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                 sumFile = globaldata->getTreeFile() + ".wsummary";
  20                 openOutputFile(sumFile, outSum);
  21
  22                 setGroups();    //sets the groups the user wants to analyze
  23                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  24                 weighted = new Weighted(tmap);
  25
  26         }
  27         catch(exception& e) {
  28                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  29                 exit(1);
  30         }
  31         catch(...) {
  32                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  33                 exit(1);
  34         }
  35 }
  36 /***********************************************************/
  37 int UnifracWeightedCommand::execute() {
  38         try {
  39
  40                 //get weighted for users tree
  41                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  42                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
  43
  44                 //create new tree with same num nodes and leaves as users
  45                 randT = new Tree();
  46
  47                 //get weighted scores for users trees
  48                 for (int i = 0; i < T.size(); i++) {
  49                         counter = 0;
  50                         rScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  51                         uScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
  52                         weightedFile = globaldata->getTreeFile()  + toString(i+1) + ".weighted";
  53                         weightedFileout = globaldata->getTreeFile() + "temp." + toString(i+1) + ".weighted";
  54
  55                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
  56
  57                         //save users score
  58                         for (int s=0; s<numComp; s++) {
  59                                 //add users score to vector of user scores
  60                                 uScores[s].push_back(userData[s]);
  61
  62                                 //save users tree score for summary file
  63                                 utreeScores.push_back(userData[s]);
  64                         }
  65
  66                         //get scores for random trees
  67                         for (int j = 0; j < iters; j++) {
  68                                 int count = 0;
  69                                 for (int r=0; r<numGroups; r++) {
  70                                         for (int l = r+1; l < numGroups; l++) {
  71                                                 //copy T[i]'s info.
  72                                                 randT->getCopy(T[i]);
  73
  74                                                 //create a random tree with same topology as T[i], but different labels
  75                                                 randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
  76                                                 //get wscore of random tree
  77                                                 randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
  78
  79                                                 //save scores
  80                                                 rScores[count].push_back(randomData[0]);
  81                                                 count++;
  82                                         }
  83                                 }
  84                         }
  85
  86                         //removeValidScoresDuplicates();
  87                         //find the signifigance of the score for summary file
  88                         for (int f = 0; f < numComp; f++) {
  89                                 //sort random scores
  90                                 sort(rScores[f].begin(), rScores[f].end());
  91
  92                                 //the index of the score higher than yours is returned
  93                                 //so if you have 1000 random trees the index returned is 100
  94                                 //then there are 900 trees with a score greater then you.
  95                                 //giving you a signifigance of 0.900
  96                                 int index = findIndex(userData[f], f);    if (index == -1) { cout << "error in UnifracWeightedCommand" << endl; exit(1); } //error code
  97
  98                                 //the signifigance is the number of trees with the users score or higher
  99                                 WScoreSig.push_back((iters-index)/(float)iters);
 100                         }
 101
 102                         //out << "Tree# " << i << endl;
 103                         calculateFreqsCumuls();
 104                         printWeightedFile();
 105
 106                         //clear data
 107                         rScores.clear();
 108                         uScores.clear();
 109                         validScores.clear();
 110                 }
 111
 112                 printWSummaryFile();
 113
 114                 //clear out users groups
 115                 globaldata->Groups.clear();
 116
 117                 delete randT;
 118
 119                 return 0;
 120
 121         }
 122         catch(exception& e) {
 123                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 124                 exit(1);
 125         }
 126         catch(...) {
 127                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 128                 exit(1);
 129         }
 130 }
 131 /***********************************************************/
 132 void UnifracWeightedCommand::printWeightedFile() {
 133         try {
 134                 vector<double> data;
 135
 136                 for(int a = 0; a < numComp; a++) {
 137                         initFile(groupComb[a]);
 138                         //print each line
 139                         for (it = validScores.begin(); it != validScores.end(); it++) {
 140                                 data.push_back(it->first);  data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 141                                 output(data);
 142                                 data.clear();
 143                         }
 144                         resetFile();
 145                 }
 146
 147                 out.close();
 148                 inFile.close();
 149                 remove(weightedFileout.c_str());
 150
 151         }
 152         catch(exception& e) {
 153                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 154                 exit(1);
 155         }
 156         catch(...) {
 157                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 158                 exit(1);
 159         }
 160 }
 161
 162
 163 /***********************************************************/
 164 void UnifracWeightedCommand::printWSummaryFile() {
 165         try {
 166                 //column headers
 167                 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 168                 cout << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 169
 170                 //format output
 171                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 172
 173                 //print each line
 174                 int count = 0;
 175                 for (int i = 0; i < T.size(); i++) {
 176                         for (int j = 0; j < numComp; j++) {
 177                                 if (WScoreSig[count] > (1/(float)iters)) {
 178                                         outSum << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << '\t'  << utreeScores[count] << '\t' << setprecision(globaldata->getIters().length()) << WScoreSig[count] << endl;
 179                                         cout << setprecision(6) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << '\t'  << utreeScores[count] << '\t' << setprecision(globaldata->getIters().length()) << WScoreSig[count] << endl;
 180                                 }else{
 181                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(globaldata->getIters().length()) << "<" << (1/float(iters)) << endl;
 182                                         cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(globaldata->getIters().length()) << "<" << (1/float(iters)) << endl;
 183                                 }
 184                                 count++;
 185                         }
 186                 }
 187                 outSum.close();
 188         }
 189         catch(exception& e) {
 190                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 191                 exit(1);
 192         }
 193         catch(...) {
 194                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 195                 exit(1);
 196         }
 197 }
 198
 199 /***********************************************************/
 200 int UnifracWeightedCommand::findIndex(float score, int index) {
 201         try{
 202                 for (int i = 0; i < rScores[index].size(); i++) {
 203                         if (rScores[index][i] >= score) {       return i;       }
 204                 }
 205                 return rScores[index].size();
 206         }
 207         catch(exception& e) {
 208                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 209                 exit(1);
 210         }
 211         catch(...) {
 212                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 213                 exit(1);
 214         }
 215 }
 216
 217 /***********************************************************/
 218 void UnifracWeightedCommand::setGroups() {
 219         try {
 220                 //if the user has not entered specific groups to analyze then do them all
 221                 if (globaldata->Groups.size() == 0) {
 222                         numGroups = tmap->getNumGroups();
 223                         for (int i=0; i < numGroups; i++) {
 224                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 225                         }
 226                 }else {
 227                         if (globaldata->getGroups() != "all") {
 228                                 //check that groups are valid
 229                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 230                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 231                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 232                                                 // erase the invalid group from globaldata->Groups
 233                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
 234                                         }
 235                                 }
 236
 237                                 //if the user only entered invalid groups
 238                                 if (globaldata->Groups.size() == 0) {
 239                                         numGroups = tmap->getNumGroups();
 240                                         for (int i=0; i < numGroups; i++) {
 241                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 242                                         }
 243                                         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;
 244                                 }else if (globaldata->Groups.size() == 1) {
 245                                         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;
 246                                         numGroups = tmap->getNumGroups();
 247                                         globaldata->Groups.clear();
 248                                         for (int i=0; i < numGroups; i++) {
 249                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 250                                         }
 251                                 }else { numGroups = globaldata->Groups.size(); }
 252                         }else { //users wants all groups
 253                                 numGroups = tmap->getNumGroups();
 254                                 globaldata->Groups.clear();
 255                                 globaldata->setGroups("");
 256                                 for (int i=0; i < numGroups; i++) {
 257                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 258                                 }
 259                         }
 260                 }
 261
 262                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 263                 numComp = 0;
 264                 for (int i=0; i<numGroups; i++) {
 265                         numComp += i;
 266                         for (int l = i+1; l < numGroups; l++) {
 267                                 //set group comparison labels
 268                                 groupComb.push_back(globaldata->Groups[i] + "-" + globaldata->Groups[l]);
 269                         }
 270                 }
 271         }
 272         catch(exception& e) {
 273                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand 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 UnifracWeightedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 278                 exit(1);
 279         }
 280 }
 281
 282 /***********************************************************/
 283
 284 void UnifracWeightedCommand::calculateFreqsCumuls() {
 285         try {
 286                 //clear out old tree values
 287                 rScoreFreq.clear();
 288                 rScoreFreq.resize(numComp);
 289                 rCumul.clear();
 290                 rCumul.resize(numComp);
 291                 validScores.clear();
 292
 293                 //calculate frequency
 294                 for (int f = 0; f < numComp; f++) {
 295                         for (int i = 0; i < rScores[f].size(); i++) { //looks like 0,0,1,1,1,2,4,7...  you want to make a map that say rScoreFreq[0] = 2, rScoreFreq[1] = 3...
 296                                 validScores[rScores[f][i]] = rScores[f][i];
 297                                 it = rScoreFreq[f].find(rScores[f][i]);
 298                                 if (it != rScoreFreq[f].end()) {
 299                                         rScoreFreq[f][rScores[f][i]]++;
 300                                 }else{
 301                                         rScoreFreq[f][rScores[f][i]] = 1;
 302                                 }
 303                         }
 304                 }
 305
 306                 //calculate rcumul
 307                 for(int a = 0; a < numComp; a++) {
 308                         float rcumul = 1.0000;
 309                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 310                         for (it = validScores.begin(); it != validScores.end(); it++) {
 311                                 //make rscoreFreq map and rCumul
 312                                 it2 = rScoreFreq[a].find(it->first);
 313                                 rCumul[a][it->first] = rcumul;
 314                                 //get percentage of random trees with that info
 315                                 if (it2 != rScoreFreq[a].end()) {  rScoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 316                                 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 317                         }
 318                 }
 319
 320         }
 321         catch(exception& e) {
 322                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function calculateFreqsCums. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 323                 exit(1);
 324         }
 325         catch(...) {
 326                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function calculateFreqsCums. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 327                 exit(1);
 328         }
 329
 330 }
 331
 332 /*****************************************************************/
 333
 334 void UnifracWeightedCommand::initFile(string label){
 335         try {
 336                 if(counter != 0){
 337                         openOutputFile(weightedFileout, out);
 338                         openInputFile(weightedFile, inFile);
 339
 340                         string inputBuffer;
 341                         getline(inFile, inputBuffer);
 342
 343                         out     <<  inputBuffer << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 344                 }else{
 345                         openOutputFile(weightedFileout, out);
 346                         out     << label + "Score" << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 347                 }
 348
 349                 out.setf(ios::fixed, ios::floatfield);
 350                 out.setf(ios::showpoint);
 351         }
 352         catch(exception& e) {
 353                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 354                 exit(1);
 355         }
 356         catch(...) {
 357                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 358                 exit(1);
 359         }
 360 }
 361
 362 /***********************************************************************/
 363
 364 void UnifracWeightedCommand::output(vector<double> data){
 365         try {
 366                 if(counter != 0){
 367                         string inputBuffer;
 368                         getline(inFile, inputBuffer);
 369
 370                         out << inputBuffer << '\t' << setprecision(6) << data[0] << setprecision(globaldata->getIters().length())  << '\t' << data[1] << '\t' << data[2] << endl;
 371                 }
 372                 else{
 373                         out << setprecision(6) << data[0] << setprecision(globaldata->getIters().length())  << '\t' << data[1] << '\t' << data[2] << endl;
 374
 375                 }
 376
 377         }
 378         catch(exception& e) {
 379                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 380                 exit(1);
 381         }
 382         catch(...) {
 383                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 384                 exit(1);
 385         }
 386 };
 387
 388 /***********************************************************************/
 389
 390 void UnifracWeightedCommand::resetFile(){
 391         try {
 392                 if(counter != 0){
 393                         out.close();
 394                         inFile.close();
 395                 }
 396                 else{
 397                         out.close();
 398                 }
 399                 counter = 1;
 400
 401                 remove(weightedFile.c_str());
 402                 rename(weightedFileout.c_str(), weightedFile.c_str());
 403         }
 404         catch(exception& e) {
 405                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 406                 exit(1);
 407         }
 408         catch(...) {
 409                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 410                 exit(1);
 411         }
 412 }
 413