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                 globaldata = GlobalData::getInstance();
  28                 abort = true;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["unweighted"] = tempOutNames;
  32                 outputTypes["uwsummary"] = tempOutNames;
  33                 outputTypes["phylip"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> UnifracUnweightedCommand::getRequiredParameters(){
  42         try {
  43                 vector<string> myArray;
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> UnifracUnweightedCommand::getRequiredFiles(){
  53         try {
  54                 string Array[] =  {"tree","group"};
  55                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  56
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 /***********************************************************/
  65 UnifracUnweightedCommand::UnifracUnweightedCommand(string option)  {
  66         try {
  67                 globaldata = GlobalData::getInstance();
  68                 abort = false;
  69                 Groups.clear();
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; }
  73
  74                 else {
  75                         //valid paramters for this command
  76                         string Array[] =  {"groups","iters","distance","random", "processors","outputdir","inputdir"};
  77                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  78
  79                         OptionParser parser(option);
  80                         map<string,string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         //initialize outputTypes
  90                         vector<string> tempOutNames;
  91                         outputTypes["unweighted"] = tempOutNames;
  92                         outputTypes["uwsummary"] = tempOutNames;
  93                         outputTypes["phylip"] = tempOutNames;
  94
  95                         if (globaldata->gTree.size() == 0) {//no trees were read
  96                                 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.unweighted command."); m->mothurOutEndLine(); abort = true;  }
  97
  98                         //if the user changes the output directory command factory will send this info to us in the output parameter
  99                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 100                                 outputDir = "";
 101                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 102                         }
 103
 104                         //check for optional parameter and set defaults
 105                         // ...at some point should added some additional type checking...
 106                         groups = validParameter.validFile(parameters, "groups", false);
 107                         if (groups == "not found") { groups = ""; }
 108                         else {
 109                                 m->splitAtDash(groups, Groups);
 110                                 globaldata->Groups = Groups;
 111                         }
 112
 113                         itersString = validParameter.validFile(parameters, "iters", false);                             if (itersString == "not found") { itersString = "1000"; }
 114                         convert(itersString, iters);
 115
 116                         string temp = validParameter.validFile(parameters, "distance", false);                  if (temp == "not found") { temp = "false"; }
 117                         phylip = m->isTrue(temp);
 118
 119                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "f"; }
 120                         random = m->isTrue(temp);
 121
 122                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 123                         convert(temp, processors);
 124
 125                         if (!random) {  iters = 0;  } //turn off random calcs
 126
 127                         //if user selects distance = true and no groups it won't calc the pairwise
 128                         if ((phylip) && (Groups.size() == 0)) {
 129                                 groups = "all";
 130                                 m->splitAtDash(groups, Groups);
 131                                 globaldata->Groups = Groups;
 132                         }
 133
 134                         if (abort == false) {
 135                                 T = globaldata->gTree;
 136                                 tmap = globaldata->gTreemap;
 137                                 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".uwsummary";
 138                                 outputNames.push_back(sumFile); outputTypes["uwsummary"].push_back(sumFile);
 139                                 m->openOutputFile(sumFile, outSum);
 140
 141                                 util = new SharedUtil();
 142                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
 143                                 util->getCombos(groupComb, globaldata->Groups, numComp);
 144
 145                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
 146
 147                                 unweighted = new Unweighted(tmap);
 148
 149                         }
 150
 151                 }
 152
 153         }
 154         catch(exception& e) {
 155                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
 156                 exit(1);
 157         }
 158 }
 159
 160 //**********************************************************************************************************************
 161
 162 void UnifracUnweightedCommand::help(){
 163         try {
 164                 m->mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
 165                 m->mothurOut("The unifrac.unweighted command parameters are groups, iters, distance, processors and random.  No parameters are required.\n");
 166                 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");
 167                 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");
 168                 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
 169                 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");
 170                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 171                 m->mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
 172                 m->mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
 173                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 174                 m->mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
 175                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 176         }
 177         catch(exception& e) {
 178                 m->errorOut(e, "UnifracUnweightedCommand", "help");
 179                 exit(1);
 180         }
 181 }
 182
 183
 184 /***********************************************************/
 185 int UnifracUnweightedCommand::execute() {
 186         try {
 187
 188                 if (abort == true) { return 0; }
 189
 190                 int start = time(NULL);
 191
 192                 userData.resize(numComp,0);  //data[0] = unweightedscore
 193                 randomData.resize(numComp,0); //data[0] = unweightedscore
 194                 //create new tree with same num nodes and leaves as users
 195
 196                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 197                 m->mothurOut("Tree#\tGroups\tUWScore\tUWSig"); m->mothurOutEndLine();
 198
 199                 //get pscores for users trees
 200                 for (int i = 0; i < T.size(); i++) {
 201                         if (m->control_pressed) {
 202                                 outSum.close();
 203                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 204                                 return 0;
 205                         }
 206
 207                         counter = 0;
 208
 209                         if (random)  {
 210                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);
 211                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 212                                 outputTypes["unweighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 213                         }
 214
 215
 216                         //get unweighted for users tree
 217                         rscoreFreq.resize(numComp);
 218                         rCumul.resize(numComp);
 219                         utreeScores.resize(numComp);
 220                         UWScoreSig.resize(numComp);
 221
 222                         userData = unweighted->getValues(T[i], processors, outputDir);  //userData[0] = unweightedscore
 223
 224                         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; }
 225
 226                         //output scores for each combination
 227                         for(int k = 0; k < numComp; k++) {
 228                                 //saves users score
 229                                 utreeScores[k].push_back(userData[k]);
 230
 231                                 //add users score to validscores
 232                                 validScores[userData[k]] = userData[k];
 233                         }
 234
 235                         //get unweighted scores for random trees - if random is false iters = 0
 236                         for (int j = 0; j < iters; j++) {
 237
 238                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 239                                 randomData = unweighted->getValues(T[i], "", "", processors, outputDir);
 240
 241                                 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; }
 242
 243                                 for(int k = 0; k < numComp; k++) {
 244                                         //add trees unweighted score to map of scores
 245                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 246                                         if (it != rscoreFreq[k].end()) {//already have that score
 247                                                 rscoreFreq[k][randomData[k]]++;
 248                                         }else{//first time we have seen this score
 249                                                 rscoreFreq[k][randomData[k]] = 1;
 250                                         }
 251
 252                                         //add randoms score to validscores
 253                                         validScores[randomData[k]] = randomData[k];
 254                                 }
 255
 256                                 //report progress
 257                                 m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 258                         }
 259
 260                         for(int a = 0; a < numComp; a++) {
 261                                 float rcumul = 1.0000;
 262
 263                                 if (random) {
 264                                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 265                                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 266                                                 //make rscoreFreq map and rCumul
 267                                                 map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 268                                                 rCumul[a][it->first] = rcumul;
 269                                                 //get percentage of random trees with that info
 270                                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 271                                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 272                                         }
 273                                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 274                                 }else           {       UWScoreSig[a].push_back(0.0);                                           }
 275
 276                         }
 277
 278                         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;  }
 279
 280                         //print output files
 281                         printUWSummaryFile(i);
 282                         if (random)  {  printUnweightedFile();  delete output;  }
 283                         if (phylip) {   createPhylipFile(i);            }
 284
 285                         rscoreFreq.clear();
 286                         rCumul.clear();
 287                         validScores.clear();
 288                         utreeScores.clear();
 289                         UWScoreSig.clear();
 290                 }
 291
 292
 293                 outSum.close();
 294
 295                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }      return 0; }
 296
 297                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.unweighted."); m->mothurOutEndLine();
 298
 299                 m->mothurOutEndLine();
 300                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 301                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 302                 m->mothurOutEndLine();
 303
 304                 return 0;
 305
 306         }
 307         catch(exception& e) {
 308                 m->errorOut(e, "UnifracUnweightedCommand", "execute");
 309                 exit(1);
 310         }
 311 }
 312 /***********************************************************/
 313 void UnifracUnweightedCommand::printUnweightedFile() {
 314         try {
 315                 vector<double> data;
 316                 vector<string> tags;
 317
 318                 tags.push_back("Score");
 319                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 320
 321                 for(int a = 0; a < numComp; a++) {
 322                         output->initFile(groupComb[a], tags);
 323                         //print each line
 324                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 325                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 326                                 output->output(data);
 327                                 data.clear();
 328                         }
 329                         output->resetFile();
 330                 }
 331         }
 332         catch(exception& e) {
 333                 m->errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 334                 exit(1);
 335         }
 336 }
 337
 338 /***********************************************************/
 339 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 340         try {
 341
 342                 //format output
 343                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 344
 345                 //print each line
 346
 347                 for(int a = 0; a < numComp; a++) {
 348                         outSum << i+1 << '\t';
 349                         m->mothurOut(toString(i+1) + "\t");
 350
 351                         if (random) {
 352                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 353                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 354                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 355                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])+ "\n");
 356                                 }else {
 357                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 358                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 359                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters))) + "\n");
 360                                 }
 361                         }else{
 362                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 363                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 364                                 m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00\n");
 365                         }
 366                 }
 367
 368         }
 369         catch(exception& e) {
 370                 m->errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 371                 exit(1);
 372         }
 373 }
 374 /***********************************************************/
 375 void UnifracUnweightedCommand::createPhylipFile(int i) {
 376         try {
 377                 string phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.dist";
 378                 outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
 379
 380                 ofstream out;
 381                 m->openOutputFile(phylipFileName, out);
 382
 383                 //output numSeqs
 384                 out << globaldata->Groups.size() << endl;
 385
 386                 //make matrix with scores in it
 387                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 388                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 389                         dists[i].resize(globaldata->Groups.size(), 0.0);
 390                 }
 391
 392                 //flip it so you can print it
 393                 int count = 0;
 394                 for (int r=0; r<globaldata->Groups.size(); r++) {
 395                         for (int l = r+1; l < globaldata->Groups.size(); l++) {
 396                                 dists[r][l] = utreeScores[count][0];
 397                                 dists[l][r] = utreeScores[count][0];
 398                                 count++;
 399                         }
 400                 }
 401
 402                 //output to file
 403                 for (int r=0; r<globaldata->Groups.size(); r++) {
 404                         //output name
 405                         string name = globaldata->Groups[r];
 406                         if (name.length() < 10) { //pad with spaces to make compatible
 407                                 while (name.length() < 10) {  name += " ";  }
 408                         }
 409                         out << name << '\t';
 410
 411                         //output distances
 412                         for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 413                         out << endl;
 414                 }
 415                 out.close();
 416         }
 417         catch(exception& e) {
 418                 m->errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 419                 exit(1);
 420         }
 421 }
 422 /***********************************************************/
 423
 424
 425