unifracweightedcommand.cpp

   1 /*
   2  *  unifracweightedcommand.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 "unifracweightedcommand.h"
  11
  12 /***********************************************************/
  13 UnifracWeightedCommand::UnifracWeightedCommand() {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16
  17                 T = globaldata->gTree;
  18                 tmap = globaldata->gTreemap;
  19                 weightedFile = globaldata->getTreeFile() + ".weighted";
  20                 openOutputFile(weightedFile, out);
  21                 sumFile = globaldata->getTreeFile() + ".wsummary";
  22                 openOutputFile(sumFile, outSum);
  23                 distFile = globaldata->getTreeFile() + ".wdistrib";
  24                 openOutputFile(distFile, outDist);
  25
  26                 numGroups = tmap->getNumGroups();
  27
  28                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
  29                 numComp = 0;
  30                 int n = 1;
  31                 for (int i=1; i<numGroups; i++) {
  32                         numComp += i;
  33                         for (int l = n; l < numGroups; l++) {
  34                                 //set group comparison labels
  35                                 groupComb.push_back(tmap->namesOfGroups[i-1]+tmap->namesOfGroups[l]);
  36                         }
  37                         n++;
  38                 }
  39
  40                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  41                 weighted = new Weighted(tmap);
  42
  43         }
  44         catch(exception& e) {
  45                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  46                 exit(1);
  47         }
  48         catch(...) {
  49                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  50                 exit(1);
  51         }
  52 }
  53 /***********************************************************/
  54 int UnifracWeightedCommand::execute() {
  55         try {
  56
  57                 //get weighted for users tree
  58                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  59                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  60                 uscoreFreq.resize(numComp);
  61                 validScores.resize(numComp);
  62                 totalrscoreFreq.resize(numComp);
  63                 uCumul.resize(numComp);
  64
  65                 //format output
  66                 outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
  67                 outDist << "Tree#" << '\t' << "Iter" << '\t' << "Groups"<< '\t' << "WScore" << endl;
  68
  69
  70                 //create new tree with same num nodes and leaves as users
  71                 randT = new Tree();
  72
  73                 //get pscores for users trees
  74                 for (int i = 0; i < T.size(); i++) {
  75                         rscoreFreq.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  76                         rCumul.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  77
  78                         cout << "Processing tree " << i+1 << endl;
  79                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
  80
  81                         //save users score
  82                         for (int s=0; s<numComp; s++) {
  83                                 //update uscoreFreq
  84                                 it = uscoreFreq[s].find(userData[s]);
  85                                 if (it == uscoreFreq[s].end()) {//new score
  86                                         uscoreFreq[s][userData[s]] = 1;
  87                                 }else{ uscoreFreq[s][userData[s]]++; }
  88
  89                                 //add user score to valid scores
  90                                 validScores[s][userData[s]] = userData[s];
  91
  92                                 //save users tree score for summary file
  93                                 utreeScores.push_back(userData[s]);
  94                         }
  95
  96                         //copy T[i]'s info.
  97                         randT->getCopy(T[i]);
  98
  99                         //get pscores for random trees
 100                         for (int j = 0; j < iters; j++) {
 101                                 //create a random tree with same topology as T[i], but different labels
 102                                 randT->assembleRandomUnifracTree();
 103                                 //get pscore of random tree
 104                                 randomData = weighted->getValues(randT);
 105
 106                                 //save ramdoms score
 107                                 for (int p=0; p<numComp; p++) {
 108                                         //add trees weighted score random score freq
 109                                         it2 = rscoreFreq[p].find(randomData[p]);
 110                                         if (it2 != rscoreFreq[p].end()) {//already have that score
 111                                                 rscoreFreq[p][randomData[p]]++;
 112                                         }else{//first time we have seen this score
 113                                                 rscoreFreq[p][randomData[p]] = 1;
 114                                         }
 115
 116                                         //add random score to valid scores
 117                                         validScores[p][randomData[p]] = randomData[p];
 118
 119                                         //output info to uwdistrib file
 120                                         outDist << i+1 << '\t' << '\t'<< j+1 << '\t' << '\t' << groupComb[p] << '\t'<< randomData[p] << endl;
 121                                 }
 122                         }
 123
 124                         saveRandomScores(); //save all random scores for weighted file
 125
 126                         //find the signifigance of the score for summary file
 127                         for (int t = 0; t < numComp; t++) {
 128                                 float rcumul = 0.0000;
 129                                 for (it = validScores[t].begin(); it != validScores[t].end(); it++) {
 130                                         //make rscoreFreq map and rCumul
 131                                         it2 = rscoreFreq[t].find(it->first);
 132                                         //get percentage of random trees with that info
 133                                         if (it2 != rscoreFreq[t].end()) {  rscoreFreq[t][it->first] /= iters; rcumul+= it2->second;  }
 134                                         else { rscoreFreq[t][it->first] = 0.0000; } //no random trees with that score
 135                                         rCumul[t][it->first] = rcumul;
 136                                 }
 137                         }
 138
 139                         //save the signifigance of the users score for printing later
 140                         for (int f = 0; f < numComp; f++) {
 141                                 WScoreSig.push_back(rCumul[f][userData[f]]);
 142                         }
 143
 144
 145                         //clear random data
 146                         rscoreFreq.clear();
 147                         rCumul.clear();
 148                 }
 149
 150                 rCumul.resize(numComp);
 151                 for (int b = 0; b < numComp; b++) {
 152                         float ucumul = 0.0000;
 153                         float rcumul = 0.0000;
 154                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 155                         for (it = validScores[b].begin(); it != validScores[b].end(); it++) {
 156                                 it2 = uscoreFreq[b].find(it->first);
 157                                 //user data has that score
 158                                 if (it2 != uscoreFreq[b].end()) { uscoreFreq[b][it->first] /= T.size(); ucumul+= it2->second;  }
 159                                 else { uscoreFreq[b][it->first] = 0.0000; } //no user trees with that score
 160                                 //make uCumul map
 161                                 uCumul[b][it->first] = ucumul;
 162
 163                                 //make rscoreFreq map and rCumul
 164                                 it2 = totalrscoreFreq[b].find(it->first);
 165                                 //get percentage of random trees with that info
 166                                 if (it2 != totalrscoreFreq[b].end()) {  totalrscoreFreq[b][it->first] /= (iters * T.size()); rcumul+= it2->second;  }
 167                                 else { totalrscoreFreq[b][it->first] = 0.0000; } //no random trees with that score
 168                                 rCumul[b][it->first] = rcumul;
 169                         }
 170                 }
 171
 172                 printWeightedFile();
 173                 printWSummaryFile();
 174
 175                 //reset randomTree parameter to 0
 176                 globaldata->setRandomTree("0");
 177
 178                 delete randT;
 179
 180                 return 0;
 181
 182         }
 183         catch(exception& e) {
 184                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187         catch(...) {
 188                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 189                 exit(1);
 190         }
 191 }
 192 /***********************************************************/
 193 void UnifracWeightedCommand::printWeightedFile() {
 194         try {
 195                 //column headers
 196
 197                 out << "Group" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 198
 199                 //format output
 200                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 201
 202                 //for each group
 203                 for (int e = 0; e < numComp; e++) {
 204                         //print each line in that group
 205                         for (it = validScores[e].begin(); it != validScores[e].end(); it++) {
 206                                 out << setprecision(6) <<  groupComb[e] << '\t' << it->first << '\t' << '\t' << uscoreFreq[e][it->first] << '\t' << uCumul[e][it->first] << '\t' << totalrscoreFreq[e][it->first] << '\t' << rCumul[e][it->first] << endl;
 207                         }
 208                 }
 209
 210                 out.close();
 211
 212         }
 213         catch(exception& e) {
 214                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 215                 exit(1);
 216         }
 217         catch(...) {
 218                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 219                 exit(1);
 220         }
 221 }
 222
 223
 224 /***********************************************************/
 225 void UnifracWeightedCommand::printWSummaryFile() {
 226         try {
 227                 //column headers
 228                 outSum << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 229
 230                 //format output
 231                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 232
 233                 //print each line
 234                 int count = 0;
 235                 for (int i = 0; i < T.size(); i++) {
 236                         for (int j = 0; j < numComp; j++) {
 237                                 outSum << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 238                                 count++;
 239                         }
 240                 }
 241                 outSum.close();
 242         }
 243         catch(exception& e) {
 244                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 245                 exit(1);
 246         }
 247         catch(...) {
 248                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 249                 exit(1);
 250         }
 251 }
 252
 253 /***********************************************************/
 254 void UnifracWeightedCommand::saveRandomScores() {
 255         try {
 256                 for (int e = 0; e < numComp; e++) {
 257                         //update total map with new random scores
 258                         for (it = rscoreFreq[e].begin(); it != rscoreFreq[e].end(); it++) {
 259                                 //does this score already exist in the total map
 260                                 it2 = totalrscoreFreq[e].find(it->first);
 261                                 //if yes then add them
 262                                 if (it2 != totalrscoreFreq[e].end()) {
 263                                         totalrscoreFreq[e][it->first] += rscoreFreq[e][it->first];
 264                                 }else{ //its a new score
 265                                         totalrscoreFreq[e][it->first] = rscoreFreq[e][it->first];
 266                                 }
 267                         }
 268                 }
 269         }
 270         catch(exception& e) {
 271                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function saveRandomScores. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 272                 exit(1);
 273         }
 274         catch(...) {
 275                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function saveRandomScores. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 276                 exit(1);
 277         }
 278 }
 279
 280 /***********************************************************/