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                 //if the user has not entered specific groups to analyze then do them all
  27                 if (globaldata->Groups.size() == 0) {
  28                         numGroups = tmap->getNumGroups();
  29                 }else {
  30                         //check that groups are valid
  31                         for (int i = 0; i < globaldata->Groups.size(); i++) {
  32                                 if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
  33                                         cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
  34                                         // erase the invalid group from globaldata->Groups
  35                                         globaldata->Groups.erase (globaldata->Groups.begin()+i);
  36                                 }
  37                         }
  38
  39                         //if the user only entered invalid groups
  40                         if (globaldata->Groups.size() == 0) {
  41                                 numGroups = tmap->getNumGroups();
  42                                 cout << "When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile." << endl;
  43                         }else if (globaldata->Groups.size() == 1) {
  44                                 cout << "When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile." << endl;
  45                                 numGroups = tmap->getNumGroups();
  46                                 globaldata->Groups.clear();
  47                         }else { numGroups = globaldata->Groups.size(); }
  48                 }
  49
  50                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
  51                 numComp = 0;
  52                 int n = 1;
  53                 for (int i=1; i<numGroups; i++) {
  54                         numComp += i;
  55                         for (int l = n; l < numGroups; l++) {
  56                                 //set group comparison labels
  57                                 if (globaldata->Groups.size() != 0) {
  58                                         groupComb.push_back(globaldata->Groups[i-1]+globaldata->Groups[l]);
  59                                 }else {
  60                                         groupComb.push_back(tmap->namesOfGroups[i-1]+tmap->namesOfGroups[l]);
  61                                 }
  62                         }
  63                         n++;
  64                 }
  65
  66                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  67                 weighted = new Weighted(tmap);
  68
  69         }
  70         catch(exception& e) {
  71                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  72                 exit(1);
  73         }
  74         catch(...) {
  75                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  76                 exit(1);
  77         }
  78 }
  79 /***********************************************************/
  80 int UnifracWeightedCommand::execute() {
  81         try {
  82
  83                 //get weighted for users tree
  84                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  85                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  86                 uscoreFreq.resize(numComp);
  87                 validScores.resize(numComp);
  88                 totalrscoreFreq.resize(numComp);
  89                 uCumul.resize(numComp);
  90
  91                 //format output
  92                 outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
  93                 outDist << "Tree#" << '\t' << "Iter" << '\t' << "Groups"<< '\t' << "WScore" << endl;
  94
  95
  96                 //create new tree with same num nodes and leaves as users
  97                 randT = new Tree();
  98
  99                 //get pscores for users trees
 100                 for (int i = 0; i < T.size(); i++) {
 101                         rscoreFreq.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 102                         rCumul.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
 103
 104                         cout << "Processing tree " << i+1 << endl;
 105                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
 106
 107                         //save users score
 108                         for (int s=0; s<numComp; s++) {
 109                                 //update uscoreFreq
 110                                 it = uscoreFreq[s].find(userData[s]);
 111                                 if (it == uscoreFreq[s].end()) {//new score
 112                                         uscoreFreq[s][userData[s]] = 1;
 113                                 }else{ uscoreFreq[s][userData[s]]++; }
 114
 115                                 //add user score to valid scores
 116                                 validScores[s][userData[s]] = userData[s];
 117
 118                                 //save users tree score for summary file
 119                                 utreeScores.push_back(userData[s]);
 120                         }
 121
 122                         //copy T[i]'s info.
 123                         randT->getCopy(T[i]);
 124
 125                         //get scores for random trees
 126                         for (int j = 0; j < iters; j++) {
 127                                 //create a random tree with same topology as T[i], but different labels
 128                                 randT->assembleRandomUnifracTree();
 129                                 //get pscore of random tree
 130                                 randomData = weighted->getValues(randT);
 131
 132                                 //save ramdoms score
 133                                 for (int p=0; p<numComp; p++) {
 134                                         //add trees weighted score random score freq
 135                                         it2 = rscoreFreq[p].find(randomData[p]);
 136                                         if (it2 != rscoreFreq[p].end()) {//already have that score
 137                                                 rscoreFreq[p][randomData[p]]++;
 138                                         }else{//first time we have seen this score
 139                                                 rscoreFreq[p][randomData[p]] = 1;
 140                                         }
 141
 142                                         //add random score to valid scores
 143                                         validScores[p][randomData[p]] = randomData[p];
 144
 145                                         //output info to uwdistrib file
 146                                         outDist << i+1 << '\t' << '\t'<< j+1 << '\t' << '\t' << groupComb[p] << '\t'<< randomData[p] << endl;
 147                                 }
 148                         }
 149
 150                         saveRandomScores(); //save all random scores for weighted file
 151
 152                         //find the signifigance of the score for summary file
 153                         for (int t = 0; t < numComp; t++) {
 154                                 float rcumul = 1.0000;
 155                                 for (it = validScores[t].begin(); it != validScores[t].end(); it++) {
 156                                         //make rscoreFreq map and rCumul
 157                                         it2 = rscoreFreq[t].find(it->first);
 158                                         rCumul[t][it->first] = rcumul;
 159                                         //get percentage of random trees with that info
 160                                         if (it2 != rscoreFreq[t].end()) {  rscoreFreq[t][it->first] /= iters; rcumul-= it2->second;  }
 161                                         else { rscoreFreq[t][it->first] = 0.0000; } //no random trees with that score
 162                                 }
 163                         }
 164
 165                         //save the signifigance of the users score for printing later
 166                         for (int f = 0; f < numComp; f++) {
 167                                 WScoreSig.push_back(rCumul[f][userData[f]]);
 168                         }
 169
 170
 171                         //clear random data
 172                         rscoreFreq.clear();
 173                         rCumul.clear();
 174                 }
 175
 176                 rCumul.resize(numComp);
 177                 for (int b = 0; b < numComp; b++) {
 178                         float ucumul = 1.0000;
 179                         float rcumul = 1.0000;
 180                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 181                         for (it = validScores[b].begin(); it != validScores[b].end(); it++) {
 182                                 it2 = uscoreFreq[b].find(it->first);
 183                                 //make uCumul map
 184                                 uCumul[b][it->first] = ucumul;
 185                                 //user data has that score
 186                                 if (it2 != uscoreFreq[b].end()) { uscoreFreq[b][it->first] /= T.size(); ucumul-= it2->second;  }
 187                                 else { uscoreFreq[b][it->first] = 0.0000; } //no user trees with that score
 188
 189                                 //make rscoreFreq map and rCumul
 190                                 it2 = totalrscoreFreq[b].find(it->first);
 191                                 rCumul[b][it->first] = rcumul;
 192                                 //get percentage of random trees with that info
 193                                 if (it2 != totalrscoreFreq[b].end()) {  totalrscoreFreq[b][it->first] /= (iters * T.size()); rcumul-= it2->second;  }
 194                                 else { totalrscoreFreq[b][it->first] = 0.0000; } //no random trees with that score
 195                                                         }
 196                 }
 197
 198                 printWeightedFile();
 199                 printWSummaryFile();
 200
 201                 //reset randomTree parameter to 0
 202                 globaldata->setRandomTree("0");
 203                 //clear out users groups
 204                 globaldata->Groups.clear();
 205
 206                 delete randT;
 207
 208                 return 0;
 209
 210         }
 211         catch(exception& e) {
 212                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 213                 exit(1);
 214         }
 215         catch(...) {
 216                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 217                 exit(1);
 218         }
 219 }
 220 /***********************************************************/
 221 void UnifracWeightedCommand::printWeightedFile() {
 222         try {
 223                 //column headers
 224
 225                 out << "Group" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 226
 227                 //format output
 228                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 229
 230                 //for each group
 231                 for (int e = 0; e < numComp; e++) {
 232                         //print each line in that group
 233                         for (it = validScores[e].begin(); it != validScores[e].end(); it++) {
 234                                 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;
 235                         }
 236                 }
 237
 238                 out.close();
 239
 240         }
 241         catch(exception& e) {
 242                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 243                 exit(1);
 244         }
 245         catch(...) {
 246                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 247                 exit(1);
 248         }
 249 }
 250
 251
 252 /***********************************************************/
 253 void UnifracWeightedCommand::printWSummaryFile() {
 254         try {
 255                 //column headers
 256                 outSum << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 257
 258                 //format output
 259                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 260
 261                 //print each line
 262                 int count = 0;
 263                 for (int i = 0; i < T.size(); i++) {
 264                         for (int j = 0; j < numComp; j++) {
 265                                 outSum << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 266                                 count++;
 267                         }
 268                 }
 269                 outSum.close();
 270         }
 271         catch(exception& e) {
 272                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 273                 exit(1);
 274         }
 275         catch(...) {
 276                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 277                 exit(1);
 278         }
 279 }
 280
 281 /***********************************************************/
 282 void UnifracWeightedCommand::saveRandomScores() {
 283         try {
 284                 for (int e = 0; e < numComp; e++) {
 285                         //update total map with new random scores
 286                         for (it = rscoreFreq[e].begin(); it != rscoreFreq[e].end(); it++) {
 287                                 //does this score already exist in the total map
 288                                 it2 = totalrscoreFreq[e].find(it->first);
 289                                 //if yes then add them
 290                                 if (it2 != totalrscoreFreq[e].end()) {
 291                                         totalrscoreFreq[e][it->first] += rscoreFreq[e][it->first];
 292                                 }else{ //its a new score
 293                                         totalrscoreFreq[e][it->first] = rscoreFreq[e][it->first];
 294                                 }
 295                         }
 296                 }
 297         }
 298         catch(exception& e) {
 299                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function saveRandomScores. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 300                 exit(1);
 301         }
 302         catch(...) {
 303                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function saveRandomScores. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 304                 exit(1);
 305         }
 306 }
 307
 308 /***********************************************************/