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                 unweightedFile = globaldata->getTreeFile() + ".unweighted";
  20                 openOutputFile(unweightedFile, out);
  21                 //column headers
  22                 out << "Comb" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
  23
  24                 sumFile = globaldata->getTreeFile() + ".uwsummary";
  25                 openOutputFile(sumFile, outSum);
  26                 //column headers
  27                 outSum << "Tree#" << '\t' << "Comb" << '\t'  <<  "UWScore" << '\t' << '\t' << "UWSig" <<  endl;
  28
  29                 setGroups(); //sets users groups to analyze
  30                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  31                 unweighted = new Unweighted(tmap);
  32
  33         }
  34         catch(exception& e) {
  35                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  36                 exit(1);
  37         }
  38         catch(...) {
  39                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  40                 exit(1);
  41         }
  42 }
  43 /***********************************************************/
  44 int UnifracUnweightedCommand::execute() {
  45         try {
  46
  47                 userData.resize(numComp,0);  //data[0] = unweightedscore
  48                 randomData.resize(numComp,0); //data[0] = unweightedscore
  49                 //create new tree with same num nodes and leaves as users
  50                 randT = new Tree();
  51
  52                 //get pscores for users trees
  53                 for (int i = 0; i < T.size(); i++) {
  54                         //get unweighted for users tree
  55                         rscoreFreq.resize(numComp);
  56                         uscoreFreq.resize(numComp);
  57                         rCumul.resize(numComp);
  58                         uCumul.resize(numComp);
  59                         validScores.resize(numComp);
  60                         utreeScores.resize(numComp);
  61                         UWScoreSig.resize(numComp);
  62
  63                         cout << "Processing tree " << i+1 << endl;
  64                         outSum << "Tree#" << i+1 << endl;
  65                         out << "Tree#" << i+1 << endl;
  66                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
  67
  68                         //output scores for each combination
  69                         for(int k = 0; k < numComp; k++) {
  70                                 //update uscoreFreq
  71                                 it = uscoreFreq[k].find(userData[k]);
  72                                 if (it == uscoreFreq[k].end()) {//new score
  73                                         uscoreFreq[k][userData[k]] = 1;
  74                                 }else{ uscoreFreq[k][userData[k]]++; }
  75
  76                                 //add users score to valid scores
  77                                 validScores[k][userData[k]] = userData[k];
  78
  79                                 //saves users score
  80                                 utreeScores[k].push_back(userData[k]);
  81                         }
  82
  83                         //copy T[i]'s info.
  84                         randT->getCopy(T[i]);
  85
  86                         //get unweighted scores for random trees
  87                         for (int j = 0; j < iters; j++) {
  88                                 int count = 0;
  89                                 for (int r=0; r<numGroups; r++) {
  90                                         for (int l = r+1; l < numGroups; l++) {
  91                                                 //we need a different getValues because when we swap the labels we only want to swap those in each parwise comparison
  92                                                 randomData = unweighted->getValues(randT, "", "");
  93
  94                                                 //add trees unweighted score to map of scores
  95                                                 it2 = rscoreFreq[count].find(randomData[count]);
  96                                                 if (it2 != rscoreFreq[count].end()) {//already have that score
  97                                                         rscoreFreq[count][randomData[count]]++;
  98                                                 }else{//first time we have seen this score
  99                                                         rscoreFreq[count][randomData[count]] = 1;
 100                                                 }
 101
 102                                                 //add randoms score to validscores
 103                                                 validScores[count][randomData[count]] = randomData[count];
 104                                                 count++;
 105                                         }
 106                                 }
 107                         }
 108
 109                 for(int a = 0; a < numComp; a++) {
 110                         float ucumul = 1.0000;
 111                         float rcumul = 1.0000;
 112                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 113                         for (it = validScores[a].begin(); it != validScores[a].end(); it++) {
 114                                 it2 = uscoreFreq[a].find(it->first);
 115                                 //make uCumul map
 116                                 uCumul[a][it->first] = ucumul;
 117                                 //user data has that score
 118                                 if (it2 != uscoreFreq[a].end()) { uscoreFreq[a][it->first] /= T.size(); ucumul-= it2->second;  }
 119                                 else { uscoreFreq[a][it->first] = 0.0000; } //no user trees with that score
 120
 121                                 //make rscoreFreq map and rCumul
 122                                 it2 = rscoreFreq[a].find(it->first);
 123                                 rCumul[a][it->first] = rcumul;
 124                                 //get percentage of random trees with that info
 125                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 126                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 127                         }
 128                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 129                 }
 130
 131                 printUnweightedFile();
 132                 printUWSummaryFile();
 133
 134                 rscoreFreq.clear();
 135                 uscoreFreq.clear();
 136                 rCumul.clear();
 137                 uCumul.clear();
 138                 validScores.clear();
 139                 utreeScores.clear();
 140                 UWScoreSig.clear();
 141         }
 142                 //reset groups parameter
 143                 globaldata->Groups.clear(); globaldata->setGroups("");
 144
 145                 delete randT;
 146
 147                 return 0;
 148
 149         }
 150         catch(exception& e) {
 151                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 152                 exit(1);
 153         }
 154         catch(...) {
 155                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 156                 exit(1);
 157         }
 158 }
 159 /***********************************************************/
 160 void UnifracUnweightedCommand::printUnweightedFile() {
 161         try {
 162                 //format output
 163                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 164
 165                 for(int a = 0; a < numComp; a++) {
 166                         //print each line
 167                         for (it = validScores[a].begin(); it != validScores[a].end(); it++) {
 168                                 out << setprecision(6) << groupComb[a] << '\t' << it->first << '\t' << '\t' << uscoreFreq[a][it->first] << '\t' << uCumul[a][it->first] << '\t' << rscoreFreq[a][it->first] << '\t' << rCumul[a][it->first] << endl;
 169                         }
 170                 }
 171                 out.close();
 172
 173         }
 174         catch(exception& e) {
 175                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 176                 exit(1);
 177         }
 178         catch(...) {
 179                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 180                 exit(1);
 181         }
 182 }
 183
 184 /***********************************************************/
 185 void UnifracUnweightedCommand::printUWSummaryFile() {
 186         try {
 187
 188                 //format output
 189                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 190
 191                 //print each line
 192                 for (int i = 0; i< T.size(); i++) {
 193                         for(int a = 0; a < numComp; a++) {
 194                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << '\t' << utreeScores[a][i] << '\t' << UWScoreSig[a][i] << endl;
 195                                 cout << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << '\t' << utreeScores[a][i] << '\t' << UWScoreSig[a][i] << endl;
 196                         }
 197                 }
 198
 199                 outSum.close();
 200         }
 201         catch(exception& e) {
 202                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 203                 exit(1);
 204         }
 205         catch(...) {
 206                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 207                 exit(1);
 208         }
 209 }
 210 /***********************************************************/
 211
 212 void UnifracUnweightedCommand::setGroups() {
 213         try {
 214                 string allGroups = "";
 215                 numGroups = 0;
 216                 //if the user has not entered specific groups to analyze then do them all
 217                 if (globaldata->Groups.size() != 0) {
 218                         if (globaldata->Groups[0] != "all") {
 219                                 //check that groups are valid
 220                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 221                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 222                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 223                                                 // erase the invalid group from globaldata->Groups
 224                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 225                                         }
 226                                 }
 227
 228                                 //if the user only entered invalid groups
 229                                 if (globaldata->Groups.size() == 0) {
 230                                         cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl;
 231                                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 232                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 233                                                 numGroups++;
 234                                                 allGroups += tmap->namesOfGroups[i];
 235                                         }
 236                                 }else {
 237                                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 238                                                 allGroups += tmap->namesOfGroups[i];
 239                                                 numGroups++;
 240                                         }
 241                                 }
 242                         }else{//user has enter "all" and wants the default groups
 243                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 244                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 245                                         numGroups++;
 246                                         allGroups += tmap->namesOfGroups[i];
 247                                 }
 248                                 globaldata->setGroups("");
 249                         }
 250                 }else {
 251                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 252                                 allGroups += tmap->namesOfGroups[i];
 253                         }
 254                         numGroups = 1;
 255                 }
 256
 257                 //calculate number of comparsions
 258                 numComp = 0;
 259                 for (int r=0; r<numGroups; r++) {
 260                         for (int l = r+1; l < numGroups; l++) {
 261                                 groupComb.push_back(globaldata->Groups[r]+globaldata->Groups[l]);
 262                                 numComp++;
 263                         }
 264                 }
 265
 266                 //ABC
 267                 if (numComp != 1) {
 268                         groupComb.push_back(allGroups);
 269                         numComp++;
 270                 }
 271         }
 272         catch(exception& e) {
 273                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 274                 exit(1);
 275         }
 276         catch(...) {
 277                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 278                 exit(1);
 279         }
 280
 281 }
 282 /*****************************************************************/
 283