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                 //column headers
  22                 //out << "Group" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
  23
  24                 sumFile = globaldata->getTreeFile() + ".wsummary";
  25                 openOutputFile(sumFile, outSum);
  26
  27                 setGroups();    //sets the groups the user wants to analyze
  28                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  29                 weighted = new Weighted(tmap);
  30
  31         }
  32         catch(exception& e) {
  33                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  34                 exit(1);
  35         }
  36         catch(...) {
  37                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  38                 exit(1);
  39         }
  40 }
  41 /***********************************************************/
  42 int UnifracWeightedCommand::execute() {
  43         try {
  44
  45                 //get weighted for users tree
  46                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  47                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  48
  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                         rScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  55                         uScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  56                         validScores.resize(numComp);
  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                                 //add users score to vector of user scores
  64                                 uScores[s].push_back(userData[s]);
  65
  66                                 //add users score to vector of valid scores
  67                                 validScores[s].push_back(userData[s]);
  68
  69                                 //save users tree score for summary file
  70                                 utreeScores.push_back(userData[s]);
  71                         }
  72
  73                         //get scores for random trees
  74                         for (int j = 0; j < iters; j++) {
  75                                 int count = 0;
  76                                 for (int r=0; r<numGroups; r++) {
  77                                         for (int l = r+1; l < numGroups; l++) {
  78                                                 //copy T[i]'s info.
  79                                                 randT->getCopy(T[i]);
  80
  81                                                 //create a random tree with same topology as T[i], but different labels
  82                                                 randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
  83                                                 //get wscore of random tree
  84                                                 randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
  85
  86                                                 //save scores
  87                                                 rScores[count].push_back(randomData[0]);
  88                                                 validScores[count][randomData[0]] = randomData[0];
  89                                                 count++;
  90                                         }
  91                                 }
  92                         }
  93
  94                         removeValidScoresDuplicates();
  95                         //find the signifigance of the score for summary file
  96                         for (int f = 0; f < numComp; f++) {
  97                                 //sort random scores
  98                                 sort(rScores[f].begin(), rScores[f].end());
  99
 100                                 //the index of the score higher than yours is returned
 101                                 //so if you have 1000 random trees the index returned is 100
 102                                 //then there are 900 trees with a score greater then you.
 103                                 //giving you a signifigance of 0.900
 104                                 int index = findIndex(userData[f], f);    if (index == -1) { cout << "error in UnifracWeightedCommand" << endl; exit(1); } //error code
 105
 106                                 //the signifigance is the number of trees with the users score or higher
 107                                 WScoreSig.push_back((iters-index)/(float)iters);
 108                         }
 109
 110                         //out << "Tree# " << i << endl;
 111                         //printWeightedFile();
 112
 113                         //clear data
 114                         rScores.clear();
 115                         uScores.clear();
 116                         validScores.clear();
 117                 }
 118
 119                 printWSummaryFile();
 120
 121                 //clear out users groups
 122                 globaldata->Groups.clear();
 123
 124                 delete randT;
 125
 126                 return 0;
 127
 128         }
 129         catch(exception& e) {
 130                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 131                 exit(1);
 132         }
 133         catch(...) {
 134                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 135                 exit(1);
 136         }
 137 }
 138 /***********************************************************
 139 void UnifracWeightedCommand::printWeightedFile() {
 140         try {
 141
 142                 //format output
 143                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 144
 145                 //for each group
 146                 for (int e = 0; e < numComp; e++) {
 147                         //print each line in that group
 148                         for (i = 0; i < validScores[e].size(); i++) {
 149                                 out << setprecision(6) <<  groupComb[e] << '\t' << validScores[e][i] << '\t' << '\t' << uscoreFreq[e][it->first] << '\t' << uCumul[e][it->first] << '\t' << rscoreFreq[e][it->first] << '\t' << rCumul[e][it->first] << endl;
 150                         }
 151                 }
 152
 153                 out.close();
 154
 155         }
 156         catch(exception& e) {
 157                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 158                 exit(1);
 159         }
 160         catch(...) {
 161                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 162                 exit(1);
 163         }
 164 }
 165
 166
 167 /***********************************************************/
 168 void UnifracWeightedCommand::printWSummaryFile() {
 169         try {
 170                 //column headers
 171                 outSum << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 172                 cout << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 173
 174                 //format output
 175                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 176
 177                 //print each line
 178                 int count = 0;
 179                 for (int i = 0; i < T.size(); i++) {
 180                         for (int j = 0; j < numComp; j++) {
 181                                 outSum << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 182                                 cout << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 183                                 count++;
 184                         }
 185                 }
 186                 outSum.close();
 187         }
 188         catch(exception& e) {
 189                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. 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 UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 194                 exit(1);
 195         }
 196 }
 197
 198 /***********************************************************/
 199 void UnifracWeightedCommand::removeValidScoresDuplicates() {
 200         try {
 201                 for (int e = 0; e < numComp; e++) {
 202                         //sort valid scores
 203                         sort(validScores[e].begin(), validScores[e].end());
 204
 205                         for (int i = 0; i< validScores[e].size()-1; i++) {
 206                                 if (validScores[e][i] == validScores[e][i+1]) { validScores[e].erase(validScores[e].begin()+i); }
 207                         }
 208                 }
 209         }
 210         catch(exception& e) {
 211                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function removeValidScoresDuplicates. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 212                 exit(1);
 213         }
 214         catch(...) {
 215                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function removeValidScoresDuplicates. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 216                 exit(1);
 217         }
 218 }
 219
 220 /***********************************************************/
 221 int UnifracWeightedCommand::findIndex(float score, int index) {
 222         try{
 223                 for (int i = 0; i < rScores[index].size(); i++) {
 224                         if (rScores[index][i] >= score) {       return i;       }
 225                 }
 226                 return rScores[index].size();
 227         }
 228         catch(exception& e) {
 229                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 230                 exit(1);
 231         }
 232         catch(...) {
 233                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 234                 exit(1);
 235         }
 236 }
 237
 238 /***********************************************************/
 239 void UnifracWeightedCommand::setGroups() {
 240         try {
 241                 //if the user has not entered specific groups to analyze then do them all
 242                 if (globaldata->Groups.size() == 0) {
 243                         numGroups = tmap->getNumGroups();
 244                         for (int i=0; i < numGroups; i++) {
 245                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 246                         }
 247                 }else {
 248                         if (globaldata->getGroups() != "all") {
 249                                 //check that groups are valid
 250                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 251                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 252                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 253                                                 // erase the invalid group from globaldata->Groups
 254                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
 255                                         }
 256                                 }
 257
 258                                 //if the user only entered invalid groups
 259                                 if (globaldata->Groups.size() == 0) {
 260                                         numGroups = tmap->getNumGroups();
 261                                         for (int i=0; i < numGroups; i++) {
 262                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 263                                         }
 264                                         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;
 265                                 }else if (globaldata->Groups.size() == 1) {
 266                                         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;
 267                                         numGroups = tmap->getNumGroups();
 268                                         globaldata->Groups.clear();
 269                                         for (int i=0; i < numGroups; i++) {
 270                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 271                                         }
 272                                 }else { numGroups = globaldata->Groups.size(); }
 273                         }else { //users wants all groups
 274                                 numGroups = tmap->getNumGroups();
 275                                 globaldata->Groups.clear();
 276                                 globaldata->setGroups("");
 277                                 for (int i=0; i < numGroups; i++) {
 278                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 279                                 }
 280                         }
 281                 }
 282
 283                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 284                 numComp = 0;
 285                 for (int i=0; i<numGroups; i++) {
 286                         numComp += i;
 287                         for (int l = i+1; l < numGroups; l++) {
 288                                 //set group comparison labels
 289                                 groupComb.push_back(globaldata->Groups[i]+globaldata->Groups[l]);
 290                         }
 291                 }
 292         }
 293         catch(exception& e) {
 294                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 295                 exit(1);
 296         }
 297         catch(...) {
 298                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 299                 exit(1);
 300         }
 301 }
 302