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