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(string option) {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 Groups.clear();
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"groups","iters"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         parser = new OptionParser();
  28                         parser->parse(option, parameters);  delete parser;
  29
  30                         ValidParameters* validParameter = new ValidParameters();
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (it4 = parameters.begin(); it4 != parameters.end(); it4++) {
  34                                 if (validParameter->isValidParameter(it4->first, myArray, it4->second) != true) {  abort = true;  }
  35                         }
  36
  37                         if (globaldata->gTree.size() == 0) {//no trees were read
  38                                 cout << "You must execute the read.tree command, before you may execute the unifrac.weighted command." << endl; abort = true;  }
  39
  40                         //check for optional parameter and set defaults
  41                         // ...at some point should added some additional type checking...
  42                         groups = validParameter->validFile(parameters, "groups", false);
  43                         if (groups == "not found") { groups = ""; }
  44                         else {
  45                                 splitAtDash(groups, Groups);
  46                                 globaldata->Groups = Groups;
  47                         }
  48
  49                         itersString = validParameter->validFile(parameters, "iters", false);                    if (itersString == "not found") { itersString = "1000"; }
  50                         convert(itersString, iters);
  51
  52                         delete validParameter;
  53
  54                         if (abort == false) {
  55                                 T = globaldata->gTree;
  56                                 tmap = globaldata->gTreemap;
  57                                 sumFile = globaldata->getTreeFile() + ".wsummary";
  58                                 openOutputFile(sumFile, outSum);
  59
  60                                 util = new SharedUtil();
  61                                 string s; //to make work with setgroups
  62                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, s, numGroups, "weighted");     //sets the groups the user wants to analyze
  63                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  64
  65                                 weighted = new Weighted(tmap);
  66
  67                         }
  68                 }
  69
  70
  71         }
  72         catch(exception& e) {
  73                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  74                 exit(1);
  75         }
  76         catch(...) {
  77                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function UnifracWeightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82
  83 void UnifracWeightedCommand::help(){
  84         try {
  85                 cout << "The unifrac.weighted command can only be executed after a successful read.tree command." << "\n";
  86                 cout << "The unifrac.weighted command parameters are groups and iters.  No parameters are required." << "\n";
  87                 cout << "The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 2 valid groups." << "\n";
  88                 cout << "The group names are separated by dashes.  The iters parameter allows you to specify how many random trees you would like compared to your tree." << "\n";
  89                 cout << "The unifrac.weighted command should be in the following format: unifrac.weighted(groups=yourGroups, iters=yourIters)." << "\n";
  90                 cout << "Example unifrac.weighted(groups=A-B-C, iters=500)." << "\n";
  91                 cout << "The default value for groups is all the groups in your groupfile, and iters is 1000." << "\n";
  92                 cout << "The unifrac.weighted command output two files: .weighted and .wsummary their descriptions are in the manual." << "\n";
  93                 cout << "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups)." << "\n" << "\n";
  94         }
  95         catch(exception& e) {
  96                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  97                 exit(1);
  98         }
  99         catch(...) {
 100                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 101                 exit(1);
 102         }
 103 }
 104
 105 /***********************************************************/
 106 int UnifracWeightedCommand::execute() {
 107         try {
 108
 109                 if (abort == true) { return 0; }
 110
 111                 Progress* reading;
 112                 reading = new Progress("Comparing to random:", iters);
 113
 114                 //get weighted for users tree
 115                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 116                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
 117
 118                 //create new tree with same num nodes and leaves as users
 119                 randT = new Tree();
 120
 121                 //get weighted scores for users trees
 122                 for (int i = 0; i < T.size(); i++) {
 123                         counter = 0;
 124                         rScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 125                         uScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 126
 127                         output = new ColumnFile(globaldata->getTreeFile()  + toString(i+1) + ".weighted", itersString);
 128
 129                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
 130
 131                         //save users score
 132                         for (int s=0; s<numComp; s++) {
 133                                 //add users score to vector of user scores
 134                                 uScores[s].push_back(userData[s]);
 135
 136                                 //save users tree score for summary file
 137                                 utreeScores.push_back(userData[s]);
 138                         }
 139
 140                         //get scores for random trees
 141                         for (int j = 0; j < iters; j++) {
 142                                 int count = 0;
 143                                 for (int r=0; r<numGroups; r++) {
 144                                         for (int l = r+1; l < numGroups; l++) {
 145                                                 //copy T[i]'s info.
 146                                                 randT->getCopy(T[i]);
 147
 148                                                 //create a random tree with same topology as T[i], but different labels
 149                                                 randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
 150                                                 //get wscore of random tree
 151                                                 randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
 152
 153                                                 //save scores
 154                                                 rScores[count].push_back(randomData[0]);
 155                                                 count++;
 156                                         }
 157                                 }
 158
 159                                 //update progress bar
 160                                 reading->update(j);
 161
 162                         }
 163
 164                         //removeValidScoresDuplicates();
 165                         //find the signifigance of the score for summary file
 166                         for (int f = 0; f < numComp; f++) {
 167                                 //sort random scores
 168                                 sort(rScores[f].begin(), rScores[f].end());
 169
 170                                 //the index of the score higher than yours is returned
 171                                 //so if you have 1000 random trees the index returned is 100
 172                                 //then there are 900 trees with a score greater then you.
 173                                 //giving you a signifigance of 0.900
 174                                 int index = findIndex(userData[f], f);    if (index == -1) { cout << "error in UnifracWeightedCommand" << endl; exit(1); } //error code
 175
 176                                 //the signifigance is the number of trees with the users score or higher
 177                                 WScoreSig.push_back((iters-index)/(float)iters);
 178                         }
 179
 180                         //out << "Tree# " << i << endl;
 181                         calculateFreqsCumuls();
 182                         printWeightedFile();
 183
 184                         delete output;
 185
 186                         //clear data
 187                         rScores.clear();
 188                         uScores.clear();
 189                         validScores.clear();
 190                 }
 191
 192                 //finish progress bar
 193                 reading->finish();
 194                 delete reading;
 195
 196                 printWSummaryFile();
 197
 198                 //clear out users groups
 199                 globaldata->Groups.clear();
 200
 201                 delete randT;
 202
 203                 return 0;
 204
 205         }
 206         catch(exception& e) {
 207                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 208                 exit(1);
 209         }
 210         catch(...) {
 211                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 212                 exit(1);
 213         }
 214 }
 215 /***********************************************************/
 216 void UnifracWeightedCommand::printWeightedFile() {
 217         try {
 218                 vector<double> data;
 219                 vector<string> tags;
 220                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 221
 222                 for(int a = 0; a < numComp; a++) {
 223                         output->initFile(groupComb[a], tags);
 224                         //print each line
 225                         for (it = validScores.begin(); it != validScores.end(); it++) {
 226                                 data.push_back(it->first);  data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 227                                 output->output(data);
 228                                 data.clear();
 229                         }
 230                         output->resetFile();
 231                 }
 232         }
 233         catch(exception& e) {
 234                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 235                 exit(1);
 236         }
 237         catch(...) {
 238                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 239                 exit(1);
 240         }
 241 }
 242
 243
 244 /***********************************************************/
 245 void UnifracWeightedCommand::printWSummaryFile() {
 246         try {
 247                 //column headers
 248                 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 249                 cout << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 250
 251                 //format output
 252                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 253
 254                 //print each line
 255                 int count = 0;
 256                 for (int i = 0; i < T.size(); i++) {
 257                         for (int j = 0; j < numComp; j++) {
 258                                 if (WScoreSig[count] > (1/(float)iters)) {
 259                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 260                                         cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 261                                 }else{
 262                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 263                                         cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 264                                 }
 265                                 count++;
 266                         }
 267                 }
 268                 outSum.close();
 269         }
 270         catch(exception& e) {
 271                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 272                 exit(1);
 273         }
 274         catch(...) {
 275                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function printWeightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 276                 exit(1);
 277         }
 278 }
 279
 280 /***********************************************************/
 281 int UnifracWeightedCommand::findIndex(float score, int index) {
 282         try{
 283                 for (int i = 0; i < rScores[index].size(); i++) {
 284                         if (rScores[index][i] >= score) {       return i;       }
 285                 }
 286                 return rScores[index].size();
 287         }
 288         catch(exception& e) {
 289                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 290                 exit(1);
 291         }
 292         catch(...) {
 293                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function findIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 294                 exit(1);
 295         }
 296 }
 297
 298 /***********************************************************/
 299
 300 void UnifracWeightedCommand::calculateFreqsCumuls() {
 301         try {
 302                 //clear out old tree values
 303                 rScoreFreq.clear();
 304                 rScoreFreq.resize(numComp);
 305                 rCumul.clear();
 306                 rCumul.resize(numComp);
 307                 validScores.clear();
 308
 309                 //calculate frequency
 310                 for (int f = 0; f < numComp; f++) {
 311                         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...
 312                                 validScores[rScores[f][i]] = rScores[f][i];
 313                                 it = rScoreFreq[f].find(rScores[f][i]);
 314                                 if (it != rScoreFreq[f].end()) {
 315                                         rScoreFreq[f][rScores[f][i]]++;
 316                                 }else{
 317                                         rScoreFreq[f][rScores[f][i]] = 1;
 318                                 }
 319                         }
 320                 }
 321
 322                 //calculate rcumul
 323                 for(int a = 0; a < numComp; a++) {
 324                         float rcumul = 1.0000;
 325                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 326                         for (it = validScores.begin(); it != validScores.end(); it++) {
 327                                 //make rscoreFreq map and rCumul
 328                                 it2 = rScoreFreq[a].find(it->first);
 329                                 rCumul[a][it->first] = rcumul;
 330                                 //get percentage of random trees with that info
 331                                 if (it2 != rScoreFreq[a].end()) {  rScoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 332                                 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 333                         }
 334                 }
 335
 336         }
 337         catch(exception& e) {
 338                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracWeightedCommand class Function calculateFreqsCums. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 339                 exit(1);
 340         }
 341         catch(...) {
 342                 cout << "An unknown error has occurred in the UnifracWeightedCommand class function calculateFreqsCums. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 343                 exit(1);
 344         }
 345
 346 }
 347
 348 /***********************************************************/
 349
 350
 351
 352
 353