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", "output", "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                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "column"; }
  94
  95                         if ((output != "column") && (output != "lt") && (output != "square")) { mothurOut(output + " is not a valid output form. Options are column, lt and square. I will use column."); mothurOutEndLine(); output = "column"; }
  96
  97                         ValidCalculators validCalculator;
  98
  99                         if (isTrue(countends) == true) {
 100                                 for (int i=0; i<Estimators.size(); i++) {
 101                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 102                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
 103                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
 104                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
 105                                         }
 106                                 }
 107                         }else {
 108                                 for (int i=0; i<Estimators.size(); i++) {
 109                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
 110                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
 111                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
 112                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
 113                                         }
 114                                 }
 115                         }
 116
 117                 }
 118
 119         }
 120         catch(exception& e) {
 121                 errorOut(e, "DistanceCommand", "DistanceCommand");
 122                 exit(1);
 123         }
 124 }
 125
 126 //**********************************************************************************************************************
 127
 128 DistanceCommand::~DistanceCommand(){
 129
 130         for(int i=0;i<lines.size();i++){
 131                 delete lines[i];
 132         }
 133
 134 }
 135
 136 //**********************************************************************************************************************
 137
 138 void DistanceCommand::help(){
 139         try {
 140                 mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
 141                 mothurOut("The dist.seqs command parameters are fasta, calc, countends, output, cutoff and processors.  \n");
 142                 mothurOut("The fasta parameter is required.\n");
 143                 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");
 144                 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");
 145                 mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 146                 mothurOut("The output parameter allows you to specify format of your distance matrix. Options are column, lt, and square. The default is column.\n");
 147                 mothurOut("The processors parameter allows you to specify number of processors to use.  The default is 1.\n");
 148                 mothurOut("The dist.seqs command should be in the following format: \n");
 149                 mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
 150                 mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
 151                 mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
 152         }
 153         catch(exception& e) {
 154                 errorOut(e, "DistanceCommand", "help");
 155                 exit(1);
 156         }
 157 }
 158 //**********************************************************************************************************************
 159
 160 int DistanceCommand::execute(){
 161         try {
 162
 163                 if (abort == true) { return 0; }
 164
 165                 int numSeqs = alignDB.getNumSeqs();
 166                 cutoff += 0.005;
 167
 168                 string outputFile;
 169
 170                 if (output == "lt") { //does the user want lower triangle phylip formatted file
 171                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "phylip.dist";
 172                         remove(outputFile.c_str());
 173
 174                         //output numSeqs to phylip formatted dist file
 175                 }else if (output == "column") { //user wants column format
 176                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "dist";
 177                         remove(outputFile.c_str());
 178                 }else { //assume square
 179                         outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "square.dist";
 180                         remove(outputFile.c_str());
 181                 }
 182
 183 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 184                 //if you don't need to fork anything
 185                 if(processors == 1){
 186                         driver(0, numSeqs, outputFile, cutoff);
 187                 }else{ //you have multiple processors
 188
 189                         for (int i = 0; i < processors; i++) {
 190                                 lines.push_back(new linePair());
 191                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 192                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 193                         }
 194
 195                         createProcesses(outputFile);
 196
 197                         map<int, int>::iterator it = processIDS.begin();
 198                         rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 199                         it++;
 200
 201                         //append and remove temp files
 202                         for (; it != processIDS.end(); it++) {
 203                                 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 204                                 remove((outputFile + toString(it->second) + ".temp").c_str());
 205                         }
 206                 }
 207 #else
 208                 ifstream inFASTA;
 209                 driver(0, numSeqs, outputFile, cutoff);
 210 #endif
 211
 212                 if (output == "square") {  convertMatrix(outputFile); }
 213
 214                 delete distCalculator;
 215
 216                 return 0;
 217
 218         }
 219         catch(exception& e) {
 220                 errorOut(e, "DistanceCommand", "execute");
 221                 exit(1);
 222         }
 223 }
 224 /**************************************************************************************************/
 225 void DistanceCommand::createProcesses(string filename) {
 226         try {
 227 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 228                 int process = 0;
 229                 processIDS.clear();
 230
 231                 //loop through and create all the processes you want
 232                 while (process != processors) {
 233                         int pid = fork();
 234
 235                         if (pid > 0) {
 236                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 237                                 process++;
 238                         }else if (pid == 0){
 239                                 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
 240                                 exit(0);
 241                         }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
 242                 }
 243
 244                 //force parent to wait until all the processes are done
 245                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 246                         int temp = it->second;
 247                         wait(&temp);
 248                 }
 249 #endif
 250         }
 251         catch(exception& e) {
 252                 errorOut(e, "DistanceCommand", "createProcesses");
 253                 exit(1);
 254         }
 255 }
 256
 257 /**************************************************************************************************/
 258 /////// need to fix to work with calcs and sequencedb
 259 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 260         try {
 261
 262                 int startTime = time(NULL);
 263
 264                 //column file
 265                 ofstream outFile(dFileName.c_str(), ios::trunc);
 266                 outFile.setf(ios::fixed, ios::showpoint);
 267                 outFile << setprecision(4);
 268
 269                 if((output == "lt") && startLine == 0){ outFile << alignDB.getNumSeqs() << endl;        }
 270
 271                 for(int i=startLine;i<endLine;i++){
 272                         if(output == "lt")      {
 273                                 string name = alignDB.get(i).getName();
 274                                 if (name.length() < 10) { //pad with spaces to make compatible
 275                                         while (name.length() < 10) {  name += " ";  }
 276                                 }
 277                                 outFile << name << '\t';
 278                         }
 279                         for(int j=0;j<i;j++){
 280                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 281                                 double dist = distCalculator->getDist();
 282
 283                                 if(dist <= cutoff){
 284                                         if (output == "column") { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 285                                 }
 286                                 if (output == "lt") {  outFile << dist << '\t'; }
 287
 288                                 if (output == "square") { //make a square column you can convert to square phylip
 289                                         outFile << alignDB.get(i).getName() << '\t' << alignDB.get(j).getName() << '\t' << dist << endl;
 290                                         outFile << alignDB.get(j).getName() << '\t' << alignDB.get(i).getName() << '\t' << dist << endl;
 291                                 }
 292
 293                         }
 294
 295                         if (output == "lt") { outFile << endl; }
 296
 297                         if(i % 100 == 0){
 298                                 mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); mothurOutEndLine();
 299                         }
 300
 301                 }
 302                 mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); mothurOutEndLine();
 303
 304                 outFile.close();
 305
 306                 return 1;
 307         }
 308         catch(exception& e) {
 309                 errorOut(e, "DistanceCommand", "driver");
 310                 exit(1);
 311         }
 312 }
 313 /**************************************************************************************************/
 314 void DistanceCommand::convertMatrix(string outputFile) {
 315         try{
 316
 317                 //sort file by first column so the distances for each row are together
 318                 string outfile = getRootName(outputFile) + "sorted.dist.temp";
 319
 320                 //use the unix sort
 321                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 322                         string command = "sort -n " + outputFile + " -o " + outfile;
 323                         system(command.c_str());
 324                 #else //sort using windows sort
 325                         string command = "sort " + outputFile + " /O " + outfile;
 326                         system(command.c_str());
 327                 #endif
 328
 329
 330                 //output to new file distance for each row and save positions in file where new row begins
 331                 ifstream in;
 332                 openInputFile(outfile, in);
 333
 334                 ofstream out;
 335                 openOutputFile(outputFile, out);
 336
 337                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 338
 339                 out << alignDB.getNumSeqs() << endl;
 340
 341                 //get first currentRow
 342                 string first, currentRow, second;
 343                 float dist;
 344                 map<string, float> rowDists; //take advantage of the fact that maps are already sorted by key
 345                 map<string, float>::iterator it;
 346
 347                 in >> first;
 348                 currentRow = first;
 349
 350                 rowDists[first] = 0.00; //distance to yourself is 0.0
 351
 352                 in.seekg(0);
 353                 //openInputFile(outfile, in);
 354
 355                 while(!in.eof()) {
 356                         in >> first >> second >> dist; gobble(in);
 357
 358                         if (first != currentRow) {
 359                                 //print out last row
 360                                 out << currentRow << '\t'; //print name
 361
 362                                 //print dists
 363                                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 364                                         out << it->second << '\t';
 365                                 }
 366                                 out << endl;
 367
 368                                 //start new row
 369                                 currentRow = first;
 370                                 rowDists.clear();
 371                                 rowDists[first] = 0.00;
 372                                 rowDists[second] = dist;
 373                         }else{
 374                                 rowDists[second] = dist;
 375                         }
 376                 }
 377                 //print out last row
 378                 out << currentRow << '\t'; //print name
 379
 380                 //print dists
 381                 for (it = rowDists.begin(); it != rowDists.end(); it++) {
 382                         out << it->second << '\t';
 383                 }
 384                 out << endl;
 385
 386                 in.close();
 387                 out.close();
 388
 389                 remove(outfile.c_str());
 390
 391         }
 392         catch(exception& e) {
 393                 errorOut(e, "DistanceCommand", "convertMatrix");
 394                 exit(1);
 395         }
 396 }
 397 /**************************************************************************************************
 398 void DistanceCommand::appendFiles(string temp, string filename) {
 399         try{
 400                 ofstream output;
 401                 ifstream input;
 402
 403                 //open output file in append mode
 404                 openOutputFileAppend(filename, output);
 405                 openInputFile(temp, input);
 406
 407                 while(char c = input.get()){
 408                         if(input.eof())         {       break;                  }
 409                         else                            {       output << c;    }
 410                 }
 411
 412                 input.close();
 413                 output.close();
 414         }
 415         catch(exception& e) {
 416                 errorOut(e, "DistanceCommand", "appendFiles");
 417                 exit(1);
 418         }
 419 }
 420 /**************************************************************************************************/
 421
 422