }else{
1049                                 rowDists[second] = dist;
1050                         }
1051                 }
1052                 //print out last row
1053                 out << currentRow << '\t'; //print name
1054
1055                 //print dists
1056                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1057                         out << it->second << '\t';
1058                 }
1059                 out << endl;
1060
1061                 in.close();
1062                 out.close();
1063
1064                 remove(outfile.c_str());
1065
1066                 return 1;
1067
1068         }
1069         catch(exception& e) {
1070                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
1071                 exit(1);
1072         }
1073 }
1074 /**************************************************************************************************/
1075 //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,
1076 //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.
1077 //also check to make sure the 2 files have the same alignment length.
1078 bool DistanceCommand::sanityCheck() {
1079         try{
1080                 bool good = true;
1081
1082                 //make sure the 2 fasta files have the same alignment length
1083                 ifstream in;
1084                 m->openInputFile(fastafile, in);
1085                 int fastaAlignLength = 0;
1086                 if (in) {
1087                         Sequence tempIn(in);
1088                         fastaAlignLength = tempIn.getAligned().length();
1089                 }
1090                 in.close();
1091
1092                 ifstream in2;
1093                 m->openInputFile(oldfastafile, in2);
1094                 int oldfastaAlignLength = 0;
1095                 if (in2) {
1096                         Sequence tempIn2(in2);
1097                         oldfastaAlignLength = tempIn2.getAligned().length();
1098                 }
1099                 in2.close();
1100
1101                 if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false;  }
1102
1103                 //read fasta file and save names as well as adding them to the alignDB
1104                 set<string> namesOldFasta;
1105
1106                 ifstream inFasta;
1107                 m->openInputFile(oldfastafile, inFasta);
1108
1109                 while (!inFasta.eof()) {
1110                         if (m->control_pressed) {  inFasta.close(); return good;  }
1111
1112                         Sequence temp(inFasta);
1113
1114                         if (temp.getName() != "") {
1115                                 namesOldFasta.insert(temp.getName());  //save name
1116                                 alignDB.push_back(temp);  //add to DB
1117                         }
1118
1119                         m->gobble(inFasta);
1120                 }
1121
1122                 inFasta.close();
1123
1124                 //read through the column file checking names and removing distances above the cutoff
1125                 ifstream inDist;
1126                 m->openInputFile(column, inDist);
1127
1128                 ofstream outDist;
1129                 string outputFile = column + ".temp";
1130                 m->openOutputFile(outputFile, outDist);
1131
1132                 string name1, name2;
1133                 float dist;
1134                 while (!inDist.eof()) {
1135                         if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
1136
1137                         inDist >> name1 >> name2 >> dist; m->gobble(inDist);
1138
1139                         //both names are in fasta file and distance is below cutoff
1140                         if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) {  good = false; break;  }
1141                         else{
1142                                 if (dist <= cutoff) {
1143                                         outDist << name1 << '\t' << name2 << '\t' << dist << endl;
1144                                 }
1145                         }
1146                 }
1147
1148                 inDist.close();
1149                 outDist.close();
1150
1151                 if (good) {
1152                         remove(column.c_str());
1153                         rename(outputFile.c_str(), column.c_str());
1154                 }else{
1155                         remove(outputFile.c_str()); //temp file is bad because file mismatch above
1156                 }
1157
1158         }
1159         catch(exception& e) {
1160                 m->errorOut(e, "DistanceCommand", "m->appendFiles");
1161                 exit(1);
1162         }
1163 }
1164 /**************************************************************************************************/
1165
1166
1167
1168