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", "outputdir","inputdir"};
  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                         map<string, string>::iterator it2;
  37
  38                         //check to make sure all parameters are valid for command
  39                         for (it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  40                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  41                         }
  42
  43                         //if the user changes the input directory command factory will send this info to us in the output parameter
  44                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  45                         if (inputDir == "not found"){   inputDir = "";          }
  46                         else {
  47                                 string path;
  48                                 it2 = parameters.find("fasta");
  49                                 //user has given a template file
  50                                 if(it2 != parameters.end()){
  51                                         path = hasPath(it2->second);
  52                                         //if the user has not given a path then, add inputdir. else leave path alone.
  53                                         if (path == "") {       parameters["fasta"] = inputDir + it2->second;           }
  54                                 }
  55                         }
  56
  57                         //check for required parameters
  58                         fastafile = validParameter.validFile(parameters, "fasta", true);
  59                         if (fastafile == "not found") { mothurOut("fasta is a required parameter for the dist.seqs command."); mothurOutEndLine(); abort = true; }
  60                         else if (fastafile == "not open") { abort = true; }
  61                         else{
  62                                 ifstream inFASTA;
  63                                 openInputFile(fastafile, inFASTA);
  64                                 alignDB = SequenceDB(inFASTA);
  65                                 inFASTA.close();
  66                         }
  67
  68                         //if the user changes the output directory command factory will send this info to us in the output parameter
  69                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  70                                 outputDir = "";
  71                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
  72                         }
  73
  74                         //check for optional parameter and set defaults
  75                         // ...at some point should added some additional type checking...
  76                         calc = validParameter.validFile(parameters, "calc", false);
  77                         if (calc == "not found") { calc = "onegap";  }
  78                         else {
  79                                  if (calc == "default")  {  calc = "onegap";  }
  80                         }
  81                         splitAtDash(calc, Estimators);
  82
  83                         string temp;
  84                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
  85                         convert(temp, countends);
  86
  87                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
  88                         convert(temp, cutoff);
  89
  90                         temp = validParameter.validFile(parameters, "processors", false);       if(temp == "not found"){        temp = "1"; }
  91                         convert(temp, processors);
  92
  93                         phylip = validParameter.validFile(parameters, "phylip", false);         if(phylip == "not found"){      phylip = "F"; }
  94
  95
  96                         ValidCalculators validCalculator;
  97
  98                         if (isTrue(countends) == true) {
  99                                 for (int i=0; i<Estimators.size(); i++) {
 100                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 101                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
 102                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
 103                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
 104                                         }
 105                                 }
 106                         }else {
 107                                 for (int i=0; i<Estimators.size(); i++) {
 108                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 109                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
 110                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
 111                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
 112                                         }
 113                                 }
 114                         }
 115
 116                 }
 117
 118         }
 119         catch(exception& e) {
 120                 errorOut(e, "DistanceCommand", "DistanceCommand");
 121                 exit(1);
 122         }
 123 }
 124
 125 //**********************************************************************************************************************
 126
 127 DistanceCommand::~DistanceCommand(){
 128
 129         for(int i=0;i<lines.size();i++){
 130                 delete lines[i];
 131         }
 132
 133 }
 134
 135 //**********************************************************************************************************************
 136
 137 void DistanceCommand::help(){
 138         try {
 139                 mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
 140                 mothurOut("The dist.seqs command parameters are fasta, calc, countends, cutoff and processors.  \n");
 141                 mothurOut("The fasta parameter is required.\n");
 142                 mothurOut("The calc parameter allows you to specify the method of calculating the distances.  Your options are: nogaps, onegap or eachgap. The default is onegap.\n");
 143                 mothurOut("The countends parameter allows you to specify whether to include terminal gaps in distance.  Your options are: T or F. The default is T.\n");
 144                 mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 145                 mothurOut("The processors parameter allows you to specify number of processors to use.  The default is 1.\n");
 146                 mothurOut("The dist.seqs command should be in the following format: \n");
 147                 mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
 148                 mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
 149                 mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
 150         }
 151         catch(exception& e) {
 152                 errorOut(e, "DistanceCommand", "help");
 153                 exit(1);
 154         }
 155 }
 156 //**********************************************************************************************************************
 157
 158 int DistanceCommand::execute(){
 159         try {
 160
 161                 if (abort == true) { return 0; }
 162
 163                 int numSeqs = alignDB.getNumSeqs();
 164                 cutoff += 0.005;
 165
 166                 string outputFile;
 167
 168                 //doses the user want the phylip formatted file as well
 169                 if (isTrue(phylip) == true) {
 170                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "phylip.dist";
 171                         remove(outputFile.c_str());
 172
 173                         //output numSeqs to phylip formatted dist file
 174                 }else { //user wants column format
 175                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "dist";
 176                         remove(outputFile.c_str());
 177                 }
 178
 179 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 180                 //if you don't need to fork anything
 181                 if(processors == 1){
 182                         driver(0, numSeqs, outputFile, cutoff);
 183                 }else{ //you have multiple processors
 184
 185                         for (int i = 0; i < processors; i++) {
 186                                 lines.push_back(new linePair());
 187                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 188                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 189                         }
 190
 191                         createProcesses(outputFile);
 192
 193                         map<int, int>::iterator it = processIDS.begin();
 194                         rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 195                         it++;
 196
 197                         //append and remove temp files
 198                         for (; it != processIDS.end(); it++) {
 199                                 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 200                                 remove((outputFile + toString(it->second) + ".temp").c_str());
 201                         }
 202                 }
 203 #else
 204                 ifstream inFASTA;
 205                 driver(0, numSeqs, outputFile, cutoff);
 206 #endif
 207
 208                 delete distCalculator;
 209
 210                 return 0;
 211
 212         }
 213         catch(exception& e) {
 214                 errorOut(e, "DistanceCommand", "execute");
 215                 exit(1);
 216         }
 217 }
 218 /**************************************************************************************************/
 219 void DistanceCommand::createProcesses(string filename) {
 220         try {
 221 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 222                 int process = 0;
 223                 processIDS.clear();
 224
 225                 //loop through and create all the processes you want
 226                 while (process != processors) {
 227                         int pid = fork();
 228
 229                         if (pid > 0) {
 230                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 231                                 process++;
 232                         }else if (pid == 0){
 233                                 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
 234                                 exit(0);
 235                         }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
 236                 }
 237
 238                 //force parent to wait until all the processes are done
 239                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 240                         int temp = it->second;
 241                         wait(&temp);
 242                 }
 243 #endif
 244         }
 245         catch(exception& e) {
 246                 errorOut(e, "DistanceCommand", "createProcesses");
 247                 exit(1);
 248         }
 249 }
 250
 251 /**************************************************************************************************/
 252 /////// need to fix to work with calcs and sequencedb
 253 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 254         try {
 255
 256                 int startTime = time(NULL);
 257
 258                 //column file
 259                 ofstream outFile(dFileName.c_str(), ios::trunc);
 260                 outFile.setf(ios::fixed, ios::showpoint);
 261                 outFile << setprecision(4);
 262
 263                 if(isTrue(phylip) && startLine == 0){   outFile << alignDB.getNumSeqs() << endl;        }
 264
 265                 for(int i=startLine;i<endLine;i++){
 266                         if(isTrue(phylip))      {
 267                                 string name = alignDB.get(i).getName();
 268                                 if (name.length() < 10) { //pad with spaces to make compatible
 269                                         while (name.length() < 10) {  name += " ";  }
 270                                 }
 271                                 outFile << name << '\t';
 272                         }
 273                         for(int j=0;j<i;j++){
 274                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 275                                 double dist = distCalculator->getDist();
 276
 277                                 if(dist <= cutoff){
 278                                         if (!isTrue(phylip)) { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 279                                 }
 280                                 if (isTrue(phylip)) {  outFile << dist << '\t'; }
 281
 282                         }
 283
 284                         if (isTrue(phylip) == true) { outFile << endl; }
 285
 286                         if(i % 100 == 0){
 287                                 mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); mothurOutEndLine();
 288                         }
 289
 290                 }
 291                 mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); mothurOutEndLine();
 292
 293                 outFile.close();
 294
 295                 return 1;
 296         }
 297         catch(exception& e) {
 298                 errorOut(e, "DistanceCommand", "driver");
 299                 exit(1);
 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                 openInputFile(temp, input);
 312
 313                 while(char c = input.get()){
 314                         if(input.eof())         {       break;                  }
 315                         else                            {       output << c;    }
 316                 }
 317
 318                 input.close();
 319                 output.close();
 320         }
 321         catch(exception& e) {
 322                 errorOut(e, "DistanceCommand", "appendFiles");
 323                 exit(1);
 324         }
 325 }
 326 /**************************************************************************************************/