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