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