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() {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16
  17                 T = globaldata->gTree;
  18                 tmap = globaldata->gTreemap;
  19                 sumFile = globaldata->getTreeFile() + ".uwsummary";
  20                 openOutputFile(sumFile, outSum);
  21
  22                 util = new SharedUtil();
  23                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  24                 util->getCombos(groupComb, globaldata->Groups, numComp);
  25                 globaldata->setGroups("");
  26
  27                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
  28
  29                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  30                 unweighted = new Unweighted(tmap);
  31
  32         }
  33         catch(exception& e) {
  34                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  35                 exit(1);
  36         }
  37         catch(...) {
  38                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  39                 exit(1);
  40         }
  41 }
  42 /***********************************************************/
  43 int UnifracUnweightedCommand::execute() {
  44         try {
  45
  46                 userData.resize(numComp,0);  //data[0] = unweightedscore
  47                 randomData.resize(numComp,0); //data[0] = unweightedscore
  48                 //create new tree with same num nodes and leaves as users
  49
  50                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
  51                 cout << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" << '\t' << "UWSig" <<  endl;
  52
  53                 //get pscores for users trees
  54                 for (int i = 0; i < T.size(); i++) {
  55                         counter = 0;
  56
  57                         output = new ColumnFile(globaldata->getTreeFile()  + toString(i+1) + ".unweighted");
  58
  59                         //get unweighted for users tree
  60                         rscoreFreq.resize(numComp);
  61                         rCumul.resize(numComp);
  62                         utreeScores.resize(numComp);
  63                         UWScoreSig.resize(numComp);
  64
  65                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
  66
  67                         //output scores for each combination
  68                         for(int k = 0; k < numComp; k++) {
  69                                 //saves users score
  70                                 utreeScores[k].push_back(userData[k]);
  71
  72                         }
  73
  74                         //get unweighted scores for random trees
  75                         for (int j = 0; j < iters; j++) {
  76                                 //we need a different getValues because when we swap the labels we only want to swap those in each parwise comparison
  77                                 randomData = unweighted->getValues(T[i], "", "");
  78
  79                                 for(int k = 0; k < numComp; k++) {
  80                                         //add trees unweighted score to map of scores
  81                                         it2 = rscoreFreq[k].find(randomData[k]);
  82                                         if (it2 != rscoreFreq[k].end()) {//already have that score
  83                                                 rscoreFreq[k][randomData[k]]++;
  84                                         }else{//first time we have seen this score
  85                                                 rscoreFreq[k][randomData[k]] = 1;
  86                                         }
  87
  88                                         //add randoms score to validscores
  89                                         validScores[randomData[k]] = randomData[k];
  90                                 }
  91
  92                         }
  93
  94                         for(int a = 0; a < numComp; a++) {
  95                                 float rcumul = 1.0000;
  96                                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
  97                                 for (it = validScores.begin(); it != validScores.end(); it++) {
  98                                         //make rscoreFreq map and rCumul
  99                                         it2 = rscoreFreq[a].find(it->first);
 100                                         rCumul[a][it->first] = rcumul;
 101                                         //get percentage of random trees with that info
 102                                         if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 103                                         else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 104                                 }
 105                                 UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 106                         }
 107
 108
 109
 110                         printUnweightedFile();
 111                         printUWSummaryFile(i);
 112
 113                         delete output;
 114                         rscoreFreq.clear();
 115                         rCumul.clear();
 116                         validScores.clear();
 117                         utreeScores.clear();
 118                         UWScoreSig.clear();
 119                 }
 120
 121                 //reset groups parameter
 122                 globaldata->Groups.clear();
 123                 outSum.close();
 124
 125                 return 0;
 126
 127         }
 128         catch(exception& e) {
 129                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 130                 exit(1);
 131         }
 132         catch(...) {
 133                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 134                 exit(1);
 135         }
 136 }
 137 /***********************************************************/
 138 void UnifracUnweightedCommand::printUnweightedFile() {
 139         try {
 140                 vector<double> data;
 141                 vector<string> tags;
 142                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 143
 144                 for(int a = 0; a < numComp; a++) {
 145                         output->initFile(groupComb[a], tags);
 146                         //print each line
 147                         for (it = validScores.begin(); it != validScores.end(); it++) {
 148                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 149                                 output->output(data);
 150                                 data.clear();
 151                         }
 152                         output->resetFile();
 153                 }
 154         }
 155         catch(exception& e) {
 156                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 157                 exit(1);
 158         }
 159         catch(...) {
 160                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 161                 exit(1);
 162         }
 163 }
 164
 165 /***********************************************************/
 166 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 167         try {
 168
 169                 //format output
 170                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 171
 172                 //print each line
 173
 174                 for(int a = 0; a < numComp; a++) {
 175                         outSum << i+1 << '\t';
 176                         cout << i+1 << '\t';
 177
 178                         if (UWScoreSig[a][0] > (1/(float)iters)) {
 179                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << UWScoreSig[a][0] << endl;
 180                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << UWScoreSig[a][0] << endl;
 181                         }else {
 182                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << "<" << (1/float(iters)) << endl;
 183                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << "<" << (1/float(iters)) << endl;
 184                         }
 185                 }
 186
 187         }
 188         catch(exception& e) {
 189                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 190                 exit(1);
 191         }
 192         catch(...) {
 193                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 194                 exit(1);
 195         }
 196 }
 197
 198 /***********************************************************/
 199
 200