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                         }
  51
  52                         //check for optional parameter and set defaults
  53                         // ...at some point should added some additional type checking...
  54                         calc = validParameter.validFile(parameters, "calc", false);
  55                         if (calc == "not found") { calc = "onegap";  }
  56                         else {
  57                                  if (calc == "default")  {  calc = "onegap";  }
  58                         }
  59                         splitAtDash(calc, Estimators);
  60
  61                         string temp;
  62                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
  63                         convert(temp, countends);
  64
  65                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
  66                         convert(temp, cutoff);
  67
  68                         temp = validParameter.validFile(parameters, "processors", false);       if(temp == "not found"){        temp = "1"; }
  69                         convert(temp, processors);
  70
  71                         phylip = validParameter.validFile(parameters, "phylip", false);         if(phylip == "not found"){      phylip = "F"; }
  72
  73
  74                         ValidCalculators validCalculator;
  75
  76                         if (isTrue(countends) == true) {
  77                                 for (int i=0; i<Estimators.size(); i++) {
  78                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
  79                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
  80                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
  81                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
  82                                         }
  83                                 }
  84                         }else {
  85                                 for (int i=0; i<Estimators.size(); i++) {
  86                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
  87                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
  88                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
  89                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
  90                                         }
  91                                 }
  92                         }
  93
  94                 }
  95
  96         }
  97         catch(exception& e) {
  98                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function DistanceCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  99                 exit(1);
 100         }
 101         catch(...) {
 102                 cout << "An unknown error has occurred in the DistanceCommand class function DistanceCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 103                 exit(1);
 104         }
 105 }
 106
 107 //**********************************************************************************************************************
 108
 109 DistanceCommand::~DistanceCommand(){
 110
 111         for(int i=0;i<lines.size();i++){
 112                 delete lines[i];
 113         }
 114
 115 }
 116
 117 //**********************************************************************************************************************
 118
 119 void DistanceCommand::help(){
 120         try {
 121                 cout << "The dist.seqs command reads a file containing sequences and creates a distance file." << "\n";
 122                 cout << "The dist.seqs command parameters are fasta, calc, countends, cutoff and processors.  " << "\n";
 123                 cout << "The fasta parameter is required." << "\n";
 124                 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";
 125                 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";
 126                 cout << "The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0." << "\n";
 127                 cout << "The processors parameter allows you to specify number of processors to use.  The default is 1." << "\n";
 128                 cout << "The dist.seqs command should be in the following format: " << "\n";
 129                 cout << "dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) " << "\n";
 130                 cout << "Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3)." << "\n";
 131                 cout << "Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc)." << "\n" << "\n";
 132         }
 133         catch(exception& e) {
 134                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 135                 exit(1);
 136         }
 137         catch(...) {
 138                 cout << "An unknown error has occurred in the DistanceCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 139                 exit(1);
 140         }
 141 }
 142
 143
 144 //**********************************************************************************************************************
 145
 146 int DistanceCommand::execute(){
 147         try {
 148
 149                 if (abort == true) { return 0; }
 150
 151                 int numSeqs = alignDB.getNumSeqs();
 152                 cutoff += 0.005;
 153
 154                 string outputFile;
 155
 156                 //doses the user want the phylip formatted file as well
 157                 if (isTrue(phylip) == true) {
 158                         outputFile = getRootName(fastafile) + "phylip.dist";
 159                         remove(outputFile.c_str());
 160
 161                         //output numSeqs to phylip formatted dist file
 162                 }else { //user wants column format
 163                         outputFile = getRootName(fastafile) + "dist";
 164                         remove(outputFile.c_str());
 165                 }
 166
 167                 //#     if defined (_WIN32)
 168                 //figure out how to implement the fork and wait commands in windows
 169                 //      driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile, cutoff);
 170                 //#     endif
 171
 172
 173 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 174                 //if you don't need to fork anything
 175                 if(processors == 1){
 176                         driver(0, numSeqs, outputFile, cutoff);
 177                 }else{ //you have multiple processors
 178
 179                         for (int i = 0; i < processors; i++) {
 180                                 lines.push_back(new linePair());
 181                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 182                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 183                         }
 184
 185                         createProcesses(outputFile);
 186
 187                         map<int, int>::iterator it = processIDS.begin();
 188                         rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 189                         it++;
 190
 191                         //append and remove temp files
 192                         for (; it != processIDS.end(); it++) {
 193                                 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 194                                 remove((outputFile + toString(it->second) + ".temp").c_str());
 195                         }
 196                 }
 197 #else
 198                 ifstream inFASTA
 199                 driver(0, numSeqs, outputFile, cutoff);
 200 #endif
 201
 202                 delete distCalculator;
 203
 204                 return 0;
 205
 206         }
 207         catch(exception& e) {
 208                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 209                 exit(1);
 210         }
 211         catch(...) {
 212                 cout << "An unknown error has occurred in the DistanceCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 213                 exit(1);
 214         }
 215 }
 216 /**************************************************************************************************/
 217 void DistanceCommand::createProcesses(string filename) {
 218         try {
 219 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 220                 int process = 0;
 221                 processIDS.clear();
 222
 223                 //loop through and create all the processes you want
 224                 while (process != processors) {
 225                         int pid = fork();
 226
 227                         if (pid > 0) {
 228                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 229                                 process++;
 230                         }else if (pid == 0){
 231                                 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
 232                                 exit(0);
 233                         }else { cout << "unable to spawn the necessary processes." << endl; exit(0); }
 234                 }
 235
 236                 //force parent to wait until all the processes are done
 237                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 238                         int temp = it->second;
 239                         wait(&temp);
 240                 }
 241 #endif
 242         }
 243         catch(exception& e) {
 244                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function createProcesses. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 245                 exit(1);
 246         }
 247         catch(...) {
 248                 cout << "An unknown error has occurred in the DistanceCommand class function createProcesses. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 249                 exit(1);
 250         }
 251 }
 252
 253 /**************************************************************************************************/
 254 /////// need to fix to work with calcs and sequencedb
 255 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 256         try {
 257
 258                 int startTime = time(NULL);
 259
 260                 //column file
 261                 ofstream outFile(dFileName.c_str(), ios::trunc);
 262                 outFile.setf(ios::fixed, ios::showpoint);
 263                 outFile << setprecision(4);
 264
 265                 if(isTrue(phylip) && startLine == 0){   outFile << alignDB.getNumSeqs() << endl;        }
 266                 for(int i=startLine;i<endLine;i++){
 267                         if(isTrue(phylip))      {       outFile << alignDB.get(i).getName() << '\t';    }
 268                         for(int j=0;j<i;j++){
 269                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 270                                 double dist = distCalculator->getDist();
 271
 272                                 if(dist <= cutoff){
 273                                         if (!isTrue(phylip)) { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 274                                 }
 275                                 if (isTrue(phylip)) {  outFile << dist << '\t'; }
 276
 277                         }
 278
 279                         if (isTrue(phylip) == true) { outFile << endl; }
 280
 281                         if(i % 100 == 0){
 282                                 cout << i << '\t' << time(NULL) - startTime << endl;
 283                         }
 284
 285                 }
 286                 cout << endLine-1 << '\t' << time(NULL) - startTime << endl;
 287
 288                 outFile.close();
 289
 290                 return 1;
 291         }
 292         catch(exception& e) {
 293                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function driver. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 294                 exit(1);
 295         }
 296         catch(...) {
 297                 cout << "An unknown error has occurred in the DistanceCommand class function driver. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 298                 exit(1);
 299         }
 300
 301 }
 302
 303 /**************************************************************************************************/
 304 void DistanceCommand::appendFiles(string temp, string filename) {
 305         try{
 306                 ofstream output;
 307                 ifstream input;
 308
 309                 //open output file in append mode
 310                 openOutputFileAppend(filename, output);
 311
 312                 //open temp file for reading
 313                 openInputFile(temp, input);
 314
 315                 string line;
 316                 //read input file and write to output file
 317                 while(input.eof() != true) {
 318                         getline(input, line); //getline removes the newline char
 319                         if (line != "") {
 320                                 output << line << endl;   // Appending back newline char
 321                         }
 322                 }
 323
 324                 input.close();
 325                 output.close();
 326         }
 327         catch(exception& e) {
 328                 cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 329                 exit(1);
 330         }
 331         catch(...) {
 332                 cout << "An unknown error has occurred in the DistanceCommand class function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 333                 exit(1);
 334         }
 335 }
 336 /**************************************************************************************************/