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
  24                 setGroups();    //sets the groups the user wants to analyze
  25                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  26                 weighted = new Weighted(tmap);
  27
  28         }
  29         catch(exception& e) {
  30                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  31                 exit(1);
  32         }
  33         catch(...) {
  34                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  35                 exit(1);
  36         }
  37 }
  38 /***********************************************************/
  39 int UnifracWeightedCommand::execute() {
  40         try {
  41
  42                 //get weighted for users tree
  43                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  44                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  45                 uscoreFreq.resize(numComp);
  46                 validScores.resize(numComp);
  47                 totalrscoreFreq.resize(numComp);
  48                 uCumul.resize(numComp);
  49
  50                 //create new tree with same num nodes and leaves as users
  51                 randT = new Tree();
  52
  53                 //get pscores for users trees
  54                 for (int i = 0; i < T.size(); i++) {
  55                         rscoreFreq.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  56                         rCumul.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  57
  58                         cout << "Processing tree " << i+1 << endl;
  59                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
  60
  61                         //save users score
  62                         for (int s=0; s<numComp; s++) {
  63                                 //update uscoreFreq
  64                                 it = uscoreFreq[s].find(userData[s]);
  65                                 if (it == uscoreFreq[s].end()) {//new score
  66                                         uscoreFreq[s][userData[s]] = 1;
  67                                 }else{ uscoreFreq[s][userData[s]]++; }
  68
  69                                 //add user score to valid scores
  70                                 validScores[s][userData[s]] = userData[s];
  71
  72                                 //save users tree score for summary file
  73                                 utreeScores.push_back(userData[s]);
  74                         }
  75
  76                         //copy T[i]'s info.
  77                         randT->getCopy(T[i]);
  78
  79                         //get scores for random trees
  80                         for (int j = 0; j < iters; j++) {
  81                                 //create a random tree with same topology as T[i], but different labels
  82                                 randT->assembleRandomUnifracTree();
  83                                 //get pscore of random tree
  84                                 randomData = weighted->getValues(randT);
  85
  86                                 //save ramdoms score
  87                                 for (int p=0; p<numComp; p++) {
  88                                         //add trees weighted score random score freq
  89                                         it2 = rscoreFreq[p].find(randomData[p]);
  90                                         if (it2 != rscoreFreq[p].end()) {//already have that score
  91                                                 rscoreFreq[p][randomData[p]]++;
  92                                         }else{//first time we have seen this score
  93                                                 rscoreFreq[p][randomData[p]] = 1;
  94                                         }
  95
  96                                         //add random score to valid scores
  97                                         validScores[p][randomData[p]] = randomData[p];
  98                                 }
  99                         }
 100
 101                         saveRandomScores(); //save all random scores for weighted file
 102
 103                         //find the signifigance of the score for summary file
 104                         for (int t = 0; t < numComp; t++) {
 105                                 float rcumul = 1.0000;
 106                                 for (it = validScores[t].begin(); it != validScores[t].end(); it++) {
 107                                         //make rscoreFreq map and rCumul
 108                                         it2 = rscoreFreq[t].find(it->first);
 109                                         rCumul[t][it->first] = rcumul;
 110                                         //get percentage of random trees with that info
 111                                         if (it2 != rscoreFreq[t].end()) {  rscoreFreq[t][it->first] /= iters; rcumul-= it2->second;  }
 112                                         else { rscoreFreq[t][it->first] = 0.0000; } //no random trees with that score
 113                                 }
 114                         }
 115
 116                         //save the signifigance of the users score for printing later
 117                         for (int f = 0; f < numComp; f++) {
 118                                 WScoreSig.push_back(rCumul[f][userData[f]]);
 119                         }
 120
 121
 122                         //clear random data
 123                         rscoreFreq.clear();
 124                         rCumul.clear();
 125                 }
 126
 127                 rCumul.resize(numComp);
 128                 for (int b = 0; b < numComp; b++) {
 129                         float ucumul = 1.0000;
 130                         float rcumul = 1.0000;
 131                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 132                         for (it = validScores[b].begin(); it != validScores[b].end(); it++) {
 133                                 it2 = uscoreFreq[b].find(it->first);
 134                                 //make uCumul map
 135                                 uCumul[b][it->first] = ucumul;
 136                                 //user data has that score
 137                                 if (it2 != uscoreFreq[b].end()) { uscoreFreq[b][it->first] /= T.size(); ucumul-= it2->second;  }
 138                                 else { uscoreFreq[b][it->first] = 0.0000; } //no user trees with that score
 139
 140                                 //make rscoreFreq map and rCumul
 141                                 it2 = totalrscoreFreq[b].find(it->first);
 142                                 rCumul[b][it->first] = rcumul;
 143                                 //get percentage of random trees with that info
 144                                 if (it2 != totalrscoreFreq[b].end()) {  totalrscoreFreq[b][it->first] /= (iters * T.size()); rcumul-= it2->second;  }
 145                                 else { totalrscoreFreq[b][it->first] = 0.0000; } //no random trees with that score
 146                                                         }
 147                 }
 148
 149                 printWeightedFile();
 150                 printWSummaryFile();
 151
 152                 //clear out users groups
 153                 globaldata->Groups.clear();
 154
 155                 delete randT;
 156
 157                 return 0;
 158
 159         }
 160         catch(exception& e) {
 161                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 162                 exit(1);
 163         }
 164         catch(...) {
 165                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 166                 exit(1);
 167         }
 168 }
 169 /***********************************************************/
 170 void UnifracWeightedCommand::printWeightedFile() {
 171         try {
 172                 //column headers
 173
 174                 out << "Group" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 175
 176                 //format output
 177                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 178
 179                 //for each group
 180                 for (int e = 0; e < numComp; e++) {
 181                         //print each line in that group
 182                         for (it = validScores[e].begin(); it != validScores[e].end(); it++) {
 183                                 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;
 184                         }
 185                 }
 186
 187                 out.close();
 188
 189         }
 190         catch(exception& e) {
 191                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 192                 exit(1);
 193         }
 194         catch(...) {
 195                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 196                 exit(1);
 197         }
 198 }
 199
 200
 201 /***********************************************************/
 202 void UnifracWeightedCommand::printWSummaryFile() {
 203         try {
 204                 //column headers
 205                 outSum << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 206                 cout << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 207
 208                 //format output
 209                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 210
 211                 //print each line
 212                 int count = 0;
 213                 for (int i = 0; i < T.size(); i++) {
 214                         for (int j = 0; j < numComp; j++) {
 215                                 outSum << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 216                                 cout << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 217                                 count++;
 218                         }
 219                 }
 220                 outSum.close();
 221         }
 222         catch(exception& e) {
 223                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 224                 exit(1);
 225         }
 226         catch(...) {
 227                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 228                 exit(1);
 229         }
 230 }
 231
 232 /***********************************************************/
 233 void UnifracWeightedCommand::saveRandomScores() {
 234         try {
 235                 for (int e = 0; e < numComp; e++) {
 236                         //update total map with new random scores
 237                         for (it = rscoreFreq[e].begin(); it != rscoreFreq[e].end(); it++) {
 238                                 //does this score already exist in the total map
 239                                 it2 = totalrscoreFreq[e].find(it->first);
 240                                 //if yes then add them
 241                                 if (it2 != totalrscoreFreq[e].end()) {
 242                                         totalrscoreFreq[e][it->first] += rscoreFreq[e][it->first];
 243                                 }else{ //its a new score
 244                                         totalrscoreFreq[e][it->first] = rscoreFreq[e][it->first];
 245                                 }
 246                         }
 247                 }
 248         }
 249         catch(exception& e) {
 250                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function saveRandomScores. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 251                 exit(1);
 252         }
 253         catch(...) {
 254                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function saveRandomScores. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 255                 exit(1);
 256         }
 257 }
 258
 259 /***********************************************************/
 260 void UnifracWeightedCommand::setGroups() {
 261         try {
 262                 //if the user has not entered specific groups to analyze then do them all
 263                 if (globaldata->Groups.size() == 0) {
 264                         numGroups = tmap->getNumGroups();
 265                 }else {
 266                         if (globaldata->getGroups() != "all") {
 267                                 //check that groups are valid
 268                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 269                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 270                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 271                                                 // erase the invalid group from globaldata->Groups
 272                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
 273                                         }
 274                                 }
 275
 276                                 //if the user only entered invalid groups
 277                                 if (globaldata->Groups.size() == 0) {
 278                                         numGroups = tmap->getNumGroups();
 279                                         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;
 280                                 }else if (globaldata->Groups.size() == 1) {
 281                                         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;
 282                                         numGroups = tmap->getNumGroups();
 283                                         globaldata->Groups.clear();
 284                                 }else { numGroups = globaldata->Groups.size(); }
 285                         }else { //users wants all groups
 286                                 numGroups = tmap->getNumGroups();
 287                                 globaldata->Groups.clear();
 288                                 globaldata->setGroups("");
 289                         }
 290                 }
 291
 292                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 293                 numComp = 0;
 294                 int n = 1;
 295                 for (int i=1; i<numGroups; i++) {
 296                         numComp += i;
 297                         for (int l = n; l < numGroups; l++) {
 298                                 //set group comparison labels
 299                                 if (globaldata->Groups.size() != 0) {
 300                                         groupComb.push_back(globaldata->Groups[i-1]+globaldata->Groups[l]);
 301                                 }else {
 302                                         groupComb.push_back(tmap->namesOfGroups[i-1]+tmap->namesOfGroups[l]);
 303                                 }
 304                         }
 305                         n++;
 306                 }
 307         }
 308         catch(exception& e) {
 309                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 310                 exit(1);
 311         }
 312         catch(...) {
 313                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 314                 exit(1);
 315         }
 316 }
 317