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() {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16
  17                 T = globaldata->gTree;
  18                 tmap = globaldata->gTreemap;
  19                 sumFile = globaldata->getTreeFile() + ".uwsummary";
  20                 openOutputFile(sumFile, outSum);
  21
  22                 util = new SharedUtil();
  23                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  24                 util->getCombos(groupComb, globaldata->Groups, numComp);
  25                 globaldata->setGroups("");
  26
  27                 //ABC
  28                 if (numComp != 1) {
  29                         groupComb.push_back(allGroups);
  30                         numComp++;
  31                 }
  32
  33
  34                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  35                 unweighted = new Unweighted(tmap);
  36
  37         }
  38         catch(exception& e) {
  39                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  40                 exit(1);
  41         }
  42         catch(...) {
  43                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  44                 exit(1);
  45         }
  46 }
  47 /***********************************************************/
  48 int UnifracUnweightedCommand::execute() {
  49         try {
  50
  51                 userData.resize(numComp,0);  //data[0] = unweightedscore
  52                 randomData.resize(numComp,0); //data[0] = unweightedscore
  53                 //create new tree with same num nodes and leaves as users
  54
  55                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
  56                 cout << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" << '\t' << "UWSig" <<  endl;
  57
  58                 //get pscores for users trees
  59                 for (int i = 0; i < T.size(); i++) {
  60                         counter = 0;
  61                         unweightedFile = globaldata->getTreeFile()  + toString(i+1) + ".unweighted";
  62                         unweightedFileout = globaldata->getTreeFile() + "temp." + toString(i+1) + ".unweighted";
  63
  64                         outSum << i+1 << '\t';
  65                         cout << i+1 << '\t';
  66
  67                         //get unweighted for users tree
  68                         rscoreFreq.resize(numComp);
  69                         rCumul.resize(numComp);
  70                         utreeScores.resize(numComp);
  71                         UWScoreSig.resize(numComp);
  72
  73                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
  74
  75                         //output scores for each combination
  76                         for(int k = 0; k < numComp; k++) {
  77                                 //saves users score
  78                                 utreeScores[k].push_back(userData[k]);
  79
  80                         }
  81
  82                         //get unweighted scores for random trees
  83                         for (int j = 0; j < iters; j++) {
  84                                 //we need a different getValues because when we swap the labels we only want to swap those in each parwise comparison
  85                                 randomData = unweighted->getValues(T[i], "", "");
  86
  87                                 for(int k = 0; k < numComp; k++) {
  88                                         //add trees unweighted score to map of scores
  89                                         it2 = rscoreFreq[k].find(randomData[k]);
  90                                         if (it2 != rscoreFreq[k].end()) {//already have that score
  91                                                 rscoreFreq[k][randomData[k]]++;
  92                                         }else{//first time we have seen this score
  93                                                 rscoreFreq[k][randomData[k]] = 1;
  94                                         }
  95
  96                                         //add randoms score to validscores
  97                                         validScores[randomData[k]] = randomData[k];
  98                                 }
  99
 100                         }
 101
 102                 for(int a = 0; a < numComp; a++) {
 103                         float rcumul = 1.0000;
 104                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 105                         for (it = validScores.begin(); it != validScores.end(); it++) {
 106                                 //make rscoreFreq map and rCumul
 107                                 it2 = rscoreFreq[a].find(it->first);
 108                                 rCumul[a][it->first] = rcumul;
 109                                 //get percentage of random trees with that info
 110                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 111                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 112                         }
 113                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 114                 }
 115
 116
 117
 118                 printUnweightedFile();
 119                 printUWSummaryFile();
 120
 121                 rscoreFreq.clear();
 122                 rCumul.clear();
 123                 validScores.clear();
 124                 utreeScores.clear();
 125                 UWScoreSig.clear();
 126         }
 127                 //reset groups parameter
 128                 globaldata->Groups.clear();
 129                 outSum.close();
 130
 131                 return 0;
 132
 133         }
 134         catch(exception& e) {
 135                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 136                 exit(1);
 137         }
 138         catch(...) {
 139                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 140                 exit(1);
 141         }
 142 }
 143 /***********************************************************/
 144 void UnifracUnweightedCommand::printUnweightedFile() {
 145         try {
 146                 vector<double> data;
 147
 148                 for(int a = 0; a < numComp; a++) {
 149                         initFile(groupComb[a]);
 150                         //print each line
 151                         for (it = validScores.begin(); it != validScores.end(); it++) {
 152                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 153                                 output(data);
 154                                 data.clear();
 155                         }
 156                         resetFile();
 157                 }
 158
 159                 out.close();
 160                 inFile.close();
 161                 remove(unweightedFileout.c_str());
 162
 163         }
 164         catch(exception& e) {
 165                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 166                 exit(1);
 167         }
 168         catch(...) {
 169                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 170                 exit(1);
 171         }
 172 }
 173
 174 /***********************************************************/
 175 void UnifracUnweightedCommand::printUWSummaryFile() {
 176         try {
 177
 178                 //format output
 179                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 180
 181                 //print each line
 182
 183                 for(int a = 0; a < numComp; a++) {
 184                         if (UWScoreSig[a][0] > (1/(float)iters)) {
 185                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << UWScoreSig[a][0] << endl;
 186                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << UWScoreSig[a][0] << endl;
 187                         }else {
 188                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << "<" << (1/float(iters)) << endl;
 189                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(globaldata->getIters().length()) << "<" << (1/float(iters)) << endl;
 190                         }
 191                 }
 192
 193         }
 194         catch(exception& e) {
 195                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 196                 exit(1);
 197         }
 198         catch(...) {
 199                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 200                 exit(1);
 201         }
 202 }
 203 /*****************************************************************/
 204
 205 void UnifracUnweightedCommand::initFile(string label){
 206         try {
 207                 if(counter != 0){
 208                         openOutputFile(unweightedFileout, out);
 209                         openInputFile(unweightedFile, inFile);
 210
 211                         string inputBuffer;
 212                         getline(inFile, inputBuffer);
 213
 214                         out     <<  inputBuffer << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 215                 }else{
 216                         openOutputFile(unweightedFileout, out);
 217                         out     << label + "Score" << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 218                 }
 219
 220                 out.setf(ios::fixed, ios::floatfield);
 221                 out.setf(ios::showpoint);
 222         }
 223         catch(exception& e) {
 224                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function initFile. 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 initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 229                 exit(1);
 230         }
 231 }
 232
 233 /***********************************************************************/
 234
 235 void UnifracUnweightedCommand::output(vector<double> data){
 236         try {
 237                 if(counter != 0){
 238                         string inputBuffer;
 239                         getline(inFile, inputBuffer);
 240 //                      out     <<  inputBuffer << setprecision(6) << '\t' << data[0] << setprecision(globaldata->getIters().length()) << '\t' << data[1] << '\t' << data[2] << endl;
 241
 242                         out << inputBuffer << '\t' << setprecision(6) << data[0] << setprecision(globaldata->getIters().length())  << '\t' << data[1] << '\t' << data[2] << endl;
 243                 }
 244                 else{
 245                         out << setprecision(6) << data[0] << setprecision(globaldata->getIters().length())  << '\t' << data[1] << '\t' << data[2] << endl;
 246                 }
 247
 248         }
 249         catch(exception& e) {
 250                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 251                 exit(1);
 252         }
 253         catch(...) {
 254                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 255                 exit(1);
 256         }
 257 };
 258
 259 /***********************************************************************/
 260
 261 void UnifracUnweightedCommand::resetFile(){
 262         try {
 263                 if(counter != 0){
 264                         out.close();
 265                         inFile.close();
 266                 }
 267                 else{
 268                         out.close();
 269                 }
 270                 counter = 1;
 271
 272                 remove(unweightedFile.c_str());
 273                 rename(unweightedFileout.c_str(), unweightedFile.c_str());
 274         }
 275         catch(exception& e) {
 276                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 277                 exit(1);
 278         }
 279         catch(...) {
 280                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 281                 exit(1);
 282         }
 283 }
 284