parsimonycommand.cpp

   1 /*
   2  *  parsimonycommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/26/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "parsimonycommand.h"
  11
  12 /***********************************************************/
  13 ParsimonyCommand::ParsimonyCommand() {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16
  17                 //randomtree will tell us if user had their own treefile or if they just want the random distribution
  18                 convert(globaldata->getRandomTree(), randomtree);
  19
  20                 //user has entered their own tree
  21                 if (randomtree == 0) {
  22                         T = globaldata->gTree;
  23                         tmap = globaldata->gTreemap;
  24                         parsFile = globaldata->getTreeFile() + ".parsimony";
  25                         openOutputFile(parsFile, out);
  26                         sumFile = globaldata->getTreeFile() + ".psummary";
  27                         openOutputFile(sumFile, outSum);
  28                         distFile = globaldata->getTreeFile() + ".pdistrib";
  29                         openOutputFile(distFile, outDist);
  30
  31                 }else { //user wants random distribution
  32                         getUserInput();
  33                         parsFile = "rd_parsimony";
  34                         openOutputFile(parsFile, out);
  35                 }
  36
  37                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  38                 pars = new Parsimony(tmap);
  39
  40         }
  41         catch(exception& e) {
  42                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  43                 exit(1);
  44         }
  45         catch(...) {
  46                 cout << "An unknown error has occurred in the ParsimonyCommand class function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  47                 exit(1);
  48         }
  49 }
  50 /***********************************************************/
  51 int ParsimonyCommand::execute() {
  52         try {
  53
  54                 //get pscore for users tree
  55                 userData.resize(1,0);  //data[0] = pscore.
  56                 randomData.resize(1,0);  //data[0] = pscore.
  57
  58                 //format output
  59                 outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
  60                 outDist << "RandomTree#" << '\t' << "ParsScore" << endl;
  61
  62                 if (randomtree == 0) {
  63                         //get pscores for users trees
  64                         for (int i = 0; i < T.size(); i++) {
  65                                 cout << "Processing tree " << i+1 << endl;
  66                                 userData = pars->getValues(T[i]);  //userData[0] = pscore
  67                                 //update uscoreFreq
  68                                 it = uscoreFreq.find(userData[0]);
  69                                 if (it == uscoreFreq.end()) {//new score
  70                                         uscoreFreq[userData[0]] = 1;
  71                                 }else{ uscoreFreq[userData[0]]++; }
  72
  73                                 //add users score to valid scores
  74                                 validScores[userData[0]] = userData[0];
  75
  76                                 //save score for summary file
  77                                 userTreeScores.push_back(userData[0]);
  78
  79                         }
  80
  81                         //get pscores for random trees
  82                         for (int j = 0; j < iters; j++) {
  83                                 //create new tree with same num nodes and leaves as users
  84                                 randT = new Tree();
  85                                 //create random relationships between nodes
  86                                 randT->assembleRandomTree();
  87                                 //get pscore of random tree
  88                                 randomData = pars->getValues(randT);
  89
  90                                 //add trees pscore to map of scores
  91                                 it2 = rscoreFreq.find(randomData[0]);
  92                                 if (it2 != rscoreFreq.end()) {//already have that score
  93                                         rscoreFreq[randomData[0]]++;
  94                                 }else{//first time we have seen this score
  95                                         rscoreFreq[randomData[0]] = 1;
  96                                 }
  97
  98                                 //add randoms score to validscores
  99                                 validScores[randomData[0]] = randomData[0];
 100
 101                                 //output info to pdistrib file
 102                                 outDist << j+1 << '\t'<< '\t' << randomData[0] << endl;
 103
 104                                 delete randT;
 105                         }
 106                 }else {
 107                         //get pscores for random trees
 108                         for (int j = 0; j < iters; j++) {
 109                                 //create new tree with same num nodes and leaves as users
 110                                 randT = new Tree();
 111                                 //create random relationships between nodes
 112                                 randT->assembleRandomTree();
 113                                 //get pscore of random tree
 114                                 randomData = pars->getValues(randT);
 115
 116                                 //add trees pscore to map of scores
 117                                 it2 = rscoreFreq.find(randomData[0]);
 118                                 if (it2 != rscoreFreq.end()) {//already have that score
 119                                         rscoreFreq[randomData[0]]++;
 120                                 }else{//first time we have seen this score
 121                                         rscoreFreq[randomData[0]] = 1;
 122                                 }
 123
 124                                 //add randoms score to validscores
 125                                 validScores[randomData[0]] = randomData[0];
 126
 127                                 delete randT;
 128                         }
 129                 }
 130
 131                 float rcumul = 0.0000;
 132                 float ucumul = 0.0000;
 133
 134                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 135                 for (it = validScores.begin(); it != validScores.end(); it++) {
 136                         if (randomtree == 0) {
 137                                 it2 = uscoreFreq.find(it->first);
 138                                 //user data has that score
 139                                 if (it2 != uscoreFreq.end()) { uscoreFreq[it->first] /= T.size(); ucumul+= it2->second;  }
 140                                 else { uscoreFreq[it->first] = 0.0000; } //no user trees with that score
 141                                 //make uCumul map
 142                                 uCumul[it->first] = ucumul;
 143                         }
 144
 145                         //make rscoreFreq map and rCumul
 146                         it2 = rscoreFreq.find(it->first);
 147                         //get percentage of random trees with that info
 148                         if (it2 != rscoreFreq.end()) {  rscoreFreq[it->first] /= iters; rcumul+= it2->second;  }
 149                         else { rscoreFreq[it->first] = 0.0000; } //no random trees with that score
 150                         rCumul[it->first] = rcumul;
 151                 }
 152
 153                 //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 154                 for (int h = 0; h < userTreeScores.size(); h++) {
 155                         UScoreSig.push_back(rCumul[userTreeScores[h]]);
 156                 }
 157
 158                 printParsimonyFile();
 159                 printUSummaryFile();
 160
 161                 //reset randomTree parameter to 0
 162                 globaldata->setRandomTree("0");
 163
 164                 return 0;
 165
 166         }
 167         catch(exception& e) {
 168                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 169                 exit(1);
 170         }
 171         catch(...) {
 172                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 173                 exit(1);
 174         }
 175 }
 176
 177 /***********************************************************/
 178 void ParsimonyCommand::printParsimonyFile() {
 179         try {
 180                 //column headers
 181                 if (randomtree == 0) {
 182                         out << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 183                 }else {
 184                         out << "Score" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 185                 }
 186
 187                 //format output
 188                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 189
 190                 //print each line
 191                 for (it = validScores.begin(); it != validScores.end(); it++) {
 192                         if (randomtree == 0) {
 193                                 out << setprecision(6) << it->first << '\t' << '\t' << uscoreFreq[it->first] << '\t' << uCumul[it->first] << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 194                         }else{
 195                                 out << setprecision(6) << it->first << '\t' << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 196                         }
 197                 }
 198
 199                 out.close();
 200
 201         }
 202         catch(exception& e) {
 203                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 204                 exit(1);
 205         }
 206         catch(...) {
 207                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 208                 exit(1);
 209         }
 210 }
 211 /***********************************************************/
 212 void ParsimonyCommand::printUSummaryFile() {
 213         try {
 214                 //column headers
 215                 outSum << "Tree#" << '\t'  <<  "ParsScore" << '\t' << '\t' << "ParsSig" <<  endl;
 216
 217                 //format output
 218                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 219
 220                 //print each line
 221                 for (int i = 0; i< T.size(); i++) {
 222                         outSum << setprecision(6) << i+1 << '\t' << '\t' << userTreeScores[i] << '\t' << UScoreSig[i] << endl;
 223                 }
 224
 225                 outSum.close();
 226         }
 227         catch(exception& e) {
 228                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 229                 exit(1);
 230         }
 231         catch(...) {
 232                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 233                 exit(1);
 234         }
 235 }
 236
 237 /***********************************************************/
 238 void ParsimonyCommand::getUserInput() {
 239         try {
 240
 241                 //create treemap
 242                 tmap = new TreeMap();
 243
 244                 cout << "Please enter the number of groups you would like to analyze: ";
 245                 cin >> numGroups;
 246
 247                 int num, count;
 248                 count = 1;
 249                 numEachGroup.resize(numGroups, 0);
 250
 251                 for (int i = 1; i <= numGroups; i++) {
 252                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 253                         cin >> num;
 254
 255                         //set tmaps seqsPerGroup
 256                         tmap->seqsPerGroup[toString(i)] = num;
 257
 258                         //set tmaps namesOfSeqs
 259                         for (int j = 0; j < num; j++) {
 260                                 tmap->namesOfSeqs.push_back(toString(count));
 261                                 tmap->treemap[toString(count)].groupname = toString(i);
 262                                 count++;
 263                         }
 264                 }
 265
 266                 //clears buffer so next command doesn't have error
 267                 string s;
 268                 getline(cin, s);
 269
 270                 //save tmap for later
 271                 globaldata->gTreemap = tmap;
 272
 273         }
 274         catch(exception& e) {
 275                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 276                 exit(1);
 277         }
 278         catch(...) {
 279                 cout << "An unknown error has occurred in the ParsimonyCommand class function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 280                 exit(1);
 281         }
 282 }
 283 /***********************************************************/
 284