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                         cout << "Processing tree " << i+1 << endl;
  56                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
  57
  58                         //save users score
  59                         for (int s=0; s<numComp; s++) {
  60                                 //add users score to vector of user scores
  61                                 uScores[s].push_back(userData[s]);
  62
  63                                 //save users tree score for summary file
  64                                 utreeScores.push_back(userData[s]);
  65                         }
  66
  67                         //get scores for random trees
  68                         for (int j = 0; j < iters; j++) {
  69                                 int count = 0;
  70                                 for (int r=0; r<numGroups; r++) {
  71                                         for (int l = r+1; l < numGroups; l++) {
  72                                                 //copy T[i]'s info.
  73                                                 randT->getCopy(T[i]);
  74
  75                                                 //create a random tree with same topology as T[i], but different labels
  76                                                 randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
  77                                                 //get wscore of random tree
  78                                                 randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
  79
  80                                                 //save scores
  81                                                 rScores[count].push_back(randomData[0]);
  82                                                 count++;
  83                                         }
  84                                 }
  85                         }
  86
  87                         //removeValidScoresDuplicates();
  88                         //find the signifigance of the score for summary file
  89                         for (int f = 0; f < numComp; f++) {
  90                                 //sort random scores
  91                                 sort(rScores[f].begin(), rScores[f].end());
  92
  93                                 //the index of the score higher than yours is returned
  94                                 //so if you have 1000 random trees the index returned is 100
  95                                 //then there are 900 trees with a score greater then you.
  96                                 //giving you a signifigance of 0.900
  97                                 int index = findIndex(userData[f], f);    if (index == -1) { cout << "error in UnifracWeightedCommand" << endl; exit(1); } //error code
  98
  99                                 //the signifigance is the number of trees with the users score or higher
 100                                 WScoreSig.push_back((iters-index)/(float)iters);
 101                         }
 102
 103                         //out << "Tree# " << i << endl;
 104                         calculateFreqsCumuls();
 105                         printWeightedFile();
 106
 107                         //clear data
 108                         rScores.clear();
 109                         uScores.clear();
 110                         validScores.clear();
 111                 }
 112
 113                 printWSummaryFile();
 114
 115                 //clear out users groups
 116                 globaldata->Groups.clear();
 117
 118                 delete randT;
 119
 120                 return 0;
 121
 122         }
 123         catch(exception& e) {
 124                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 125                 exit(1);
 126         }
 127         catch(...) {
 128                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 129                 exit(1);
 130         }
 131 }
 132 /***********************************************************/
 133 void UnifracWeightedCommand::printWeightedFile() {
 134         try {
 135                 vector<double> data;
 136
 137                 for(int a = 0; a < numComp; a++) {
 138                         initFile(groupComb[a]);
 139                         //print each line
 140                         for (it = validScores.begin(); it != validScores.end(); it++) {
 141                                 data.push_back(it->first);  data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 142                                 output(data);
 143                                 data.clear();
 144                         }
 145                         resetFile();
 146                 }
 147
 148                 out.close();
 149                 inFile.close();
 150                 remove(weightedFileout.c_str());
 151
 152         }
 153         catch(exception& e) {
 154                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 155                 exit(1);
 156         }
 157         catch(...) {
 158                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 159                 exit(1);
 160         }
 161 }
 162
 163
 164 /***********************************************************/
 165 void UnifracWeightedCommand::printWSummaryFile() {
 166         try {
 167                 //column headers
 168                 outSum << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 169                 cout << "Tree#" << '\t' << "Groups" << '\t' << '\t' << "WScore" << '\t' << '\t' << "WSig" <<  endl;
 170
 171                 //format output
 172                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 173
 174                 //print each line
 175                 int count = 0;
 176                 for (int i = 0; i < T.size(); i++) {
 177                         for (int j = 0; j < numComp; j++) {
 178                                 outSum << setprecision(globaldata->getIters().length()) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 179                                 cout << setprecision(globaldata->getIters().length()) << i+1 << '\t' << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << WScoreSig[count] << endl;
 180                                 count++;
 181                         }
 182                 }
 183                 outSum.close();
 184         }
 185         catch(exception& e) {
 186                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 187                 exit(1);
 188         }
 189         catch(...) {
 190                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 191                 exit(1);
 192         }
 193 }
 194
 195 /***********************************************************/
 196 int UnifracWeightedCommand::findIndex(float score, int index) {
 197         try{
 198                 for (int i = 0; i < rScores[index].size(); i++) {
 199                         if (rScores[index][i] >= score) {       return i;       }
 200                 }
 201                 return rScores[index].size();
 202         }
 203         catch(exception& e) {
 204                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 205                 exit(1);
 206         }
 207         catch(...) {
 208                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 209                 exit(1);
 210         }
 211 }
 212
 213 /***********************************************************/
 214 void UnifracWeightedCommand::setGroups() {
 215         try {
 216                 //if the user has not entered specific groups to analyze then do them all
 217                 if (globaldata->Groups.size() == 0) {
 218                         numGroups = tmap->getNumGroups();
 219                         for (int i=0; i < numGroups; i++) {
 220                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 221                         }
 222                 }else {
 223                         if (globaldata->getGroups() != "all") {
 224                                 //check that groups are valid
 225                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 226                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 227                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 228                                                 // erase the invalid group from globaldata->Groups
 229                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
 230                                         }
 231                                 }
 232
 233                                 //if the user only entered invalid groups
 234                                 if (globaldata->Groups.size() == 0) {
 235                                         numGroups = tmap->getNumGroups();
 236                                         for (int i=0; i < numGroups; i++) {
 237                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 238                                         }
 239                                         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;
 240                                 }else if (globaldata->Groups.size() == 1) {
 241                                         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;
 242                                         numGroups = tmap->getNumGroups();
 243                                         globaldata->Groups.clear();
 244                                         for (int i=0; i < numGroups; i++) {
 245                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 246                                         }
 247                                 }else { numGroups = globaldata->Groups.size(); }
 248                         }else { //users wants all groups
 249                                 numGroups = tmap->getNumGroups();
 250                                 globaldata->Groups.clear();
 251                                 globaldata->setGroups("");
 252                                 for (int i=0; i < numGroups; i++) {
 253                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 254                                 }
 255                         }
 256                 }
 257
 258                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 259                 numComp = 0;
 260                 for (int i=0; i<numGroups; i++) {
 261                         numComp += i;
 262                         for (int l = i+1; l < numGroups; l++) {
 263                                 //set group comparison labels
 264                                 groupComb.push_back(globaldata->Groups[i]+globaldata->Groups[l]);
 265                         }
 266                 }
 267         }
 268         catch(exception& e) {
 269                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 270                 exit(1);
 271         }
 272         catch(...) {
 273                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 274                 exit(1);
 275         }
 276 }
 277
 278 /***********************************************************/
 279
 280 void UnifracWeightedCommand::calculateFreqsCumuls() {
 281         try {
 282                 //clear out old tree values
 283                 rScoreFreq.clear();
 284                 rScoreFreq.resize(numComp);
 285                 rCumul.clear();
 286                 rCumul.resize(numComp);
 287                 validScores.clear();
 288
 289                 //calculate frequency
 290                 for (int f = 0; f < numComp; f++) {
 291                         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...
 292                                 validScores[rScores[f][i]] = rScores[f][i];
 293                                 it = rScoreFreq[f].find(rScores[f][i]);
 294                                 if (it != rScoreFreq[f].end()) {
 295                                         rScoreFreq[f][rScores[f][i]]++;
 296                                 }else{
 297                                         rScoreFreq[f][rScores[f][i]] = 1;
 298                                 }
 299                         }
 300                 }
 301
 302                 //calculate rcumul
 303                 for(int a = 0; a < numComp; a++) {
 304                         float rcumul = 1.0000;
 305                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 306                         for (it = validScores.begin(); it != validScores.end(); it++) {
 307                                 //make rscoreFreq map and rCumul
 308                                 it2 = rScoreFreq[a].find(it->first);
 309                                 rCumul[a][it->first] = rcumul;
 310                                 //get percentage of random trees with that info
 311                                 if (it2 != rScoreFreq[a].end()) {  rScoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 312                                 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 313                         }
 314                 }
 315
 316         }
 317         catch(exception& e) {
 318                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function calculateFreqsCums. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 319                 exit(1);
 320         }
 321         catch(...) {
 322                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function calculateFreqsCums. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 323                 exit(1);
 324         }
 325
 326 }
 327
 328 /*****************************************************************/
 329
 330 void UnifracWeightedCommand::initFile(string label){
 331         try {
 332                 if(counter != 0){
 333                         openOutputFile(weightedFileout, out);
 334                         openInputFile(weightedFile, inFile);
 335
 336                         string inputBuffer;
 337                         getline(inFile, inputBuffer);
 338
 339                         out     <<  inputBuffer << '\t' << label + "Score" << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 340                 }else{
 341                         openOutputFile(weightedFileout, out);
 342                         out     << label + "Score" << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 343                 }
 344
 345                 out.setf(ios::fixed, ios::floatfield);
 346                 out.setf(ios::showpoint);
 347         }
 348         catch(exception& e) {
 349                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 350                 exit(1);
 351         }
 352         catch(...) {
 353                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 354                 exit(1);
 355         }
 356 }
 357
 358 /***********************************************************************/
 359
 360 void UnifracWeightedCommand::output(vector<double> data){
 361         try {
 362                 if(counter != 0){
 363                         string inputBuffer;
 364                         getline(inFile, inputBuffer);
 365
 366                         out     <<  inputBuffer << setprecision(globaldata->getIters().length()) << '\t' << data[0] << '\t' << data[1] << '\t' << data[2] << endl;
 367                 }
 368                 else{
 369                         out << setprecision(globaldata->getIters().length()) << data[0] << '\t' << data[1] << '\t' << data[2] << endl;
 370                 }
 371
 372         }
 373         catch(exception& e) {
 374                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 375                 exit(1);
 376         }
 377         catch(...) {
 378                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 379                 exit(1);
 380         }
 381 };
 382
 383 /***********************************************************************/
 384
 385 void UnifracWeightedCommand::resetFile(){
 386         try {
 387                 if(counter != 0){
 388                         out.close();
 389                         inFile.close();
 390                 }
 391                 else{
 392                         out.close();
 393                 }
 394                 counter = 1;
 395
 396                 remove(weightedFile.c_str());
 397                 rename(weightedFileout.c_str(), weightedFile.c_str());
 398         }
 399         catch(exception& e) {
 400                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 401                 exit(1);
 402         }
 403         catch(...) {
 404                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 405                 exit(1);
 406         }
 407 }
 408