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                 if (!alignDB.sameLength()) {  m->mothurOut("[ERROR]: your sequences are not the same length, aborting."); m->mothurOutEndLine(); return 0; }
 251
 252                 string outputFile;
 253
 254                 if (output == "lt") { //does the user want lower triangle phylip formatted file
 255                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
 256                         remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
 257
 258                         //output numSeqs to phylip formatted dist file
 259                 }else if (output == "column") { //user wants column format
 260                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
 261                         outputTypes["column"].push_back(outputFile);
 262
 263                         //so we don't accidentally overwrite
 264                         if (outputFile == column) {
 265                                 string tempcolumn = column + ".old";
 266                                 rename(column.c_str(), tempcolumn.c_str());
 267                         }
 268
 269                         remove(outputFile.c_str());
 270                 }else { //assume square
 271                         outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
 272                         remove(outputFile.c_str());
 273                         outputTypes["phylip"].push_back(outputFile);
 274                 }
 275
 276
 277 #ifdef USE_MPI
 278
 279                 int pid, start, end;
 280                 int tag = 2001;
 281
 282                 MPI_Status status;
 283                 MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
 284                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 285
 286                 //each process gets where it should start and stop in the file
 287                 if (output != "square") {
 288                         start = int (sqrt(float(pid)/float(processors)) * numSeqs);
 289                         end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
 290                 }else{
 291                         start = int ((float(pid)/float(processors)) * numSeqs);
 292                         end = int ((float(pid+1)/float(processors)) * numSeqs);
 293                 }
 294
 295                 if (output == "column") {
 296                         MPI_File outMPI;
 297                         int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 298
 299                         //char* filename = new char[outputFile.length()];
 300                         //memcpy(filename, outputFile.c_str(), outputFile.length());
 301
 302                         char filename[1024];
 303                         strcpy(filename, outputFile.c_str());
 304
 305                         MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
 306                         //delete filename;
 307
 308                         if (pid == 0) { //you are the root process
 309
 310                                 //do your part
 311                                 string outputMyPart;
 312
 313                                 driverMPI(start, end, outMPI, cutoff);
 314
 315                                 if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator;  return 0; }
 316
 317                                 //wait on chidren
 318                                 for(int i = 1; i < processors; i++) {
 319                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 320
 321                                         char buf[5];
 322                                         MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
 323                                 }
 324                         }else { //you are a child process
 325                                 //do your part
 326                                 driverMPI(start, end, outMPI, cutoff);
 327
 328                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 329
 330                                 char buf[5];
 331                                 strcpy(buf, "done");
 332                                 //tell parent you are done.
 333                                 MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
 334                         }
 335
 336                         MPI_File_close(&outMPI);
 337
 338                 }else { //lower triangle format
 339                         if (pid == 0) { //you are the root process
 340
 341                                 //do your part
 342                                 string outputMyPart;
 343                                 unsigned long int mySize;
 344
 345                                 if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
 346                                 else { driverMPI(start, end, outputFile, mySize, output); }
 347
 348                                 if (m->control_pressed) {  outputTypes.clear();  delete distCalculator;  return 0; }
 349
 350                                 int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 351                                 MPI_File outMPI;
 352                                 MPI_File inMPI;
 353
 354                                 //char* filename = new char[outputFile.length()];
 355                                 //memcpy(filename, outputFile.c_str(), outputFile.length());
 356
 357                                 char filename[1024];
 358                                 strcpy(filename, outputFile.c_str());
 359
 360                                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 361                                 //delete filename;
 362
 363                                 //wait on chidren
 364                                 for(int b = 1; b < processors; b++) {
 365                                         unsigned long int fileSize;
 366
 367                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
 368
 369                                         MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
 370
 371                                         string outTemp = outputFile + toString(b) + ".temp";
 372
 373                                         char* buf = new char[outTemp.length()];
 374                                         memcpy(buf, outTemp.c_str(), outTemp.length());
 375
 376                                         MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
 377                                         delete buf;
 378
 379                                         int count = 0;
 380                                         while (count < fileSize) {
 381                                                 char buf2[1];
 382                                                 MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
 383                                                 MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
 384                                                 count += 1;
 385                                         }
 386
 387                                         MPI_File_close(&inMPI); //deleted on close
 388                                 }
 389
 390                                 MPI_File_close(&outMPI);
 391                         }else { //you are a child process
 392                                 //do your part
 393                                 unsigned long int size;
 394                                 if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
 395                                 else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
 396
 397                                 if (m->control_pressed) { delete distCalculator;  return 0; }
 398
 399                                 //tell parent you are done.
 400                                 MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
 401                         }
 402                 }
 403                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 404 #else
 405
 406         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 407                 //if you don't need to fork anything
 408                 if(processors == 1){
 409                         if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 410                         else { driver(0, numSeqs, outputFile, "square");  }
 411                 }else{ //you have multiple processors
 412
 413                         unsigned long int numDists = 0;
 414
 415                         if (output == "square") {
 416                                  numDists = numSeqs * numSeqs;
 417                         }else {
 418                                 for(int i=0;i<numSeqs;i++){
 419                                         for(int j=0;j<i;j++){
 420                                                 numDists++;
 421                                                 if (numDists > processors) { break; }
 422                                         }
 423                                 }
 424                         }
 425
 426                         if (numDists < processors) { processors = numDists; }
 427
 428                         for (int i = 0; i < processors; i++) {
 429                                 distlinePair tempLine;
 430                                 lines.push_back(tempLine);
 431                                 if (output != "square") {
 432                                         lines[i].start = int (sqrt(float(i)/float(processors)) * numSeqs);
 433                                         lines[i].end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 434                                 }else{
 435                                         lines[i].start = int ((float(i)/float(processors)) * numSeqs);
 436                                         lines[i].end = int ((float(i+1)/float(processors)) * numSeqs);
 437                                 }
 438
 439                         }
 440
 441                         createProcesses(outputFile);
 442                 }
 443         #else
 444                 //ifstream inFASTA;
 445                 if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 446                 else { driver(0, numSeqs, outputFile, "square");  }
 447         #endif
 448
 449 #endif
 450                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 451
 452                 #ifdef USE_MPI
 453                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 454
 455                         if (pid == 0) { //only one process should output to screen
 456                 #endif
 457
 458                 //if (output == "square") {  convertMatrix(outputFile); }
 459
 460                 ifstream fileHandle;
 461                 fileHandle.open(outputFile.c_str());
 462                 if(fileHandle) {
 463                         m->gobble(fileHandle);
 464                         if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff.");  m->mothurOutEndLine(); }
 465                 }
 466
 467                 //append the old column file to the new one
 468                 if ((oldfastafile != "") && (column != ""))  {
 469                         //we had to rename the column file so we didnt overwrite above, but we want to keep old name
 470                         if (outputFile == column) {
 471                                 string tempcolumn = column + ".old";
 472                                 m->appendFiles(tempcolumn, outputFile);
 473                                 remove(tempcolumn.c_str());
 474                         }else{
 475                                 m->appendFiles(outputFile, column);
 476                                 remove(outputFile.c_str());
 477                                 outputFile = column;
 478                         }
 479
 480                         if (outputDir != "") {
 481                                 string newOutputName = outputDir + m->getSimpleName(outputFile);
 482                                 rename(outputFile.c_str(), newOutputName.c_str());
 483                                 remove(outputFile.c_str());
 484                                 outputFile = newOutputName;
 485                         }
 486                 }
 487
 488
 489                 #ifdef USE_MPI
 490                         }
 491                 #endif
 492
 493                 if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
 494
 495                 delete distCalculator;
 496
 497                 //set phylip file as new current phylipfile
 498                 string current = "";
 499                 itTypes = outputTypes.find("phylip");
 500                 if (itTypes != outputTypes.end()) {
 501                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setPhylipFile(current); }
 502                 }
 503
 504                 //set column file as new current columnfile
 505                 itTypes = outputTypes.find("column");
 506                 if (itTypes != outputTypes.end()) {
 507                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setColumnFile(current); }
 508                 }
 509
 510                 m->mothurOutEndLine();
 511                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 512                 m->mothurOut(outputFile); m->mothurOutEndLine();
 513                 m->mothurOutEndLine();
 514                 m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 515
 516
 517                 if (m->isTrue(compress)) {
 518                         m->mothurOut("Compressing..."); m->mothurOutEndLine();
 519                         m->mothurOut("(Replacing " + outputFile + " with " + outputFile + ".gz)"); m->mothurOutEndLine();
 520                         system(("gzip -v " + outputFile).c_str());
 521                         outputNames.push_back(outputFile + ".gz");
 522                 }else { outputNames.push_back(outputFile); }
 523
 524                 return 0;
 525
 526         }
 527         catch(exception& e) {
 528                 m->errorOut(e, "DistanceCommand", "execute");
 529                 exit(1);
 530         }
 531 }
 532 /**************************************************************************************************/
 533 void DistanceCommand::createProcesses(string filename) {
 534         try {
 535 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 536                 int process = 1;
 537                 processIDS.clear();
 538
 539                 //loop through and create all the processes you want
 540                 while (process != processors) {
 541                         int pid = fork();
 542
 543                         if (pid > 0) {
 544                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 545                                 process++;
 546                         }else if (pid == 0){
 547                                 if (output != "square") {  driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", cutoff); }
 548                                 else { driver(lines[process].start, lines[process].end, filename + toString(getpid()) + ".temp", "square"); }
 549                                 exit(0);
 550                         }else {
 551                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes. Error code: " + toString(pid)); m->mothurOutEndLine();
 552                                 perror(" : ");
 553                                 for (int i=0;i<processIDS.size();i++) {  int temp = processIDS[i]; kill (temp, SIGINT); }
 554                                 exit(0);
 555                         }
 556                 }
 557
 558                 //parent does its part
 559                 if (output != "square") {  driver(lines[0].start, lines[0].end, filename, cutoff); }
 560                 else { driver(lines[0].start, lines[0].end, filename, "square"); }
 561
 562
 563                 //force parent to wait until all the processes are done
 564                 for (int i=0;i<processIDS.size();i++) {
 565                         int temp = processIDS[i];
 566                         wait(&temp);
 567                 }
 568
 569                 //append and remove temp files
 570                 for (int i=0;i<processIDS.size();i++) {
 571                         m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
 572                         remove((filename + toString(processIDS[i]) + ".temp").c_str());
 573                 }
 574 #endif
 575         }
 576         catch(exception& e) {
 577                 m->errorOut(e, "DistanceCommand", "createProcesses");
 578                 exit(1);
 579         }
 580 }
 581
 582 /**************************************************************************************************/
 583 /////// need to fix to work with calcs and sequencedb
 584 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 585         try {
 586
 587                 int startTime = time(NULL);
 588
 589                 //column file
 590                 ofstream outFile(dFileName.c_str(), ios::trunc);
 591                 outFile.setf(ios::fixed, ios::showpoint);
 592                 outFile << setprecision(4);
 593
 594                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 595
 596                 for(int i=startLine;i<endLine;i++){
 597                         if(output == "lt")      {
 598                                 string name = alignDB.get(i).getName();
 599                                 if (name.length() < 10) { //pad with spaces to make compatible
 600                                         while (name.length() < 10) {  name += " ";  }
 601                                 }
 602                                 outFile << name << '\t';
 603                         }
 604                         for(int j=0;j<i;j++){
 605
 606                                 if (m->control_pressed) { outFile.close(); return 0;  }
 607
 608                                 //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
 609                                 //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
 610                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 611
 612                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 613                                 double dist = distCalculator->getDist();
 614
 615                                 if(dist <= cutoff){
 616                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 617                                 }
 618                                 if (output == "lt") {  outFile << dist << '\t'; }
 619                         }
 620
 621                         if (output == "lt") { outFile << endl; }
 622
 623                         if(i % 100 == 0){
 624                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 625                         }
 626
 627                 }
 628                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 629
 630                 outFile.close();
 631
 632                 return 1;
 633         }
 634         catch(exception& e) {
 635                 m->errorOut(e, "DistanceCommand", "driver");
 636                 exit(1);
 637         }
 638 }
 639 /**************************************************************************************************/
 640 /////// need to fix to work with calcs and sequencedb
 641 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 642         try {
 643
 644                 int startTime = time(NULL);
 645
 646                 //column file
 647                 ofstream outFile(dFileName.c_str(), ios::trunc);
 648                 outFile.setf(ios::fixed, ios::showpoint);
 649                 outFile << setprecision(4);
 650
 651                 if(startLine == 0){     outFile << alignDB.getNumSeqs() << endl;        }
 652
 653                 for(int i=startLine;i<endLine;i++){
 654
 655                         string name = alignDB.get(i).getName();
 656                         //pad with spaces to make compatible
 657                         if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 658
 659                         outFile << name << '\t';
 660
 661                         for(int j=0;j<alignDB.getNumSeqs();j++){
 662
 663                                 if (m->control_pressed) { outFile.close(); return 0;  }
 664
 665                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 666                                 double dist = distCalculator->getDist();
 667
 668                                 outFile << dist << '\t';
 669                         }
 670
 671                         outFile << endl;
 672
 673                         if(i % 100 == 0){
 674                                 m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 675                         }
 676
 677                 }
 678                 m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 679
 680                 outFile.close();
 681
 682                 return 1;
 683         }
 684         catch(exception& e) {
 685                 m->errorOut(e, "DistanceCommand", "driver");
 686                 exit(1);
 687         }
 688 }
 689 #ifdef USE_MPI
 690 /**************************************************************************************************/
 691 /////// need to fix to work with calcs and sequencedb
 692 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 693         try {
 694                 MPI_Status status;
 695                 int startTime = time(NULL);
 696
 697                 string outputString = "";
 698
 699                 for(int i=startLine;i<endLine;i++){
 700
 701                         for(int j=0;j<i;j++){
 702
 703                                 if (m->control_pressed) {  return 0;  }
 704
 705                                 //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
 706                                 //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
 707                                 if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
 708
 709                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 710                                 double dist = distCalculator->getDist();
 711
 712                                 if(dist <= cutoff){
 713                                          outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
 714                                 }
 715                         }
 716
 717                         if(i % 100 == 0){
 718                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 719                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 720                         }
 721
 722
 723                         //send results to parent
 724                         int length = outputString.length();
 725
 726                         char* buf = new char[length];
 727                         memcpy(buf, outputString.c_str(), length);
 728
 729                         MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
 730                         outputString = "";
 731                         delete buf;
 732
 733                 }
 734
 735                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 736                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 737                 return 1;
 738         }
 739         catch(exception& e) {
 740                 m->errorOut(e, "DistanceCommand", "driverMPI");
 741                 exit(1);
 742         }
 743 }
 744 /**************************************************************************************************/
 745 /////// need to fix to work with calcs and sequencedb
 746 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
 747         try {
 748                 MPI_Status status;
 749
 750                 MPI_File outMPI;
 751                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 752
 753                 //char* filename = new char[file.length()];
 754                 //memcpy(filename, file.c_str(), file.length());
 755
 756                 char filename[1024];
 757                 strcpy(filename, file.c_str());
 758
 759                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 760                 //delete filename;
 761
 762                 int startTime = time(NULL);
 763
 764                 string outputString = "";
 765                 size = 0;
 766
 767                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 768
 769                 for(int i=startLine;i<endLine;i++){
 770
 771                         string name = alignDB.get(i).getName();
 772                         if (name.length() < 10) { //pad with spaces to make compatible
 773                                 while (name.length() < 10) {  name += " ";  }
 774                         }
 775                         outputString += name + "\t";
 776
 777                         for(int j=0;j<i;j++){
 778
 779                                 if (m->control_pressed) {  return 0;  }
 780
 781                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 782                                 double dist = distCalculator->getDist();
 783
 784                                 outputString += toString(dist) + "\t";
 785                         }
 786
 787                         outputString += "\n";
 788
 789
 790                         if(i % 100 == 0){
 791                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 792                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 793                         }
 794
 795
 796                         //send results to parent
 797                         int length = outputString.length();
 798                         char* buf = new char[length];
 799                         memcpy(buf, outputString.c_str(), length);
 800
 801                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 802                         size += outputString.length();
 803                         outputString = "";
 804                         delete buf;
 805                 }
 806
 807                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 808                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 809                 MPI_File_close(&outMPI);
 810
 811                 return 1;
 812         }
 813         catch(exception& e) {
 814                 m->errorOut(e, "DistanceCommand", "driverMPI");
 815                 exit(1);
 816         }
 817 }
 818 /**************************************************************************************************/
 819 /////// need to fix to work with calcs and sequencedb
 820 int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
 821         try {
 822                 MPI_Status status;
 823
 824                 MPI_File outMPI;
 825                 int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 826
 827                 //char* filename = new char[file.length()];
 828                 //memcpy(filename, file.c_str(), file.length());
 829
 830                 char filename[1024];
 831                 strcpy(filename, file.c_str());
 832
 833                 MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
 834                 //delete filename;
 835
 836                 int startTime = time(NULL);
 837
 838                 string outputString = "";
 839                 size = 0;
 840
 841                 if(startLine == 0){     outputString += toString(alignDB.getNumSeqs()) + "\n";  }
 842
 843                 for(int i=startLine;i<endLine;i++){
 844
 845                         string name = alignDB.get(i).getName();
 846                         if (name.length() < 10) { //pad with spaces to make compatible
 847                                 while (name.length() < 10) {  name += " ";  }
 848                         }
 849                         outputString += name + "\t";
 850
 851                         for(int j=0;j<alignDB.getNumSeqs();j++){
 852
 853                                 if (m->control_pressed) {  return 0;  }
 854
 855                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 856                                 double dist = distCalculator->getDist();
 857
 858                                 outputString += toString(dist) + "\t";
 859                         }
 860
 861                         outputString += "\n";
 862
 863
 864                         if(i % 100 == 0){
 865                                 //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 866                                 cout << i << '\t' << (time(NULL) - startTime) << endl;
 867                         }
 868
 869
 870                         //send results to parent
 871                         int length = outputString.length();
 872                         char* buf = new char[length];
 873                         memcpy(buf, outputString.c_str(), length);
 874
 875                         MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
 876                         size += outputString.length();
 877                         outputString = "";
 878                         delete buf;
 879                 }
 880
 881                 //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
 882                 cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
 883                 MPI_File_close(&outMPI);
 884
 885                 return 1;
 886         }
 887         catch(exception& e) {
 888                 m->errorOut(e, "DistanceCommand", "driverMPI");
 889                 exit(1);
 890         }
 891 }
 892 #endif
 893 /**************************************************************************************************
 894 int DistanceCommand::convertMatrix(string outputFile) {
 895         try{
 896
 897                 //sort file by first column so the distances for each row are together
 898                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 899
 900                 //use the unix sort
 901                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 902                         string command = "sort -n " + outputFile + " -o " + outfile;
 903                         system(command.c_str());
 904                 #else //sort using windows sort
 905                         string command = "sort " + outputFile + " /O " + outfile;
 906                         system(command.c_str());
 907                 #endif
 908
 909
 910                 //output to new file distance for each row and save positions in file where new row begins
 911                 ifstream in;
 912                 m->openInputFile(outfile, in);
 913
 914                 ofstream out;
 915                 m->openOutputFile(outputFile, out);
 916
 917                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 918
 919                 out << alignDB.getNumSeqs() << endl;
 920
 921                 //get first currentRow
 922                 string first, currentRow, second;
 923                 float dist;
 924                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 925                 map<string, float>::iterator it;
 926
 927                 in >> first;
 928                 currentRow = first;
 929
 930                 rowDists[first] = 0.00; //distance to yourself is 0.0
 931
 932                 in.seekg(0);
 933                 //m->openInputFile(outfile, in);
 934
 935                 while(!in.eof()) {
 936                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
 937
 938                         in >> first >> second >> dist; m->gobble(in);
 939
 940                         if (first != currentRow) {
 941                                 //print out last row
 942                                 out << currentRow << '\t'; //print name
 943
 944                                 //print dists
 945                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 946                                         out << it->second << '\t';
 947                                 }
 948                                 out << endl;
 949
 950                                 //start new row
 951                                 currentRow = first;
 952                                 rowDists.clear();
 953                                 rowDists[first] = 0.00;
 954                                 rowDists[second] = dist;
 955                         }else{
 956                                 rowDists[second] = dist;
 957                         }
 958                 }
 959                 //print out last row
 960                 out << currentRow << '\t'; //print name
 961
 962                 //print dists
 963                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 964                         out << it->second << '\t';
 965                 }
 966                 out << endl;
 967
 968                 in.close();
 969                 out.close();
 970
 971                 remove(outfile.c_str());
 972
 973                 return 1;
 974
 975         }
 976         catch(exception& e) {
 977                 m->errorOut(e, "DistanceCommand", "convertMatrix");
 978                 exit(1);
 979         }
 980 }
 981 /**************************************************************************************************
 982 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 983         try{
 984
 985                 //sort file by first column so the distances for each row are together
 986                 string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 987
 988                 //use the unix sort
 989                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 990                         string command = "sort -n " + outputFile + " -o " + outfile;
 991                         system(command.c_str());
 992                 #else //sort using windows sort
 993                         string command = "sort " + outputFile + " /O " + outfile;
 994                         system(command.c_str());
 995                 #endif
 996
 997
 998                 //output to new file distance for each row and save positions in file where new row begins
 999                 ifstream in;
1000                 m->openInputFile(outfile, in);
1001
1002                 ofstream out;
1003                 m->openOutputFile(outputFile, out);
1004
1005                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1006
1007                 out << alignDB.getNumSeqs() << endl;
1008
1009                 //get first currentRow
1010                 string first, currentRow, second;
1011                 float dist;
1012                 int i, j;
1013                 i = 0; j = 0;
1014                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
1015                 map<string, float>::iterator it;
1016
1017                 in >> first;
1018                 currentRow = first;
1019
1020                 rowDists[first] = 0.00; //distance to yourself is 0.0
1021
1022                 in.seekg(0);
1023                 //m->openInputFile(outfile, in);
1024
1025                 while(!in.eof()) {
1026                         if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
1027
1028                         in >> first >> second >> dist; m->gobble(in);
1029
1030                         if (first != currentRow) {
1031                                 //print out last row
1032                                 out << currentRow << '\t'; //print name
1033
1034                                 //print dists
1035                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1036                                         if (j >= i) { break; }
1037                                         out << it->second << '\t';
1038                                         j++;
1039                                 }
1040                                 out << endl;
1041
1042                                 //start new row
1043                                 currentRow = first;
1044                                 rowDists.clear();
1045                                 rowDists[first] = 0.00;
1046                                 rowDists[second] = dist;
1047                                 j = 0;
1048                                 i++;
1049                         }else{
1050                                 rowDists[second] = dist;
1051                         }
1052                 }
1053                 //print out last row
1054                 out << currentRow << '\t'; //print name
1055
1056                 //print dists
1057                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
1058                         out << it->second << '\t';
1059                 }
1060                 out << endl;
1061
1062                 in.close();
1063                 out.close();
1064
1065                 remove(outfile.c_str());
1066
1067                 return 1;
1068
1069         }
1070         catch(exception& e) {
1071                 m->errorOut(e, "DistanceCommand", "convertToLowerTriangle");
1072                 exit(1);
1073         }
1074 }
1075 /**************************************************************************************************/
1076 //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,
1077 //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.
1078 //also check to make sure the 2 files have the same alignment length.
1079 bool DistanceCommand::sanityCheck() {
1080         try{
1081                 bool good = true;
1082
1083                 //make sure the 2 fasta files have the same alignment length
1084                 ifstream in;
1085                 m->openInputFile(fastafile, in);
1086                 int fastaAlignLength = 0;
1087                 if (in) {
1088                         Sequence tempIn(in);
1089                         fastaAlignLength = tempIn.getAligned().length();
1090                 }
1091                 in.close();
1092
1093                 ifstream in2;
1094                 m->openInputFile(oldfastafile, in2);
1095                 int oldfastaAlignLength = 0;
1096                 if (in2) {
1097                         Sequence tempIn2(in2);
1098                         oldfastaAlignLength = tempIn2.getAligned().length();
1099                 }
1100                 in2.close();
1101
1102                 if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false;  }
1103
1104                 //read fasta file and save names as well as adding them to the alignDB
1105                 set<string> namesOldFasta;
1106
1107                 ifstream inFasta;
1108                 m->openInputFile(oldfastafile, inFasta);
1109
1110                 while (!inFasta.eof()) {
1111                         if (m->control_pressed) {  inFasta.close(); return good;  }
1112
1113                         Sequence temp(inFasta);
1114
1115                         if (temp.getName() != "") {
1116                                 namesOldFasta.insert(temp.getName());  //save name
1117                                 alignDB.push_back(temp);  //add to DB
1118                         }
1119
1120                         m->gobble(inFasta);
1121                 }
1122
1123                 inFasta.close();
1124
1125                 //read through the column file checking names and removing distances above the cutoff
1126                 ifstream inDist;
1127                 m->openInputFile(column, inDist);
1128
1129                 ofstream outDist;
1130                 string outputFile = column + ".temp";
1131                 m->openOutputFile(outputFile, outDist);
1132
1133                 string name1, name2;
1134                 float dist;
1135                 while (!inDist.eof()) {
1136                         if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
1137
1138                         inDist >> name1 >> name2 >> dist; m->gobble(inDist);
1139
1140                         //both names are in fasta file and distance is below cutoff
1141                         if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) {  good = false; break;  }
1142                         else{
1143                                 if (dist <= cutoff) {
1144                                         outDist << name1 << '\t' << name2 << '\t' << dist << endl;
1145                                 }
1146                         }
1147                 }
1148
1149                 inDist.close();
1150                 outDist.close();
1151
1152                 if (good) {
1153                         remove(column.c_str());
1154                         rename(outputFile.c_str(), column.c_str());
1155                 }else{
1156                         remove(outputFile.c_str()); //temp file is bad because file mismatch above
1157                 }
1158
1159                 return good;
1160
1161         }
1162         catch(exception& e) {
1163                 m->errorOut(e, "DistanceCommand", "sanityCheck");
1164                 exit(1);
1165         }
1166 }
1167 /**************************************************************************************************/
1168
1169
1170
1171