unifracunweightedcommand.cpp

   1 /*
   2  *  unifracunweightedcommand.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 "unifracunweightedcommand.h"
  11
  12 /***********************************************************/
  13 UnifracUnweightedCommand::UnifracUnweightedCommand(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.unweighted 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() + ".uwsummary";
  58                                 openOutputFile(sumFile, outSum);
  59
  60                                 util = new SharedUtil();
  61                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  62                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  63
  64                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
  65
  66                                 unweighted = new Unweighted(tmap);
  67
  68                         }
  69
  70                 }
  71
  72         }
  73         catch(exception& e) {
  74                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  75                 exit(1);
  76         }
  77         catch(...) {
  78                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  79                 exit(1);
  80         }
  81 }
  82
  83 //**********************************************************************************************************************
  84
  85 void UnifracUnweightedCommand::help(){
  86         try {
  87                 cout << "The unifrac.unweighted command can only be executed after a successful read.tree command." << "\n";
  88                 cout << "The unifrac.unweighted command parameters are groups and iters.  No parameters are required." << "\n";
  89                 cout << "The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 1 valid group." << "\n";
  90                 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";
  91                 cout << "The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters)." << "\n";
  92                 cout << "Example unifrac.unweighted(groups=A-B-C, iters=500)." << "\n";
  93                 cout << "The default value for groups is all the groups in your groupfile, and iters is 1000." << "\n";
  94                 cout << "The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual." << "\n";
  95                 cout << "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups)." << "\n" << "\n";
  96         }
  97         catch(exception& e) {
  98                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  99                 exit(1);
 100         }
 101         catch(...) {
 102                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 103                 exit(1);
 104         }
 105 }
 106
 107
 108 /***********************************************************/
 109 int UnifracUnweightedCommand::execute() {
 110         try {
 111
 112                 if (abort == true) { return 0; }
 113
 114                 userData.resize(numComp,0);  //data[0] = unweightedscore
 115                 randomData.resize(numComp,0); //data[0] = unweightedscore
 116                 //create new tree with same num nodes and leaves as users
 117
 118                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 119                 cout << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" << '\t' << "UWSig" <<  endl;
 120
 121                 //get pscores for users trees
 122                 for (int i = 0; i < T.size(); i++) {
 123                         counter = 0;
 124
 125                         output = new ColumnFile(globaldata->getTreeFile()  + toString(i+1) + ".unweighted", itersString);
 126
 127                         //get unweighted for users tree
 128                         rscoreFreq.resize(numComp);
 129                         rCumul.resize(numComp);
 130                         utreeScores.resize(numComp);
 131                         UWScoreSig.resize(numComp);
 132
 133                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
 134
 135                         //output scores for each combination
 136                         for(int k = 0; k < numComp; k++) {
 137                                 //saves users score
 138                                 utreeScores[k].push_back(userData[k]);
 139
 140                         }
 141
 142                         //get unweighted scores for random trees
 143                         for (int j = 0; j < iters; j++) {
 144                                 //we need a different getValues because when we swap the labels we only want to swap those in each parwise comparison
 145                                 randomData = unweighted->getValues(T[i], "", "");
 146
 147                                 for(int k = 0; k < numComp; k++) {
 148                                         //add trees unweighted score to map of scores
 149                                         it2 = rscoreFreq[k].find(randomData[k]);
 150                                         if (it2 != rscoreFreq[k].end()) {//already have that score
 151                                                 rscoreFreq[k][randomData[k]]++;
 152                                         }else{//first time we have seen this score
 153                                                 rscoreFreq[k][randomData[k]] = 1;
 154                                         }
 155
 156                                         //add randoms score to validscores
 157                                         validScores[randomData[k]] = randomData[k];
 158                                 }
 159
 160                         }
 161
 162                         for(int a = 0; a < numComp; a++) {
 163                                 float rcumul = 1.0000;
 164                                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 165                                 for (it = validScores.begin(); it != validScores.end(); it++) {
 166                                         //make rscoreFreq map and rCumul
 167                                         it2 = rscoreFreq[a].find(it->first);
 168                                         rCumul[a][it->first] = rcumul;
 169                                         //get percentage of random trees with that info
 170                                         if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 171                                         else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 172                                 }
 173                                 UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 174                         }
 175
 176
 177
 178                         printUnweightedFile();
 179                         printUWSummaryFile(i);
 180
 181                         delete output;
 182                         rscoreFreq.clear();
 183                         rCumul.clear();
 184                         validScores.clear();
 185                         utreeScores.clear();
 186                         UWScoreSig.clear();
 187                 }
 188
 189                 //reset groups parameter
 190                 globaldata->Groups.clear();
 191                 outSum.close();
 192
 193                 return 0;
 194
 195         }
 196         catch(exception& e) {
 197                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 198                 exit(1);
 199         }
 200         catch(...) {
 201                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 202                 exit(1);
 203         }
 204 }
 205 /***********************************************************/
 206 void UnifracUnweightedCommand::printUnweightedFile() {
 207         try {
 208                 vector<double> data;
 209                 vector<string> tags;
 210                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 211
 212                 for(int a = 0; a < numComp; a++) {
 213                         output->initFile(groupComb[a], tags);
 214                         //print each line
 215                         for (it = validScores.begin(); it != validScores.end(); it++) {
 216                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 217                                 output->output(data);
 218                                 data.clear();
 219                         }
 220                         output->resetFile();
 221                 }
 222         }
 223         catch(exception& e) {
 224                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 225                 exit(1);
 226         }
 227         catch(...) {
 228                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 229                 exit(1);
 230         }
 231 }
 232
 233 /***********************************************************/
 234 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 235         try {
 236
 237                 //format output
 238                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 239
 240                 //print each line
 241
 242                 for(int a = 0; a < numComp; a++) {
 243                         outSum << i+1 << '\t';
 244                         cout << i+1 << '\t';
 245
 246                         if (UWScoreSig[a][0] > (1/(float)iters)) {
 247                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 248                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 249                         }else {
 250                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 251                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 252                         }
 253                 }
 254
 255         }
 256         catch(exception& e) {
 257                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 258                 exit(1);
 259         }
 260         catch(...) {
 261                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 262                 exit(1);
 263         }
 264 }
 265
 266 /***********************************************************/
 267
 268