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