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                                                 if (globaldata->Groups.size() != 0) {
  82                                                         //create a random tree with same topology as T[i], but different labels
  83                                                         randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
  84                                                         //get wscore of random tree
  85                                                         randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
  86                                                 }else {
  87                                                         //create a random tree with same topology as T[i], but different labels
  88                                                         randT->assembleRandomUnifracTree(tmap->namesOfGroups[r], tmap->namesOfGroups[l]);
  89                                                         //get wscore of random tree
  90                                                         randomData = weighted->getValues(randT, tmap->namesOfGroups[r], tmap->namesOfGroups[l]);
  91                                                 }
  92                                                 //save scores
  93                                                 rScores[count].push_back(randomData[0]);
  94                                                 validScores[count][randomData[0]] = randomData[0];
  95                                                 count++;
  96                                         }
  97                                 }
  98                         }
  99
 100                         removeValidScoresDuplicates();
 101                         //find the signifigance of the score for summary file
 102                         for (int f = 0; f < numComp; f++) {
 103                                 //sort random scores
 104                                 sort(rScores[f].begin(), rScores[f].end());
 105
 106                                 //the index of the score higher than yours is returned
 107                                 //so if you have 1000 random trees the index returned is 100
 108                                 //then there are 900 trees with a score greater then you.
 109                                 //giving you a signifigance of 0.900
 110                                 int index = findIndex(userData[f], f);    if (index == -1) { cout << "error in UnifracWeightedCommand" << endl; exit(1); } //error code
 111
 112                                 //the signifigance is the number of trees with the users score or higher
 113                                 WScoreSig.push_back((iters-index)/(float)iters);
 114                         }
 115
 116                         out << "Tree# " << i << endl;
 117                         //printWeightedFile();
 118
 119                         //clear data
 120                         rScores.clear();
 121                         uScores.clear();
 122                         validScores.clear();
 123                 }
 124
 125                 printWSummaryFile();
 126
 127                 //clear out users groups
 128                 globaldata->Groups.clear();
 129
 130                 delete randT;
 131
 132                 return 0;
 133
 134         }
 135         catch(exception& e) {
 136                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 137                 exit(1);
 138         }
 139         catch(...) {
 140                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 141                 exit(1);
 142         }
 143 }
 144 /***********************************************************
 145 void UnifracWeightedCommand::printWeightedFile() {
 146         try {
 147
 148                 //format output
 149                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 150
 151                 //for each group
 152                 for (int e = 0; e < numComp; e++) {
 153                         //print each line in that group
 154                         for (i = 0; i < validScores[e].size(); i++) {
 155                                 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;
 156                         }
 157                 }
 158
 159                 out.close();
 160
 161         }
 162         catch(exception& e) {
 163                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 164                 exit(1);
 165         }
 166         catch(...) {
 167                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 168                 exit(1);
 169         }
 170 }
 171
 172
 173 /***********************************************************/
 174 void UnifracWeightedCommand::printWSummaryFile() {
 175         try {
 176                 //column headers
 177                 outSum << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 178                 cout << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 179
 180                 //format output
 181                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 182
 183                 //print each line
 184                 int count = 0;
 185                 for (int i = 0; i < T.size(); i++) {
 186                         for (int j = 0; j < numComp; j++) {
 187                                 outSum << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 188                                 cout << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 189                                 count++;
 190                         }
 191                 }
 192                 outSum.close();
 193         }
 194         catch(exception& e) {
 195                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 196                 exit(1);
 197         }
 198         catch(...) {
 199                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 200                 exit(1);
 201         }
 202 }
 203
 204 /***********************************************************/
 205 void UnifracWeightedCommand::removeValidScoresDuplicates() {
 206         try {
 207                 for (int e = 0; e < numComp; e++) {
 208                         //sort valid scores
 209                         sort(validScores[e].begin(), validScores[e].end());
 210
 211                         for (int i = 0; i< validScores[e].size()-1; i++) {
 212                                 if (validScores[e][i] == validScores[e][i+1]) { validScores[e].erase(validScores[e].begin()+i); }
 213                         }
 214                 }
 215         }
 216         catch(exception& e) {
 217                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function removeValidScoresDuplicates. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 218                 exit(1);
 219         }
 220         catch(...) {
 221                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function removeValidScoresDuplicates. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 222                 exit(1);
 223         }
 224 }
 225
 226 /***********************************************************/
 227 int UnifracWeightedCommand::findIndex(float score, int index) {
 228         try{
 229                 for (int i = 0; i < rScores[index].size(); i++) {
 230                         if (rScores[index][i] >= score) {       return i;       }
 231                 }
 232                 return rScores[index].size();
 233         }
 234         catch(exception& e) {
 235                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 236                 exit(1);
 237         }
 238         catch(...) {
 239                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 240                 exit(1);
 241         }
 242 }
 243
 244 /***********************************************************/
 245 void UnifracWeightedCommand::setGroups() {
 246         try {
 247                 //if the user has not entered specific groups to analyze then do them all
 248                 if (globaldata->Groups.size() == 0) {
 249                         numGroups = tmap->getNumGroups();
 250                 }else {
 251                         if (globaldata->getGroups() != "all") {
 252                                 //check that groups are valid
 253                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 254                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 255                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 256                                                 // erase the invalid group from globaldata->Groups
 257                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
 258                                         }
 259                                 }
 260
 261                                 //if the user only entered invalid groups
 262                                 if (globaldata->Groups.size() == 0) {
 263                                         numGroups = tmap->getNumGroups();
 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                                 }else { numGroups = globaldata->Groups.size(); }
 270                         }else { //users wants all groups
 271                                 numGroups = tmap->getNumGroups();
 272                                 globaldata->Groups.clear();
 273                                 globaldata->setGroups("");
 274                         }
 275                 }
 276
 277                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 278                 numComp = 0;
 279                 int n = 1;
 280                 for (int i=1; i<numGroups; i++) {
 281                         numComp += i;
 282                         for (int l = n; l < numGroups; l++) {
 283                                 //set group comparison labels
 284                                 if (globaldata->Groups.size() != 0) {
 285                                         groupComb.push_back(globaldata->Groups[i-1]+globaldata->Groups[l]);
 286                                 }else {
 287                                         groupComb.push_back(tmap->namesOfGroups[i-1]+tmap->namesOfGroups[l]);
 288                                 }
 289                         }
 290                         n++;
 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