distancecommand.cpp

   1 /*
   2  *  distancecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/7/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "distancecommand.h"
  11 #include "ignoregaps.h"
  12 #include "eachgapdist.h"
  13 #include "eachgapignore.h"
  14 #include "onegapdist.h"
  15 #include "onegapignore.h"
  16
  17 //**********************************************************************************************************************
  18
  19 DistanceCommand::DistanceCommand(string option){
  20         try {
  21                 abort = false;
  22                 Estimators.clear();
  23
  24                 //allow user to run help
  25                 if(option == "help") { help(); abort = true; }
  26
  27                 else {
  28                         //valid paramters for this command
  29                         string Array[] =  {"fasta", "phylip", "calc", "countends", "cutoff", "processors"};
  30                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  31
  32                         OptionParser parser(option);
  33                         map<string, string> parameters = parser.getParameters();
  34
  35                         ValidParameters validParameter;
  36
  37                         //check to make sure all parameters are valid for command
  38                         for (map<string, string>::iterator it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  39                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  40                         }
  41
  42                         //check for required parameters
  43                         fastafile = validParameter.validFile(parameters, "fasta", true);
  44                         if (fastafile == "not found") { cout << "fasta is a required parameter for the dist.seqs command." << endl; abort = true; }
  45                         else if (fastafile == "not open") { abort = true; }
  46                         else{
  47                                 ifstream inFASTA;
  48                                 openInputFile(fastafile, inFASTA);
  49                                 alignDB = SequenceDB(inFASTA);
  50                                 inFASTA.close();
  51                         }
  52
  53                         //check for optional parameter and set defaults
  54                         // ...at some point should added some additional type checking...
  55                         calc = validParameter.validFile(parameters, "calc", false);
  56                         if (calc == "not found") { calc = "onegap";  }
  57                         else {
  58                                  if (calc == "default")  {  calc = "onegap";  }
  59                         }
  60                         splitAtDash(calc, Estimators);
  61
  62                         string temp;
  63                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
  64                         convert(temp, countends);
  65
  66                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
  67                         convert(temp, cutoff);
  68
  69                         temp = validParameter.validFile(parameters, "processors", false);       if(temp == "not found"){        temp = "1"; }
  70                         convert(temp, processors);
  71
  72                         phylip = validParameter.validFile(parameters, "phylip", false);         if(phylip == "not found"){      phylip = "F"; }
  73
  74
  75                         ValidCalculators validCalculator;
  76
  77                         if (isTrue(countends) == true) {
  78                                 for (int i=0; i<Estimators.size(); i++) {
  79                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
  80                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
  81                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
  82                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
  83                                         }
  84                                 }
  85                         }else {
  86                                 for (int i=0; i<Estimators.size(); i++) {
  87                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
  88                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
  89                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
  90                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
  91                                         }
  92                                 }
  93                         }
  94
  95                 }
  96
  97         }
  98         catch(exception& e) {
  99                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function DistanceCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 100                 exit(1);
 101         }
 102         catch(...) {
 103                 cout << "An unknown error has occurred in the DistanceCommand class function DistanceCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 104                 exit(1);
 105         }
 106 }
 107
 108 //**********************************************************************************************************************
 109
 110 DistanceCommand::~DistanceCommand(){
 111
 112         for(int i=0;i<lines.size();i++){
 113                 delete lines[i];
 114         }
 115
 116 }
 117
 118 //**********************************************************************************************************************
 119
 120 void DistanceCommand::help(){
 121         try {
 122                 cout << "The dist.seqs command reads a file containing sequences and creates a distance file." << "\n";
 123                 cout << "The dist.seqs command parameters are fasta, calc, countends, cutoff and processors.  " << "\n";
 124                 cout << "The fasta parameter is required." << "\n";
 125                 cout << "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";
 126                 cout << "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";
 127                 cout << "The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0." << "\n";
 128                 cout << "The processors parameter allows you to specify number of processors to use.  The default is 1." << "\n";
 129                 cout << "The dist.seqs command should be in the following format: " << "\n";
 130                 cout << "dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) " << "\n";
 131                 cout << "Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3)." << "\n";
 132                 cout << "Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc)." << "\n" << "\n";
 133         }
 134         catch(exception& e) {
 135                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 136                 exit(1);
 137         }
 138         catch(...) {
 139                 cout << "An unknown error has occurred in the DistanceCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 140                 exit(1);
 141         }
 142 }
 143
 144
 145 //**********************************************************************************************************************
 146
 147 int DistanceCommand::execute(){
 148         try {
 149
 150                 if (abort == true) { return 0; }
 151
 152                 int numSeqs = alignDB.getNumSeqs();
 153                 cutoff += 0.005;
 154
 155                 string outputFile;
 156
 157                 //doses the user want the phylip formatted file as well
 158                 if (isTrue(phylip) == true) {
 159                         outputFile = getRootName(fastafile) + "phylip.dist";
 160                         remove(outputFile.c_str());
 161
 162                         //output numSeqs to phylip formatted dist file
 163                 }else { //user wants column format
 164                         outputFile = getRootName(fastafile) + "dist";
 165                         remove(outputFile.c_str());
 166                 }
 167
 168                 //#     if defined (_WIN32)
 169                 //figure out how to implement the fork and wait commands in windows
 170                 //      driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile, cutoff);
 171                 //#     endif
 172
 173
 174 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 175                 //if you don't need to fork anything
 176                 if(processors == 1){
 177                         driver(0, numSeqs, outputFile, cutoff);
 178                 }else{ //you have multiple processors
 179
 180                         for (int i = 0; i < processors; i++) {
 181                                 lines.push_back(new linePair());
 182                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 183                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 184                         }
 185
 186                         createProcesses(outputFile);
 187
 188                         map<int, int>::iterator it = processIDS.begin();
 189                         rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 190                         it++;
 191
 192                         //append and remove temp files
 193                         for (; it != processIDS.end(); it++) {
 194                                 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 195                                 remove((outputFile + toString(it->second) + ".temp").c_str());
 196                         }
 197                 }
 198 #else
 199                 ifstream inFASTA;
 200                 driver(0, numSeqs, outputFile, cutoff);
 201 #endif
 202
 203                 delete distCalculator;
 204
 205                 return 0;
 206
 207         }
 208         catch(exception& e) {
 209                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 210                 exit(1);
 211         }
 212         catch(...) {
 213                 cout << "An unknown error has occurred in the DistanceCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 214                 exit(1);
 215         }
 216 }
 217 /**************************************************************************************************/
 218 void DistanceCommand::createProcesses(string filename) {
 219         try {
 220 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 221                 int process = 0;
 222                 processIDS.clear();
 223
 224                 //loop through and create all the processes you want
 225                 while (process != processors) {
 226                         int pid = fork();
 227
 228                         if (pid > 0) {
 229                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 230                                 process++;
 231                         }else if (pid == 0){
 232                                 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
 233                                 exit(0);
 234                         }else { cout << "unable to spawn the necessary processes." << endl; exit(0); }
 235                 }
 236
 237                 //force parent to wait until all the processes are done
 238                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 239                         int temp = it->second;
 240                         wait(&temp);
 241                 }
 242 #endif
 243         }
 244         catch(exception& e) {
 245                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function createProcesses. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 246                 exit(1);
 247         }
 248         catch(...) {
 249                 cout << "An unknown error has occurred in the DistanceCommand class function createProcesses. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 250                 exit(1);
 251         }
 252 }
 253
 254 /**************************************************************************************************/
 255 /////// need to fix to work with calcs and sequencedb
 256 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 257         try {
 258
 259                 int startTime = time(NULL);
 260
 261                 //column file
 262                 ofstream outFile(dFileName.c_str(), ios::trunc);
 263                 outFile.setf(ios::fixed, ios::showpoint);
 264                 outFile << setprecision(4);
 265
 266                 if(isTrue(phylip) && startLine == 0){   outFile << alignDB.getNumSeqs() << endl;        }
 267                 for(int i=startLine;i<endLine;i++){
 268                         if(isTrue(phylip))      {       outFile << alignDB.get(i).getName() << '\t';    }
 269                         for(int j=0;j<i;j++){
 270                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 271                                 double dist = distCalculator->getDist();
 272
 273                                 if(dist <= cutoff){
 274                                         if (!isTrue(phylip)) { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 275                                 }
 276                                 if (isTrue(phylip)) {  outFile << dist << '\t'; }
 277
 278                         }
 279
 280                         if (isTrue(phylip) == true) { outFile << endl; }
 281
 282                         if(i % 100 == 0){
 283                                 cout << i << '\t' << time(NULL) - startTime << endl;
 284                         }
 285
 286                 }
 287                 cout << endLine-1 << '\t' << time(NULL) - startTime << endl;
 288
 289                 outFile.close();
 290
 291                 return 1;
 292         }
 293         catch(exception& e) {
 294                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function driver. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 295                 exit(1);
 296         }
 297         catch(...) {
 298                 cout << "An unknown error has occurred in the DistanceCommand class function driver. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 299                 exit(1);
 300         }
 301
 302 }
 303
 304 /**************************************************************************************************/
 305 void DistanceCommand::appendFiles(string temp, string filename) {
 306         try{
 307                 ofstream output;
 308                 ifstream input;
 309
 310                 //open output file in append mode
 311                 openOutputFileAppend(filename, output);
 312                 openInputFile(temp, input);
 313
 314                 while(char c = input.get()){
 315                         if(input.eof())         {       break;                  }
 316                         else                            {       output << c;    }
 317                 }
 318
 319                 input.close();
 320                 output.close();
 321         }
 322         catch(exception& e) {
 323                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 324                 exit(1);
 325         }
 326         catch(...) {
 327                 cout << "An unknown error has occurred in the DistanceCommand class function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 328                 exit(1);
 329         }
 330 }
 331 /**************************************************************************************************/