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::setParameters(){
  19         try {
  20                 CommandParameter pcolumn("column", "InputTypes", "", "", "none", "none", "OldFastaColumn",false,false); parameters.push_back(pcolumn);
  21                 CommandParameter poldfasta("oldfasta", "InputTypes", "", "", "none", "none", "OldFastaColumn",false,false); parameters.push_back(poldfasta);
  22                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  23                 CommandParameter poutput("output", "Multiple", "column-lt-square", "column", "", "", "",false,false); parameters.push_back(poutput);
  24                 CommandParameter pcalc("calc", "Multiple", "nogaps-eachgap-onegap", "onegap", "", "", "",false,false); parameters.push_back(pcalc);
  25                 CommandParameter pcountends("countends", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pcountends);
  26                 CommandParameter pcompress("compress", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pcompress);
  27                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  28                 CommandParameter pcutoff("cutoff", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pcutoff);
  29                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  30                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  31
  32                 vector<string> myArray;
  33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "DistanceCommand", "setParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 string DistanceCommand::getHelpString(){
  43         try {
  44                 string helpString = "";
  45                 helpString += "The dist.seqs command reads a file containing sequences and creates a distance file.\n";
  46                 helpString += "The dist.seqs command parameters are fasta, oldfasta, column, calc, countends, output, compress, cutoff and processors.  \n";
  47                 helpString += "The fasta parameter is required, unless you have a valid current fasta file.\n";
  48                 helpString += "The oldfasta and column parameters allow you to append the distances calculated to the column file.\n";
  49                 helpString += "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";
  50                 helpString += "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";
  51                 helpString += "The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n";
  52                 helpString += "The output parameter allows you to specify format of your distance matrix. Options are column, lt, and square. The default is column.\n";
  53                 helpString += "The processors parameter allows you to specify number of processors to use.  The default is 1.\n";
  54                 helpString += "The compress parameter allows you to indicate that you want the resulting distance file compressed.  The default is false.\n";
  55                 helpString += "The dist.seqs command should be in the following format: \n";
  56                 helpString += "dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n";
  57                 helpString += "Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n";
  58                 helpString += "Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n";
  59                 return helpString;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "DistanceCommand", "getHelpString");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67 DistanceCommand::DistanceCommand(){
  68         try {
  69                 abort = true; calledHelp = true;
  70                 setParameters();
  71                 vector<string> tempOutNames;
  72                 outputTypes["phylip"] = tempOutNames;
  73                 outputTypes["column"] = tempOutNames;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "DistanceCommand", "DistanceCommand");
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81 DistanceCommand::DistanceCommand(string option) {
  82         try {
  83                 abort = false; calledHelp = false;
  84                 Estimators.clear();
  85
  86                 //allow user to run help
  87                 if(option == "help") { help(); abort = true; calledHelp = true; }
  88                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  89
  90                 else {
  91                         vector<string> myArray = setParameters();
  92
  93                         OptionParser parser(option);
  94                         map<string, string> parameters = parser.getParameters();
  95
  96                         ValidParameters validParameter("dist.seqs");
  97                         map<string, string>::iterator it2;
  98
  99                         //check to make sure all parameters are valid for command
 100                         for (it2 = parameters.begin(); it2 != parameters.end(); it2++) {
 101                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
 102                         }
 103
 104                         //initialize outputTypes
 105                         vector<string> tempOutNames;
 106                         outputTypes["phylip"] = tempOutNames;
 107                         outputTypes["column"] = tempOutNames;
 108
 109                         //if the user changes the input directory command factory will send this info to us in the output parameter
 110                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 111                         if (inputDir == "not found"){   inputDir = "";          }
 112                         else {
 113                                 string path;
 114                                 it2 = parameters.find("fasta");
 115                                 //user has given a template file
 116                                 if(it2 != parameters.end()){
 117                                         path = m->hasPath(it2->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["fasta"] = inputDir + it2->second;           }
 120                                 }
 121
 122                                 it2 = parameters.find("oldfasta");
 123                                 //user has given a template file
 124                                 if(it2 != parameters.end()){
 125                                         path = m->hasPath(it2->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["oldfasta"] = inputDir + it2->second;                }
 128                                 }
 129
 130                                 it2 = parameters.find("column");
 131                                 //user has given a template file
 132                                 if(it2 != parameters.end()){
 133                                         path = m->hasPath(it2->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                                         if (path == "") {       parameters["column"] = inputDir + it2->second;          }
 136                                 }
 137                         }
 138
 139                         //check for required parameters
 140                         fastafile = validParameter.validFile(parameters, "fasta", true);
 141                         if (fastafile == "not found") {
 142                                 fastafile = m->getFastaFile();
 143                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine();
 144                                         ifstream inFASTA;
 145                                         m->openInputFile(fastafile, inFASTA);
 146                                         alignDB = SequenceDB(inFASTA);
 147                                         inFASTA.close();
 148                                 }else {         m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 149                         }else if (fastafile == "not open") { abort = true; }
 150                         else{
 151                                 ifstream inFASTA;
 152                                 m->openInputFile(fastafile, inFASTA);
 153                                 alignDB = SequenceDB(inFASTA);
 154                                 inFASTA.close();
 155                         }
 156
 157                         oldfastafile = validParameter.validFile(parameters, "oldfasta", true);
 158                         if (oldfastafile == "not found") { oldfastafile = ""; }
 159                         else if (oldfastafile == "not open") { abort = true; }
 160
 161                         column = validParameter.validFile(parameters, "column", true);
 162                         if (column == "not found") { column = ""; }
 163                         else if (column == "not open") { abort = true; }
 164
 165                         //if the user changes the output directory command factory will send this info to us in the output parameter
 166                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 167                                 outputDir = "";
 168                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 169                         }
 170
 171                         //check for optional parameter and set defaults
 172                         // ...at some point should added some additional type checking...
 173                         calc = validParameter.validFile(parameters, "calc", false);
 174                         if (calc == "not found") { calc = "onegap";  }
 175                         else {
 176                                  if (calc == "default")  {  calc = "onegap";  }
 177                         }
 178                         m->splitAtDash(calc, Estimators);
 179
 180                         string temp;
 181                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
 182                         convert(temp, countends);
 183
 184                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
 185                         convert(temp, cutoff);
 186
 187                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 188                         m->setProcessors(temp);
 189                         convert(temp, processors);
 190
 191                         temp = validParameter.validFile(parameters, "compress", false);         if(temp == "not found"){  temp = "F"; }
 192                         convert(temp, compress);
 193
 194                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "column"; }
 195
 196                         if (((column != "") && (oldfastafile == "")) || ((column == "") && (oldfastafile != ""))) { m->mothurOut("If you provide column or oldfasta, you must provide both."); m->mothurOutEndLine(); abort=true; }
 197
 198                         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; }
 199
 200                         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"; }
 201
 202                         ValidCalculators validCalculator;
 203
 204                         if (m->isTrue(countends) == true) {
 205                                 for (int i=0; i<Estimators.size(); i++) {
 206                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 207                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
 208                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
 209                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
 210                                         }
 211                                 }
 212                         }else {
 213                                 for (int i=0; i<Estimators.size(); i++) {
 214                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 215                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
 216                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
 217                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
 218                                         }
 219                                 }
 220                         }
 221
 222                 }
 223
 224         }
 225         catch(exception& e) {
 226                 m->errorOut(e, "DistanceCommand", "DistanceCommand");
 227                 exit(1);
 228         }
 229 }
 230 //**********************************************************************************************************************
 231
 232 int DistanceCommand::execute(){
 233         try {
 234
 235                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 236
 237                 int startTime = time(NULL);
 238
 239                 //save number of new sequence
 240                 numNewFasta = alignDB.getNumSeqs();
 241
 242                 //sanity check the oldfasta and column file as well as add oldfasta sequences to alignDB
 243                 if ((oldfastafile != "") && (column != ""))  {  if (!(sanityCheck())) { return 0; }  }
 244
 245                 if (m->control_pressed) { return 0; }
 246
 247                 int numSeqs = alignDB.getNumSeqs();
 248                 cutoff += 0.005;
 249
 250                 string outputFile;
 251
 252                 if (output == "lt") { //does the user want lower triangle phylip formatted file
 253                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
 254                         remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
 255
 256                         //output numSeqs to phylip formatted dist file
 257                 }else if (output == "column") { //user wants column format
 258                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
 259                         outputTypes["column"].push_back(outputFile);
 260
 261                         //so we don't accidentally overwrite
 262                         if (outputFile == column) {
 263                                 string tempcolumn = column + ".old";
 264                                 rename(column.c_str(), tempcolumn.c_str());
 265                         }
 266
 267                         remove(outputFile.c_str());
 268                 }else { //assume square
 269                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
 270                         remove(outputFile.c_str());
 271                         outputTypes["phylip"].push_back(outputFile);
 272                 }
 273
 274
 275 #ifdef USE_MPI
 276
 277                 int pid, start, end;
 278                 int tag = 2001;
 279
 280                 MPI_Status status;
 281                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 282                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 283
 284                 //each process gets where it should start and stop in the file
 285                 if (output != "square") {
 286                         start = int (sqrt(float(pid)/float(processors)) * numSeqs);
 287                         end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
 288                 }else{
 289                         start = int ((float(pid)/float(processors)) * numSeqs);
 290                         end = int ((float(pid+1)/float(processors)) * numSeqs);
 291                 }
 292
 293                 if (output == "column") {
 294                         MPI_File outMPI;
 295                         int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 296
 297                         //char* filename = new char[outputFile.length()];
 298                         //memcpy(filename, outputFile.c_str(), outputFile.length());
 299
 300                         char filename[1024];
 301                         strcpy(filename, outputFile.c_str());
 302
 303                         MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
 304                         //delete filename;
 305
 306                         if (pid == 0) { //you are the root process
 307
 308                                 //do your part
 309                                 string outputMyPart;
 310
 311                                 driverMPI(start, end, outMPI, cutoff);
 312
 313                                 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator;  return 0; }
 314
 315                                 //wait on chidren
 316                                 for(int i = 1; i < processors; i++) {
 317                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 318
 319                                         char buf[5];
 320                                         MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 321                                 }
 322                         }else { //you are a child process
 323                                 //do your part
 324                                 driverMPI(start, end, outMPI, cutoff);
 325
 326                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 327
 328                                 char buf[5];
 329                                 strcpy(buf, "done");
 330                                 //tell parent you are done.
 331                                 MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 332                         }
 333
 334                         MPI_File_close(&outMPI);
 335
 336                 }else { //lower triangle format
 337                         if (pid == 0) { //you are the root process
 338
 339                                 //do your part
 340                                 string outputMyPart;
 341                                 unsigned long int mySize;
 342
 343                                 if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
 344                                 else { driverMPI(start, end, outputFile, mySize, output); }
 345
 346                                 if (m->control_pressed) {  outputTypes.clear();  delete distCalculator;  return 0; }
 347
 348                                 int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 349                                 MPI_File outMPI;
 350                                 MPI_File inMPI;
 351
 352                                 //char* filename = new char[outputFile.length()];
 353                                 //memcpy(filename, outputFile.c_str(), outputFile.length());
 354
 355                                 char filename[1024];
 356                                 strcpy(filename, outputFile.c_str());
 357
 358                                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 359                                 //delete filename;
 360
 361                                 //wait on chidren
 362                                 for(int b = 1; b < processors; b++) {
 363                                         unsigned long int fileSize;
 364
 365                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 366
 367                                         MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
 368
 369                                         string outTemp = outputFile + toString(b) + ".temp";
 370
 371                                         char* buf = new char[outTemp.length()];
 372                                         memcpy(buf, outTemp.c_str(), outTemp.length());
 373
 374                                         MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 375                                         delete buf;
 376
 377                                         int count = 0;
 378                                         while (count < fileSize) {
 379                                                 char buf2[1];
 380                                                 MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
 381                                                 MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
 382                                                 count += 1;
 383                                         }
 384
 385                                         MPI_File_close(&inMPI); //deleted on close
 386                                 }
 387
 388                                 MPI_File_close(&outMPI);
 389                         }else { //you are a child process
 390                                 //do your part
 391                                 unsigned long int size;
 392                                 if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
 393                                 else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
 394
 395                                 if (m->control_pressed) { delete distCalculator;  return 0; }
 396
 397                                 //tell parent you are done.
 398                                 MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
 399                         }
 400                 }
 401                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 402 #else
 403
 404         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 405                 //if you don't need to fork anything
 406                 if(processors == 1){
 407                         if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 408                         else { driver(0, numSeqs, outputFile, "square");  }
 409                 }else{ //you have multiple processors
 410
 411                         unsigned long int numDists = 0;
 412
 413                         if (output == "square") {
 414                                  numDists = numSeqs * numSeqs;
 415                         }else {
 416                                 for(int i=0;i<numSeqs;i++){
 417                                         for(int j=0;j<i;j++){
 418                                                 numDists++;
 419                                                 if (numDists > processors) { break; }
 420                                         }
 421                                 }
 422                         }
 423
 424                         if (numDists < processors) { processors = numDists; }
 425
 426                         for (int i = 0; i < processors; i++) {
 427                                 distlinePair tempLine;
 428                                 lines.push_back(tempLine);
 429                                 if (output != "square") {
 430                                         lines[i].start = int (sqrt(float(i)/float(processors)) * numSeqs);
 431                                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 432                                 }else{
 433                                         lines[i].start = int ((float(i)/float(processors)) * numSeqs);
 434                                         lines[i].end = int ((float(i+1)/float(processors)) * numSeqs);
 435                                 }
 436
 437                         }
 438
 439                         createProcesses(outputFile);
 440                 }
 441         #else
 442                 //ifstream inFASTA;
 443                 if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 444                 else { driver(0, numSeqs, outputFile, "square");  }
 445         #endif
 446
 447 #endif
 448                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 449
 450                 #ifdef USE_MPI
 451                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 452
 453                         if (pid == 0) { //only one process should output to screen
 454                 #endif
 455
 456                 //if (output == "square") {  convertMatrix(outputFile); }
 457
 458                 ifstream fileHandle;
 459                 fileHandle.open(outputFile.c_str());
 460                 if(fileHandle) {
 461                         m->gobble(fileHandle);
 462                         if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff.");  m->mothurOutEndLine(); }
 463                 }
 464
 465                 //append the old column file to the new one
 466                 if ((oldfastafile != "") && (column != ""))  {
 467                         //we had to rename the column file so we didnt overwrite above, but we want to keep old name
 468                         if (outputFile == column) {
 469                                 string tempcolumn = column + ".old";
 470                                 m->appendFiles(tempcolumn, outputFile);
 471                                 remove(tempcolumn.c_str());
 472                         }else{
 473                                 m->appendFiles(outputFile, column);
 474                                 remove(outputFile.c_str());
 475                                 outputFile = column;
 476                         }
 477
 478                         if (outputDir != "") {
 479                                 string newOutputName = outputDir + m->getSimpleName(outputFile);
 480                                 rename(outputFile.c_str(), newOutputName.c_str());
 481                                 remove(outputFile.c_str());
 482                                 outputFile = newOutputName;
 483                         }
 484                 }
 485
 486
 487                 #ifdef USE_MPI
 488                         }
 489                 #endif
 490
 491                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 492
 493                 delete distCalculator;
 494
 495                 //set phylip file as new current phylipfile
 496                 string current = "";
 497                 itTypes = outputTypes.find("phylip");
 498                 if (itTypes != outputTypes.end()) {
 499                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
 500                 }
 501
 502                 //set column file as new current columnfile
 503                 itTypes = outputTypes.find("column");
 504                 if (itTypes != outputTypes.end()) {
 505                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setColumnFile(current); }
 506                 }
 507
 508                 m->mothurOutEndLine();
 509                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 510                 m->mothurOut(outputFile); m->mothurOutEndLine();
 511                 m->mothurOutEndLine();
 512                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 513
 514
 515                 if (m->isTrue(compress)) {
 516                         m->mothurOut("Compressing..."); m->mothurOutEndLine();
 517                         m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 518                         system(("gzip -v " + outputFile).c_str());
 519                         outputNames.push_back(outputFile + ".gz");
 520                 }else { outputNames.push_back(outputFile); }
 521
 522                 return 0;
 523
 524         }
 525         catch(exception& e) {
 526                 m->errorOut(e, "DistanceCommand", "execute");
 527                 exit(1);
 528         }
 529 }
 530 /**************************************************************************************************/
 531 void DistanceCommand::createProcesses(string filename) {
 532         try {
 533 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 534                 int process = 1;
 535                 processIDS.clear();
 536
 537                 //loop through and create all the processes you want
 538                 while (process != processors) {
 539                         int pid = fork();
 540
 541                         if (pid > 0) {
 542                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 543                                 process++;
 544                         }else if (pid == 0){
 545                                 if (output != "square") {  driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", cutoff); }
 546                                 else { driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", "square"); }
 547                                 exit(0);
 548                         }else {
 549                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
 550                                 perror(" : ");
 551                                 for (int i=0;i<processIDS.size();i++) {  int temp = processIDS[i]; kill (temp, SIGINT); }
 552                                 exit(0);
 553                         }
 554                 }
 555
 556                 //parent does its part
 557                 if (output != "square") {  driver(lines[0].start, lines[0].end, filename, cutoff); }
 558                 else { driver(lines[0].start, lines[0].end, filename, "square"); }
 559
 560
 561                 //force parent to wait until all the processes are done
 562                 for (int i=0;i<processIDS.size();i++) {
 563                         int temp = processIDS[i];
 564                         wait(&temp);
 565                 }
 566
 567                 //append and remove temp files
 568                 for (int i=0;i<processIDS.size();i++) {
 569                         m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
 570                         remove((filename + toString(processIDS[i]) + ".temp").c_str());
 571                 }
 572 #endif
 573         }
 574         catch(exception& e) {
 575                 m->errorOut(e, "DistanceCommand", "createProcesses");
 576                 exit(1);
 577         }
 578 }
 579
 580 /**************************************************************************************************/
 581 /////// need to fix to work with calcs and sequencedb
 582 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 583         try {
 584
 585                 int startTime = time(NULL);
 586
 587                 //column file
 588                 ofstream outFile(dFileName.c_str(), ios::trunc);
 589                 outFile.setf(ios::fixed, ios::showpoint);
 590                 outFile << setprecision(4);
 591
 592                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 593
 594                 for(int i=startLine;i<endLine;i++){
 595                         if(output == "lt")      {
 596                                 string name = alignDB.get(i).getName();
 597                                 if (name.length() < 10) { //pad with spaces to make compatible
 598                                         while (name.length() < 10) {  name += " ";  }
 599                                 }
 600                                 outFile << name << '\t';
 601                         }
 602                         for(int j=0;j<i;j++){
 603
 604                                 if (m->control_pressed) { outFile.close(); return 0;  }
 605
 606                                 //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
 607                                 //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
 608                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 609
 610                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 611                                 double dist = distCalculator->getDist();
 612
 613                                 if(dist <= cutoff){
 614                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 615                                 }
 616                                 if (output == "lt") {  outFile << dist << '\t'; }
 617                         }
 618
 619                         if (output == "lt") { outFile << endl; }
 620
 621                         if(i % 100 == 0){
 622                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 623                         }
 624
 625                 }
 626                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 627
 628                 outFile.close();
 629
 630                 return 1;
 631         }
 632         catch(exception& e) {
 633                 m->errorOut(e, "DistanceCommand", "driver");
 634                 exit(1);
 635         }
 636 }
 637 /**************************************************************************************************/
 638 /////// need to fix to work with calcs and sequencedb
 639 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 640         try {
 641
 642                 int startTime = time(NULL);
 643
 644                 //column file
 645                 ofstream outFile(dFileName.c_str(), ios::trunc);
 646                 outFile.setf(ios::fixed, ios::showpoint);
 647                 outFile << setprecision(4);
 648
 649                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 650
 651                 for(int i=startLine;i<endLine;i++){
 652
 653                         string name = alignDB.get(i).getName();
 654                         //pad with spaces to make compatible
 655                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 656
 657                         outFile << name << '\t';
 658
 659                         for(int j=0;j<alignDB.getNumSeqs();j++){
 660
 661                                 if (m->control_pressed) { outFile.close(); return 0;  }
 662
 663                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 664                                 double dist = distCalculator->getDist();
 665
 666                                 outFile << dist << '\t';
 667                         }
 668
 669                         outFile << endl;
 670
 671                         if(i % 100 == 0){
 672                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 673                         }
 674
 675                 }
 676                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 677
 678                 outFile.close();
 679
 680                 return 1;
 681         }
 682         catch(exception& e) {
 683                 m->errorOut(e, "DistanceCommand", "driver");
 684                 exit(1);
 685         }
 686 }
 687 #ifdef USE_MPI
 688 /**************************************************************************************************/
 689 /////// need to fix to work with calcs and sequencedb
 690 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 691         try {
 692                 MPI_Status status;
 693                 int startTime = time(NULL);
 694
 695                 string outputString = "";
 696
 697                 for(int i=startLine;i<endLine;i++){
 698
 699                         for(int j=0;j<i;j++){
 700
 701                                 if (m->control_pressed) {  return 0;  }
 702
 703                                 //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
 704                                 //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
 705                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 706
 707                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 708                                 double dist = distCalculator->getDist();
 709
 710                                 if(dist <= cutoff){
 711                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 712                                 }
 713                         }
 714
 715                         if(i % 100 == 0){
 716                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 717                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 718                         }
 719
 720
 721                         //send results to parent
 722                         int length = outputString.length();
 723
 724                         char* buf = new char[length];
 725                         memcpy(buf, outputString.c_str(), length);
 726
 727                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 728                         outputString = "";
 729                         delete buf;
 730
 731                 }
 732
 733                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 734                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 735                 return 1;
 736         }
 737         catch(exception& e) {
 738                 m->errorOut(e, "DistanceCommand", "driverMPI");
 739                 exit(1);
 740         }
 741 }
 742 /**************************************************************************************************/
 743 /////// need to fix to work with calcs and sequencedb
 744 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 745         try {
 746                 MPI_Status status;
 747
 748                 MPI_File outMPI;
 749                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 750
 751                 //char* filename = new char[file.length()];
 752                 //memcpy(filename, file.c_str(), file.length());
 753
 754                 char filename[1024];
 755                 strcpy(filename, file.c_str());
 756
 757                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 758                 //delete filename;
 759
 760                 int startTime = time(NULL);
 761
 762                 string outputString = "";
 763                 size = 0;
 764
 765                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 766
 767                 for(int i=startLine;i<endLine;i++){
 768
 769                         string name = alignDB.get(i).getName();
 770                         if (name.length() < 10) { //pad with spaces to make compatible
 771                                 while (name.length() < 10) {  name += " ";  }
 772                         }
 773                         outputString += name + "\t";
 774
 775                         for(int j=0;j<i;j++){
 776
 777                                 if (m->control_pressed) {  return 0;  }
 778
 779                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 780                                 double dist = distCalculator->getDist();
 781
 782                                 outputString += toString(dist) + "\t";
 783                         }
 784
 785                         outputString += "\n";
 786
 787
 788                         if(i % 100 == 0){
 789                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 790                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 791                         }
 792
 793
 794                         //send results to parent
 795                         int length = outputString.length();
 796                         char* buf = new char[length];
 797                         memcpy(buf, outputString.c_str(), length);
 798
 799                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 800                         size += outputString.length();
 801                         outputString = "";
 802                         delete buf;
 803                 }
 804
 805                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 806                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 807                 MPI_File_close(&outMPI);
 808
 809                 return 1;
 810         }
 811         catch(exception& e) {
 812                 m->errorOut(e, "DistanceCommand", "driverMPI");
 813                 exit(1);
 814         }
 815 }
 816 /**************************************************************************************************/
 817 /////// need to fix to work with calcs and sequencedb
 818 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 819         try {
 820                 MPI_Status status;
 821
 822                 MPI_File outMPI;
 823                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 824
 825                 //char* filename = new char[file.length()];
 826                 //memcpy(filename, file.c_str(), file.length());
 827
 828                 char filename[1024];
 829                 strcpy(filename, file.c_str());
 830
 831                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 832                 //delete filename;
 833
 834                 int startTime = time(NULL);
 835
 836                 string outputString = "";
 837                 size = 0;
 838
 839                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 840
 841                 for(int i=startLine;i<endLine;i++){
 842
 843                         string name = alignDB.get(i).getName();
 844                         if (name.length() < 10) { //pad with spaces to make compatible
 845                                 while (name.length() < 10) {  name += " ";  }
 846                         }
 847                         outputString += name + "\t";
 848
 849                         for(int j=0;j<alignDB.getNumSeqs();j++){
 850
 851                                 if (m->control_pressed) {  return 0;  }
 852
 853                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 854                                 double dist = distCalculator->getDist();
 855
 856                                 outputString += toString(dist) + "\t";
 857                         }
 858
 859                         outputString += "\n";
 860
 861
 862                         if(i % 100 == 0){
 863                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 864                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 865                         }
 866
 867
 868                         //send results to parent
 869                         int length = outputString.length();
 870                         char* buf = new char[length];
 871                         memcpy(buf, outputString.c_str(), length);
 872
 873                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 874                         size += outputString.length();
 875                         outputString = "";
 876                         delete buf;
 877                 }
 878
 879                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 880                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 881                 MPI_File_close(&outMPI);
 882
 883                 return 1;
 884         }
 885         catch(exception& e) {
 886                 m->errorOut(e, "DistanceCommand", "driverMPI");
 887                 exit(1);
 888         }
 889 }
 890 #endif
 891 /**************************************************************************************************
 892 int DistanceCommand::convertMatrix(string outputFile) {
 893         try{
 894
 895                 //sort file by first column so the distances for each row are together
 896                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 897
 898                 //use the unix sort
 899                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 900                         string command = "sort -n " + outputFile + " -o " + outfile;
 901                         system(command.c_str());
 902                 #else //sort using windows sort
 903                         string command = "sort " + outputFile + " /O " + outfile;
 904                         system(command.c_str());
 905                 #endif
 906
 907
 908                 //output to new file distance for each row and save positions in file where new row begins
 909                 ifstream in;
 910                 m->openInputFile(outfile, in);
 911
 912                 ofstream out;
 913                 m->openOutputFile(outputFile, out);
 914
 915                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 916
 917                 out << alignDB.getNumSeqs() << endl;
 918
 919                 //get first currentRow
 920                 string first, currentRow, second;
 921                 float dist;
 922                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 923                 map<string, float>::iterator it;
 924
 925                 in >> first;
 926                 currentRow = first;
 927
 928                 rowDists[first] = 0.00; //distance to yourself is 0.0
 929
 930                 in.seekg(0);
 931                 //m->openInputFile(outfile, in);
 932
 933                 while(!in.eof()) {
 934                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 935
 936                         in >> first >> second >> dist; m->gobble(in);
 937
 938                         if (first != currentRow) {
 939                                 //print out last row
 940                                 out << currentRow << '\t'; //print name
 941
 942                                 //print dists
 943                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 944                                         out << it->second << '\t';
 945                                 }
 946                                 out << endl;
 947
 948                                 //start new row
 949                                 currentRow = first;
 950                                 rowDists.clear();
 951                                 rowDists[first] = 0.00;
 952                                 rowDists[second] = dist;
 953                         }else{
 954                                 rowDists[second] = dist;
 955                         }
 956                 }
 957                 //print out last row
 958                 out << currentRow << '\t'; //print name
 959
 960                 //print dists
 961                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 962                         out << it->second << '\t';
 963                 }
 964                 out << endl;
 965
 966                 in.close();
 967                 out.close();
 968
 969                 remove(outfile.c_str());
 970
 971                 return 1;
 972
 973         }
 974         catch(exception& e) {
 975                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 976                 exit(1);
 977         }
 978 }
 979 /**************************************************************************************************
 980 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 981         try{
 982
 983                 //sort file by first column so the distances for each row are together
 984                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 985
 986                 //use the unix sort
 987                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 988                         string command = "sort -n " + outputFile + " -o " + outfile;
 989                         system(command.c_str());
 990                 #else //sort using windows sort
 991                         string command = "sort " + outputFile + " /O " + outfile;
 992                         system(command.c_str());
 993                 #endif
 994
 995
 996                 //output to new file distance for each row and save positions in file where new row begins
 997                 ifstream in;
 998                 m->openInputFile(outfile, in);
 999
1000                 ofstream out;
1001                 m->openOutputFile(outputFile, out);
1002
1003                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1004
1005                 out << alignDB.getNumSeqs() << endl;
1006
1007                 //get first currentRow
1008                 string first, currentRow, second;
1009                 float dist;
1010                 int i, j;
1011                 i = 0; j = 0;
1012                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
1013                 map<string, float>::iterator it;
1014
1015                 in >> first;
1016                 currentRow = first;
1017
1018                 rowDists[first] = 0.00; //distance to yourself is 0.0
1019
1020                 in.seekg(0);
1021                 //m->openInputFile(outfile, in);
1022
1023                 while(!in.eof()) {
1024                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
1025
1026                         in >> first >> second >> dist; m->gobble(in);
1027
1028                         if (first != currentRow) {
1029                                 //print out last row
1030                                 out << currentRow << '\t'; //print name
1031
1032                                 //print dists
1033                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1034                                         if (j >= i) { break; }
1035                                         out << it->second << '\t';
1036                                         j++;
1037                                 }
1038                                 out << endl;
1039
1040                                 //start new row
1041                                 currentRow = first;
1042                                 rowDists.clear();
1043                                 rowDists[first] = 0.00;
1044                                 rowDists[second] = dist;
1045                                 j = 0;
1046                                 i++;
1047                         }else{
1048                                 rowDists[second] = dist;
1049                         }
1050                 }
1051                 //print out last row
1052                 out << currentRow << '\t'; //print name
1053
1054                 //print dists
1055                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1056                         out << it->second << '\t';
1057                 }
1058                 out << endl;
1059
1060                 in.close();
1061                 out.close();
1062
1063                 remove(outfile.c_str());
1064
1065                 return 1;
1066
1067         }
1068         catch(exception& e) {
1069                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
1070                 exit(1);
1071         }
1072 }
1073 /**************************************************************************************************/
1074 //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,
1075 //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.
1076 //also check to make sure the 2 files have the same alignment length.
1077 bool DistanceCommand::sanityCheck() {
1078         try{
1079                 bool good = true;
1080
1081                 //make sure the 2 fasta files have the same alignment length
1082                 ifstream in;
1083                 m->openInputFile(fastafile, in);
1084                 int fastaAlignLength = 0;
1085                 if (in) {
1086                         Sequence tempIn(in);
1087                         fastaAlignLength = tempIn.getAligned().length();
1088                 }
1089                 in.close();
1090
1091                 ifstream in2;
1092                 m->openInputFile(oldfastafile, in2);
1093                 int oldfastaAlignLength = 0;
1094                 if (in2) {
1095                         Sequence tempIn2(in2);
1096                         oldfastaAlignLength = tempIn2.getAligned().length();
1097                 }
1098                 in2.close();
1099
1100                 if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false;  }
1101
1102                 //read fasta file and save names as well as adding them to the alignDB
1103                 set<string> namesOldFasta;
1104
1105                 ifstream inFasta;
1106                 m->openInputFile(oldfastafile, inFasta);
1107
1108                 while (!inFasta.eof()) {
1109                         if (m->control_pressed) {  inFasta.close(); return good;  }
1110
1111                         Sequence temp(inFasta);
1112
1113                         if (temp.getName() != "") {
1114                                 namesOldFasta.insert(temp.getName());  //save name
1115                                 alignDB.push_back(temp);  //add to DB
1116                         }
1117
1118                         m->gobble(inFasta);
1119                 }
1120
1121                 inFasta.close();
1122
1123                 //read through the column file checking names and removing distances above the cutoff
1124                 ifstream inDist;
1125                 m->openInputFile(column, inDist);
1126
1127                 ofstream outDist;
1128                 string outputFile = column + ".temp";
1129                 m->openOutputFile(outputFile, outDist);
1130
1131                 string name1, name2;
1132                 float dist;
1133                 while (!inDist.eof()) {
1134                         if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
1135
1136                         inDist >> name1 >> name2 >> dist; m->gobble(inDist);
1137
1138                         //both names are in fasta file and distance is below cutoff
1139                         if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) {  good = false; break;  }
1140                         else{
1141                                 if (dist <= cutoff) {
1142                                         outDist << name1 << '\t' << name2 << '\t' << dist << endl;
1143                                 }
1144                         }
1145                 }
1146
1147                 inDist.close();
1148                 outDist.close();
1149
1150                 if (good) {
1151                         remove(column.c_str());
1152                         rename(outputFile.c_str(), column.c_str());
1153                 }else{
1154                         remove(outputFile.c_str()); //temp file is bad because file mismatch above
1155                 }
1156
1157                 return good;
1158
1159         }
1160         catch(exception& e) {
1161                 m->errorOut(e, "DistanceCommand", "sanityCheck");
1162                 exit(1);
1163         }
1164 }
1165 /**************************************************************************************************/
1166
1167
1168
1169