unifracunweightedcommand.cpp

   1 /*
   2  *  unifracunweightedcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 2/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "unifracunweightedcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> UnifracUnweightedCommand::getValidParameters(){
  14         try {
  15                 string Array[] =  {"groups","iters","distance","random", "processors","outputdir","inputdir"};
  16                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "UnifracUnweightedCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 UnifracUnweightedCommand::UnifracUnweightedCommand(){
  26         try {
  27                 //initialize outputTypes
  28                 vector<string> tempOutNames;
  29                 outputTypes["unweighted"] = tempOutNames;
  30                 outputTypes["uwsummary"] = tempOutNames;
  31                 outputTypes["phylip"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> UnifracUnweightedCommand::getRequiredParameters(){
  40         try {
  41                 vector<string> myArray;
  42                 return myArray;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredParameters");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 vector<string> UnifracUnweightedCommand::getRequiredFiles(){
  51         try {
  52                 string Array[] =  {"tree","group"};
  53                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  54
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 /***********************************************************/
  63 UnifracUnweightedCommand::UnifracUnweightedCommand(string option)  {
  64         try {
  65                 globaldata = GlobalData::getInstance();
  66                 abort = false;
  67                 Groups.clear();
  68
  69                 //allow user to run help
  70                 if(option == "help") { help(); abort = true; }
  71
  72                 else {
  73                         //valid paramters for this command
  74                         string Array[] =  {"groups","iters","distance","random", "processors","outputdir","inputdir"};
  75                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  76
  77                         OptionParser parser(option);
  78                         map<string,string> parameters = parser.getParameters();
  79
  80                         ValidParameters validParameter;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["unweighted"] = tempOutNames;
  90                         outputTypes["uwsummary"] = tempOutNames;
  91                         outputTypes["phylip"] = tempOutNames;
  92
  93                         if (globaldata->gTree.size() == 0) {//no trees were read
  94                                 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.unweighted command."); m->mothurOutEndLine(); abort = true;  }
  95
  96                         //if the user changes the output directory command factory will send this info to us in the output parameter
  97                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  98                                 outputDir = "";
  99                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 100                         }
 101
 102                         //check for optional parameter and set defaults
 103                         // ...at some point should added some additional type checking...
 104                         groups = validParameter.validFile(parameters, "groups", false);
 105                         if (groups == "not found") { groups = ""; }
 106                         else {
 107                                 m->splitAtDash(groups, Groups);
 108                                 globaldata->Groups = Groups;
 109                         }
 110
 111                         itersString = validParameter.validFile(parameters, "iters", false);                             if (itersString == "not found") { itersString = "1000"; }
 112                         convert(itersString, iters);
 113
 114                         string temp = validParameter.validFile(parameters, "distance", false);                  if (temp == "not found") { temp = "false"; }
 115                         phylip = m->isTrue(temp);
 116
 117                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "f"; }
 118                         random = m->isTrue(temp);
 119
 120                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 121                         convert(temp, processors);
 122
 123                         if (!random) {  iters = 0;  } //turn off random calcs
 124
 125                         //if user selects distance = true and no groups it won't calc the pairwise
 126                         if ((phylip) && (Groups.size() == 0)) {
 127                                 groups = "all";
 128                                 m->splitAtDash(groups, Groups);
 129                                 globaldata->Groups = Groups;
 130                         }
 131
 132                         if (abort == false) {
 133                                 T = globaldata->gTree;
 134                                 tmap = globaldata->gTreemap;
 135                                 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".uwsummary";
 136                                 outputNames.push_back(sumFile); outputTypes["uwsummary"].push_back(sumFile);
 137                                 m->openOutputFile(sumFile, outSum);
 138
 139                                 util = new SharedUtil();
 140                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
 141                                 util->getCombos(groupComb, globaldata->Groups, numComp);
 142
 143                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
 144
 145                                 unweighted = new Unweighted(tmap);
 146
 147                         }
 148
 149                 }
 150
 151         }
 152         catch(exception& e) {
 153                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
 154                 exit(1);
 155         }
 156 }
 157
 158 //**********************************************************************************************************************
 159
 160 void UnifracUnweightedCommand::help(){
 161         try {
 162                 m->mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
 163                 m->mothurOut("The unifrac.unweighted command parameters are groups, iters, distance, processors and random.  No parameters are required.\n");
 164                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 1 valid group.\n");
 165                 m->mothurOut("The group names are separated by dashes.  The iters parameter allows you to specify how many random trees you would like compared to your tree.\n");
 166                 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
 167                 m->mothurOut("The random parameter allows you to shut off the comparison to random trees. The default is false, meaning compare don't your trees with randomly generated trees.\n");
 168                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 169                 m->mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
 170                 m->mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
 171                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 172                 m->mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
 173                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 174         }
 175         catch(exception& e) {
 176                 m->errorOut(e, "UnifracUnweightedCommand", "help");
 177                 exit(1);
 178         }
 179 }
 180
 181
 182 /***********************************************************/
 183 int UnifracUnweightedCommand::execute() {
 184         try {
 185
 186                 if (abort == true) { return 0; }
 187
 188                 int start = time(NULL);
 189
 190                 userData.resize(numComp,0);  //data[0] = unweightedscore
 191                 randomData.resize(numComp,0); //data[0] = unweightedscore
 192                 //create new tree with same num nodes and leaves as users
 193
 194                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 195                 m->mothurOut("Tree#\tGroups\tUWScore\tUWSig"); m->mothurOutEndLine();
 196
 197                 //get pscores for users trees
 198                 for (int i = 0; i < T.size(); i++) {
 199                         if (m->control_pressed) {
 200                                 outSum.close();
 201                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 202                                 return 0;
 203                         }
 204
 205                         counter = 0;
 206
 207                         if (random)  {
 208                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);
 209                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 210                                 outputTypes["unweighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 211                         }
 212
 213
 214                         //get unweighted for users tree
 215                         rscoreFreq.resize(numComp);
 216                         rCumul.resize(numComp);
 217                         utreeScores.resize(numComp);
 218                         UWScoreSig.resize(numComp);
 219
 220                         userData = unweighted->getValues(T[i], processors, outputDir);  //userData[0] = unweightedscore
 221
 222                         if (m->control_pressed) { if (random) { delete output;  } outSum.close();  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  }return 0; }
 223
 224                         //output scores for each combination
 225                         for(int k = 0; k < numComp; k++) {
 226                                 //saves users score
 227                                 utreeScores[k].push_back(userData[k]);
 228
 229                                 //add users score to validscores
 230                                 validScores[userData[k]] = userData[k];
 231                         }
 232
 233                         //get unweighted scores for random trees - if random is false iters = 0
 234                         for (int j = 0; j < iters; j++) {
 235
 236                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 237                                 randomData = unweighted->getValues(T[i], "", "", processors, outputDir);
 238
 239                                 if (m->control_pressed) { if (random) { delete output;  } outSum.close(); for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 240
 241                                 for(int k = 0; k < numComp; k++) {
 242                                         //add trees unweighted score to map of scores
 243                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 244                                         if (it != rscoreFreq[k].end()) {//already have that score
 245                                                 rscoreFreq[k][randomData[k]]++;
 246                                         }else{//first time we have seen this score
 247                                                 rscoreFreq[k][randomData[k]] = 1;
 248                                         }
 249
 250                                         //add randoms score to validscores
 251                                         validScores[randomData[k]] = randomData[k];
 252                                 }
 253
 254                                 //report progress
 255                                 m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 256                         }
 257
 258                         for(int a = 0; a < numComp; a++) {
 259                                 float rcumul = 1.0000;
 260
 261                                 if (random) {
 262                                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 263                                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 264                                                 //make rscoreFreq map and rCumul
 265                                                 map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 266                                                 rCumul[a][it->first] = rcumul;
 267                                                 //get percentage of random trees with that info
 268                                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 269                                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 270                                         }
 271                                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 272                                 }else           {       UWScoreSig[a].push_back(0.0);                                           }
 273
 274                         }
 275
 276                         if (m->control_pressed) { if (random) { delete output;  } outSum.close(); for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0;  }
 277
 278                         //print output files
 279                         printUWSummaryFile(i);
 280                         if (random)  {  printUnweightedFile();  delete output;  }
 281                         if (phylip) {   createPhylipFile(i);            }
 282
 283                         rscoreFreq.clear();
 284                         rCumul.clear();
 285                         validScores.clear();
 286                         utreeScores.clear();
 287                         UWScoreSig.clear();
 288                 }
 289
 290
 291                 outSum.close();
 292
 293                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }      return 0; }
 294
 295                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.unweighted."); m->mothurOutEndLine();
 296
 297                 m->mothurOutEndLine();
 298                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 299                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 300                 m->mothurOutEndLine();
 301
 302                 return 0;
 303
 304         }
 305         catch(exception& e) {
 306                 m->errorOut(e, "UnifracUnweightedCommand", "execute");
 307                 exit(1);
 308         }
 309 }
 310 /***********************************************************/
 311 void UnifracUnweightedCommand::printUnweightedFile() {
 312         try {
 313                 vector<double> data;
 314                 vector<string> tags;
 315
 316                 tags.push_back("Score");
 317                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 318
 319                 for(int a = 0; a < numComp; a++) {
 320                         output->initFile(groupComb[a], tags);
 321                         //print each line
 322                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 323                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 324                                 output->output(data);
 325                                 data.clear();
 326                         }
 327                         output->resetFile();
 328                 }
 329         }
 330         catch(exception& e) {
 331                 m->errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 332                 exit(1);
 333         }
 334 }
 335
 336 /***********************************************************/
 337 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 338         try {
 339
 340                 //format output
 341                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 342
 343                 //print each line
 344
 345                 for(int a = 0; a < numComp; a++) {
 346                         outSum << i+1 << '\t';
 347                         m->mothurOut(toString(i+1) + "\t");
 348
 349                         if (random) {
 350                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 351                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 352                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 353                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])+ "\n");
 354                                 }else {
 355                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 356                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 357                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters))) + "\n");
 358                                 }
 359                         }else{
 360                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 361                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 362                                 m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00\n");
 363                         }
 364                 }
 365
 366         }
 367         catch(exception& e) {
 368                 m->errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 369                 exit(1);
 370         }
 371 }
 372 /***********************************************************/
 373 void UnifracUnweightedCommand::createPhylipFile(int i) {
 374         try {
 375                 string phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.dist";
 376                 outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
 377
 378                 ofstream out;
 379                 m->openOutputFile(phylipFileName, out);
 380
 381                 //output numSeqs
 382                 out << globaldata->Groups.size() << endl;
 383
 384                 //make matrix with scores in it
 385                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 386                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 387                         dists[i].resize(globaldata->Groups.size(), 0.0);
 388                 }
 389
 390                 //flip it so you can print it
 391                 int count = 0;
 392                 for (int r=0; r<globaldata->Groups.size(); r++) {
 393                         for (int l = r+1; l < globaldata->Groups.size(); l++) {
 394                                 dists[r][l] = utreeScores[count][0];
 395                                 dists[l][r] = utreeScores[count][0];
 396                                 count++;
 397                         }
 398                 }
 399
 400                 //output to file
 401                 for (int r=0; r<globaldata->Groups.size(); r++) {
 402                         //output name
 403                         string name = globaldata->Groups[r];
 404                         if (name.length() < 10) { //pad with spaces to make compatible
 405                                 while (name.length() < 10) {  name += " ";  }
 406                         }
 407                         out << name << '\t';
 408
 409                         //output distances
 410                         for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 411                         out << endl;
 412                 }
 413                 out.close();
 414         }
 415         catch(exception& e) {
 416                 m->errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 417                 exit(1);
 418         }
 419 }
 420 /***********************************************************/
 421
 422
 423