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                 //set phylip file as new current phylipfile
 505                 string current = "";
 506                 itTypes = outputTypes.find("phylip");
 507                 if (itTypes != outputTypes.end()) {
 508                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
 509                 }
 510
 511                 //set column file as new current columnfile
 512                 itTypes = outputTypes.find("column");
 513                 if (itTypes != outputTypes.end()) {
 514                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setColumnFile(current); }
 515                 }
 516
 517                 m->mothurOutEndLine();
 518                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 519                 m->mothurOut(outputFile); m->mothurOutEndLine();
 520                 m->mothurOutEndLine();
 521                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 522
 523
 524                 if (m->isTrue(compress)) {
 525                         m->mothurOut("Compressing..."); m->mothurOutEndLine();
 526                         m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 527                         system(("gzip -v " + outputFile).c_str());
 528                         outputNames.push_back(outputFile + ".gz");
 529                 }else { outputNames.push_back(outputFile); }
 530
 531                 return 0;
 532
 533         }
 534         catch(exception& e) {
 535                 m->errorOut(e, "DistanceCommand", "execute");
 536                 exit(1);
 537         }
 538 }
 539 /**************************************************************************************************/
 540 void DistanceCommand::createProcesses(string filename) {
 541         try {
 542 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 543                 int process = 1;
 544                 processIDS.clear();
 545
 546                 //loop through and create all the processes you want
 547                 while (process != processors) {
 548                         int pid = fork();
 549
 550                         if (pid > 0) {
 551                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 552                                 process++;
 553                         }else if (pid == 0){
 554                                 if (output != "square") {  driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", cutoff); }
 555                                 else { driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", "square"); }
 556                                 exit(0);
 557                         }else {
 558                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
 559                                 perror(" : ");
 560                                 for (int i=0;i<processIDS.size();i++) {  int temp = processIDS[i]; kill (temp, SIGINT); }
 561                                 exit(0);
 562                         }
 563                 }
 564
 565                 //parent does its part
 566                 if (output != "square") {  driver(lines[0].start, lines[0].end, filename, cutoff); }
 567                 else { driver(lines[0].start, lines[0].end, filename, "square"); }
 568
 569
 570                 //force parent to wait until all the processes are done
 571                 for (int i=0;i<processIDS.size();i++) {
 572                         int temp = processIDS[i];
 573                         wait(&temp);
 574                 }
 575
 576                 //append and remove temp files
 577                 for (int i=0;i<processIDS.size();i++) {
 578                         m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
 579                         remove((filename + toString(processIDS[i]) + ".temp").c_str());
 580                 }
 581 #endif
 582         }
 583         catch(exception& e) {
 584                 m->errorOut(e, "DistanceCommand", "createProcesses");
 585                 exit(1);
 586         }
 587 }
 588
 589 /**************************************************************************************************/
 590 /////// need to fix to work with calcs and sequencedb
 591 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 592         try {
 593
 594                 int startTime = time(NULL);
 595
 596                 //column file
 597                 ofstream outFile(dFileName.c_str(), ios::trunc);
 598                 outFile.setf(ios::fixed, ios::showpoint);
 599                 outFile << setprecision(4);
 600
 601                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 602
 603                 for(int i=startLine;i<endLine;i++){
 604                         if(output == "lt")      {
 605                                 string name = alignDB.get(i).getName();
 606                                 if (name.length() < 10) { //pad with spaces to make compatible
 607                                         while (name.length() < 10) {  name += " ";  }
 608                                 }
 609                                 outFile << name << '\t';
 610                         }
 611                         for(int j=0;j<i;j++){
 612
 613                                 if (m->control_pressed) { outFile.close(); return 0;  }
 614
 615                                 //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
 616                                 //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
 617                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 618
 619                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 620                                 double dist = distCalculator->getDist();
 621
 622                                 if(dist <= cutoff){
 623                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 624                                 }
 625                                 if (output == "lt") {  outFile << dist << '\t'; }
 626                         }
 627
 628                         if (output == "lt") { outFile << endl; }
 629
 630                         if(i % 100 == 0){
 631                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 632                         }
 633
 634                 }
 635                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 636
 637                 outFile.close();
 638
 639                 return 1;
 640         }
 641         catch(exception& e) {
 642                 m->errorOut(e, "DistanceCommand", "driver");
 643                 exit(1);
 644         }
 645 }
 646 /**************************************************************************************************/
 647 /////// need to fix to work with calcs and sequencedb
 648 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 649         try {
 650
 651                 int startTime = time(NULL);
 652
 653                 //column file
 654                 ofstream outFile(dFileName.c_str(), ios::trunc);
 655                 outFile.setf(ios::fixed, ios::showpoint);
 656                 outFile << setprecision(4);
 657
 658                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 659
 660                 for(int i=startLine;i<endLine;i++){
 661
 662                         string name = alignDB.get(i).getName();
 663                         //pad with spaces to make compatible
 664                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 665
 666                         outFile << name << '\t';
 667
 668                         for(int j=0;j<alignDB.getNumSeqs();j++){
 669
 670                                 if (m->control_pressed) { outFile.close(); return 0;  }
 671
 672                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 673                                 double dist = distCalculator->getDist();
 674
 675                                 outFile << dist << '\t';
 676                         }
 677
 678                         outFile << endl;
 679
 680                         if(i % 100 == 0){
 681                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 682                         }
 683
 684                 }
 685                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 686
 687                 outFile.close();
 688
 689                 return 1;
 690         }
 691         catch(exception& e) {
 692                 m->errorOut(e, "DistanceCommand", "driver");
 693                 exit(1);
 694         }
 695 }
 696 #ifdef USE_MPI
 697 /**************************************************************************************************/
 698 /////// need to fix to work with calcs and sequencedb
 699 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 700         try {
 701                 MPI_Status status;
 702                 int startTime = time(NULL);
 703
 704                 string outputString = "";
 705
 706                 for(int i=startLine;i<endLine;i++){
 707
 708                         for(int j=0;j<i;j++){
 709
 710                                 if (m->control_pressed) {  return 0;  }
 711
 712                                 //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
 713                                 //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
 714                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 715
 716                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 717                                 double dist = distCalculator->getDist();
 718
 719                                 if(dist <= cutoff){
 720                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 721                                 }
 722                         }
 723
 724                         if(i % 100 == 0){
 725                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 726                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 727                         }
 728
 729
 730                         //send results to parent
 731                         int length = outputString.length();
 732
 733                         char* buf = new char[length];
 734                         memcpy(buf, outputString.c_str(), length);
 735
 736                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 737                         outputString = "";
 738                         delete buf;
 739
 740                 }
 741
 742                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 743                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 744                 return 1;
 745         }
 746         catch(exception& e) {
 747                 m->errorOut(e, "DistanceCommand", "driverMPI");
 748                 exit(1);
 749         }
 750 }
 751 /**************************************************************************************************/
 752 /////// need to fix to work with calcs and sequencedb
 753 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 754         try {
 755                 MPI_Status status;
 756
 757                 MPI_File outMPI;
 758                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 759
 760                 //char* filename = new char[file.length()];
 761                 //memcpy(filename, file.c_str(), file.length());
 762
 763                 char filename[1024];
 764                 strcpy(filename, file.c_str());
 765
 766                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 767                 //delete filename;
 768
 769                 int startTime = time(NULL);
 770
 771                 string outputString = "";
 772                 size = 0;
 773
 774                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 775
 776                 for(int i=startLine;i<endLine;i++){
 777
 778                         string name = alignDB.get(i).getName();
 779                         if (name.length() < 10) { //pad with spaces to make compatible
 780                                 while (name.length() < 10) {  name += " ";  }
 781                         }
 782                         outputString += name + "\t";
 783
 784                         for(int j=0;j<i;j++){
 785
 786                                 if (m->control_pressed) {  return 0;  }
 787
 788                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 789                                 double dist = distCalculator->getDist();
 790
 791                                 outputString += toString(dist) + "\t";
 792                         }
 793
 794                         outputString += "\n";
 795
 796
 797                         if(i % 100 == 0){
 798                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 799                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 800                         }
 801
 802
 803                         //send results to parent
 804                         int length = outputString.length();
 805                         char* buf = new char[length];
 806                         memcpy(buf, outputString.c_str(), length);
 807
 808                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 809                         size += outputString.length();
 810                         outputString = "";
 811                         delete buf;
 812                 }
 813
 814                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 815                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 816                 MPI_File_close(&outMPI);
 817
 818                 return 1;
 819         }
 820         catch(exception& e) {
 821                 m->errorOut(e, "DistanceCommand", "driverMPI");
 822                 exit(1);
 823         }
 824 }
 825 /**************************************************************************************************/
 826 /////// need to fix to work with calcs and sequencedb
 827 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 828         try {
 829                 MPI_Status status;
 830
 831                 MPI_File outMPI;
 832                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 833
 834                 //char* filename = new char[file.length()];
 835                 //memcpy(filename, file.c_str(), file.length());
 836
 837                 char filename[1024];
 838                 strcpy(filename, file.c_str());
 839
 840                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 841                 //delete filename;
 842
 843                 int startTime = time(NULL);
 844
 845                 string outputString = "";
 846                 size = 0;
 847
 848                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 849
 850                 for(int i=startLine;i<endLine;i++){
 851
 852                         string name = alignDB.get(i).getName();
 853                         if (name.length() < 10) { //pad with spaces to make compatible
 854                                 while (name.length() < 10) {  name += " ";  }
 855                         }
 856                         outputString += name + "\t";
 857
 858                         for(int j=0;j<alignDB.getNumSeqs();j++){
 859
 860                                 if (m->control_pressed) {  return 0;  }
 861
 862                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 863                                 double dist = distCalculator->getDist();
 864
 865                                 outputString += toString(dist) + "\t";
 866                         }
 867
 868                         outputString += "\n";
 869
 870
 871                         if(i % 100 == 0){
 872                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 873                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 874                         }
 875
 876
 877                         //send results to parent
 878                         int length = outputString.length();
 879                         char* buf = new char[length];
 880                         memcpy(buf, outputString.c_str(), length);
 881
 882                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 883                         size += outputString.length();
 884                         outputString = "";
 885                         delete buf;
 886                 }
 887
 888                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 889                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 890                 MPI_File_close(&outMPI);
 891
 892                 return 1;
 893         }
 894         catch(exception& e) {
 895                 m->errorOut(e, "DistanceCommand", "driverMPI");
 896                 exit(1);
 897         }
 898 }
 899 #endif
 900 /**************************************************************************************************
 901 int DistanceCommand::convertMatrix(string outputFile) {
 902         try{
 903
 904                 //sort file by first column so the distances for each row are together
 905                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 906
 907                 //use the unix sort
 908                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 909                         string command = "sort -n " + outputFile + " -o " + outfile;
 910                         system(command.c_str());
 911                 #else //sort using windows sort
 912                         string command = "sort " + outputFile + " /O " + outfile;
 913                         system(command.c_str());
 914                 #endif
 915
 916
 917                 //output to new file distance for each row and save positions in file where new row begins
 918                 ifstream in;
 919                 m->openInputFile(outfile, in);
 920
 921                 ofstream out;
 922                 m->openOutputFile(outputFile, out);
 923
 924                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 925
 926                 out << alignDB.getNumSeqs() << endl;
 927
 928                 //get first currentRow
 929                 string first, currentRow, second;
 930                 float dist;
 931                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 932                 map<string, float>::iterator it;
 933
 934                 in >> first;
 935                 currentRow = first;
 936
 937                 rowDists[first] = 0.00; //distance to yourself is 0.0
 938
 939                 in.seekg(0);
 940                 //m->openInputFile(outfile, in);
 941
 942                 while(!in.eof()) {
 943                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 944
 945                         in >> first >> second >> dist; m->gobble(in);
 946
 947                         if (first != currentRow) {
 948                                 //print out last row
 949                                 out << currentRow << '\t'; //print name
 950
 951                                 //print dists
 952                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 953                                         out << it->second << '\t';
 954                                 }
 955                                 out << endl;
 956
 957                                 //start new row
 958                                 currentRow = first;
 959                                 rowDists.clear();
 960                                 rowDists[first] = 0.00;
 961                                 rowDists[second] = dist;
 962                         }else{
 963                                 rowDists[second] = dist;
 964                         }
 965                 }
 966                 //print out last row
 967                 out << currentRow << '\t'; //print name
 968
 969                 //print dists
 970                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 971                         out << it->second << '\t';
 972                 }
 973                 out << endl;
 974
 975                 in.close();
 976                 out.close();
 977
 978                 remove(outfile.c_str());
 979
 980                 return 1;
 981
 982         }
 983         catch(exception& e) {
 984                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 985                 exit(1);
 986         }
 987 }
 988 /**************************************************************************************************
 989 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 990         try{
 991
 992                 //sort file by first column so the distances for each row are together
 993                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 994
 995                 //use the unix sort
 996                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 997                         string command = "sort -n " + outputFile + " -o " + outfile;
 998                         system(command.c_str());
 999                 #else //sort using windows sort
1000                         string command = "sort " + outputFile + " /O " + outfile;
1001                         system(command.c_str());
1002                 #endif
1003
1004
1005                 //output to new file distance for each row and save positions in file where new row begins
1006                 ifstream in;
1007                 m->openInputFile(outfile, in);
1008
1009                 ofstream out;
1010                 m->openOutputFile(outputFile, out);
1011
1012                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1013
1014                 out << alignDB.getNumSeqs() << endl;
1015
1016                 //get first currentRow
1017                 string first, currentRow, second;
1018                 float dist;
1019                 int i, j;
1020                 i = 0; j = 0;
1021                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
1022                 map<string, float>::iterator it;
1023
1024                 in >> first;
1025                 currentRow = first;
1026
1027                 rowDists[first] = 0.00; //distance to yourself is 0.0
1028
1029                 in.seekg(0);
1030                 //m->openInputFile(outfile, in);
1031
1032                 while(!in.eof()) {
1033                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
1034
1035                         in >> first >> second >> dist; m->gobble(in);
1036
1037                         if (first != currentRow) {
1038                                 //print out last row
1039                                 out << currentRow << '\t'; //print name
1040
1041                                 //print dists
1042                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1043                                         if (j >= i) { break; }
1044                                         out << it->second << '\t';
1045                                         j++;
1046                                 }
1047                                 out << endl;
1048
1049                                 //start new row
1050                                 currentRow = first;
1051                                 rowDists.clear();
1052                                 rowDists[first] = 0.00;
1053                                 rowDists[second] = dist;
1054                                 j = 0;
1055                                 i++;
1056                         }else{
1057                                 rowDists[second] = dist;
1058                         }
1059                 }
1060                 //print out last row
1061                 out << currentRow << '\t'; //print name
1062
1063                 //print dists
1064                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1065                         out << it->second << '\t';
1066                 }
1067                 out << endl;
1068
1069                 in.close();
1070                 out.close();
1071
1072                 remove(outfile.c_str());
1073
1074                 return 1;
1075
1076         }
1077         catch(exception& e) {
1078                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
1079                 exit(1);
1080         }
1081 }
1082 /**************************************************************************************************/
1083 //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,
1084 //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.
1085 //also check to make sure the 2 files have the same alignment length.
1086 bool DistanceCommand::sanityCheck() {
1087         try{
1088                 bool good = true;
1089
1090                 //make sure the 2 fasta files have the same alignment length
1091                 ifstream in;
1092                 m->openInputFile(fastafile, in);
1093                 int fastaAlignLength = 0;
1094                 if (in) {
1095                         Sequence tempIn(in);
1096                         fastaAlignLength = tempIn.getAligned().length();
1097                 }
1098                 in.close();
1099
1100                 ifstream in2;
1101                 m->openInputFile(oldfastafile, in2);
1102                 int oldfastaAlignLength = 0;
1103                 if (in2) {
1104                         Sequence tempIn2(in2);
1105                         oldfastaAlignLength = tempIn2.getAligned().length();
1106                 }
1107                 in2.close();
1108
1109                 if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false;  }
1110
1111                 //read fasta file and save names as well as adding them to the alignDB
1112                 set<string> namesOldFasta;
1113
1114                 ifstream inFasta;
1115                 m->openInputFile(oldfastafile, inFasta);
1116
1117                 while (!inFasta.eof()) {
1118                         if (m->control_pressed) {  inFasta.close(); return good;  }
1119
1120                         Sequence temp(inFasta);
1121
1122                         if (temp.getName() != "") {
1123                                 namesOldFasta.insert(temp.getName());  //save name
1124                                 alignDB.push_back(temp);  //add to DB
1125                         }
1126
1127                         m->gobble(inFasta);
1128                 }
1129
1130                 inFasta.close();
1131
1132                 //read through the column file checking names and removing distances above the cutoff
1133                 ifstream inDist;
1134                 m->openInputFile(column, inDist);
1135
1136                 ofstream outDist;
1137                 string outputFile = column + ".temp";
1138                 m->openOutputFile(outputFile, outDist);
1139
1140                 string name1, name2;
1141                 float dist;
1142                 while (!inDist.eof()) {
1143                         if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
1144
1145                         inDist >> name1 >> name2 >> dist; m->gobble(inDist);
1146
1147                         //both names are in fasta file and distance is below cutoff
1148                         if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) {  good = false; break;  }
1149                         else{
1150                                 if (dist <= cutoff) {
1151                                         outDist << name1 << '\t' << name2 << '\t' << dist << endl;
1152                                 }
1153                         }
1154                 }
1155
1156                 inDist.close();
1157                 outDist.close();
1158
1159                 if (good) {
1160                         remove(column.c_str());
1161                         rename(outputFile.c_str(), column.c_str());
1162                 }else{
1163                         remove(outputFile.c_str()); //temp file is bad because file mismatch above
1164                 }
1165
1166                 return good;
1167
1168         }
1169         catch(exception& e) {
1170                 m->errorOut(e, "DistanceCommand", "sanityCheck");
1171                 exit(1);
1172         }
1173 }
1174 /**************************************************************************************************/
1175
1176
1177
1178