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                 randomtree = globaldata->getRandomTree();
  19
  20                 //user has entered their own tree
  21                 if (randomtree == "") {
  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                         savetmap = globaldata->gTreemap;
  33                         getUserInput();
  34                         parsFile = randomtree + ".rd_parsimony";
  35                         openOutputFile(parsFile, out);
  36                 }
  37
  38                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  39                 pars = new Parsimony(tmap);
  40
  41         }
  42         catch(exception& e) {
  43                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  44                 exit(1);
  45         }
  46         catch(...) {
  47                 cout << "An unknown error has occurred in the ParsimonyCommand class function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  48                 exit(1);
  49         }
  50 }
  51 /***********************************************************/
  52 int ParsimonyCommand::execute() {
  53         try {
  54
  55                 //get pscore for users tree
  56                 userData.resize(1,0);  //data[0] = pscore.
  57                 randomData.resize(1,0);  //data[0] = pscore.
  58
  59                 //format output
  60                 outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
  61                 outDist << "RandomTree#" << '\t' << "ParsScore" << endl;
  62
  63                 if (randomtree == "") {
  64                         //get pscores for users trees
  65                         for (int i = 0; i < T.size(); i++) {
  66                                 cout << "Processing tree " << i+1 << endl;
  67                                 userData = pars->getValues(T[i]);  //userData[0] = pscore
  68                                 //update uscoreFreq
  69                                 it = uscoreFreq.find(userData[0]);
  70                                 if (it == uscoreFreq.end()) {//new score
  71                                         uscoreFreq[userData[0]] = 1;
  72                                 }else{ uscoreFreq[userData[0]]++; }
  73
  74                                 //add users score to valid scores
  75                                 validScores[userData[0]] = userData[0];
  76
  77                                 //save score for summary file
  78                                 userTreeScores.push_back(userData[0]);
  79
  80                         }
  81
  82                         //get pscores for random trees
  83                         for (int j = 0; j < iters; j++) {
  84                                 //create new tree with same num nodes and leaves as users
  85                                 randT = new Tree();
  86                                 //create random relationships between nodes
  87                                 randT->assembleRandomTree();
  88                                 //get pscore of random tree
  89                                 randomData = pars->getValues(randT);
  90
  91                                 //add trees pscore to map of scores
  92                                 it2 = rscoreFreq.find(randomData[0]);
  93                                 if (it2 != rscoreFreq.end()) {//already have that score
  94                                         rscoreFreq[randomData[0]]++;
  95                                 }else{//first time we have seen this score
  96                                         rscoreFreq[randomData[0]] = 1;
  97                                 }
  98
  99                                 //add randoms score to validscores
 100                                 validScores[randomData[0]] = randomData[0];
 101
 102                                 //output info to pdistrib file
 103                                 outDist << j+1 << '\t'<< '\t' << randomData[0] << endl;
 104
 105                                 delete randT;
 106                         }
 107                 }else {
 108                         //get pscores for random trees
 109                         for (int j = 0; j < iters; j++) {
 110                                 //create new tree with same num nodes and leaves as users
 111                                 randT = new Tree();
 112                                 //create random relationships between nodes
 113                                 randT->assembleRandomTree();
 114                                 //get pscore of random tree
 115                                 randomData = pars->getValues(randT);
 116
 117                                 //add trees pscore to map of scores
 118                                 it2 = rscoreFreq.find(randomData[0]);
 119                                 if (it2 != rscoreFreq.end()) {//already have that score
 120                                         rscoreFreq[randomData[0]]++;
 121                                 }else{//first time we have seen this score
 122                                         rscoreFreq[randomData[0]] = 1;
 123                                 }
 124
 125                                 //add randoms score to validscores
 126                                 validScores[randomData[0]] = randomData[0];
 127
 128                                 delete randT;
 129                         }
 130                 }
 131
 132                 float rcumul = 0.0000;
 133                 float ucumul = 0.0000;
 134
 135                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 136                 for (it = validScores.begin(); it != validScores.end(); it++) {
 137                         if (randomtree == "") {
 138                                 it2 = uscoreFreq.find(it->first);
 139                                 //user data has that score
 140                                 if (it2 != uscoreFreq.end()) { uscoreFreq[it->first] /= T.size(); ucumul+= it2->second;  }
 141                                 else { uscoreFreq[it->first] = 0.0000; } //no user trees with that score
 142                                 //make uCumul map
 143                                 uCumul[it->first] = ucumul;
 144                         }
 145
 146                         //make rscoreFreq map and rCumul
 147                         it2 = rscoreFreq.find(it->first);
 148                         //get percentage of random trees with that info
 149                         if (it2 != rscoreFreq.end()) {  rscoreFreq[it->first] /= iters; rcumul+= it2->second;  }
 150                         else { rscoreFreq[it->first] = 0.0000; } //no random trees with that score
 151                         rCumul[it->first] = rcumul;
 152                 }
 153
 154                 //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 155                 for (int h = 0; h < userTreeScores.size(); h++) {
 156                         UScoreSig.push_back(rCumul[userTreeScores[h]]);
 157                 }
 158
 159                 printParsimonyFile();
 160                 printUSummaryFile();
 161
 162                 //reset globaldata's treemap if you just did random distrib
 163                 if (randomtree != "") { globaldata->gTreemap = savetmap; }
 164
 165                 //reset randomTree parameter to ""
 166                 globaldata->setRandomTree("");
 167
 168                 return 0;
 169
 170         }
 171         catch(exception& e) {
 172                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 173                 exit(1);
 174         }
 175         catch(...) {
 176                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 177                 exit(1);
 178         }
 179 }
 180
 181 /***********************************************************/
 182 void ParsimonyCommand::printParsimonyFile() {
 183         try {
 184                 //column headers
 185                 if (randomtree == "") {
 186                         out << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 187                 }else {
 188                         out << "Score" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 189                 }
 190
 191                 //format output
 192                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 193
 194                 //print each line
 195                 for (it = validScores.begin(); it != validScores.end(); it++) {
 196                         if (randomtree == "") {
 197                                 out << setprecision(6) << it->first << '\t' << '\t' << uscoreFreq[it->first] << '\t' << uCumul[it->first] << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 198                         }else{
 199                                 out << setprecision(6) << it->first << '\t' << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 200                         }
 201                 }
 202
 203                 out.close();
 204
 205         }
 206         catch(exception& e) {
 207                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 208                 exit(1);
 209         }
 210         catch(...) {
 211                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 212                 exit(1);
 213         }
 214 }
 215 /***********************************************************/
 216 void ParsimonyCommand::printUSummaryFile() {
 217         try {
 218                 //column headers
 219                 outSum << "Tree#" << '\t'  <<  "ParsScore" << '\t' << '\t' << "ParsSig" <<  endl;
 220
 221                 //format output
 222                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 223
 224                 //print each line
 225                 for (int i = 0; i< T.size(); i++) {
 226                         outSum << setprecision(6) << i+1 << '\t' << '\t' << userTreeScores[i] << '\t' << UScoreSig[i] << endl;
 227                 }
 228
 229                 outSum.close();
 230         }
 231         catch(exception& e) {
 232                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 233                 exit(1);
 234         }
 235         catch(...) {
 236                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 237                 exit(1);
 238         }
 239 }
 240
 241 /***********************************************************/
 242 void ParsimonyCommand::getUserInput() {
 243         try {
 244
 245                 //create treemap
 246                 tmap = new TreeMap();
 247
 248                 cout << "Please enter the number of groups you would like to analyze: ";
 249                 cin >> numGroups;
 250
 251                 int num, count;
 252                 count = 1;
 253                 numEachGroup.resize(numGroups, 0);
 254
 255                 for (int i = 1; i <= numGroups; i++) {
 256                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 257                         cin >> num;
 258
 259                         //set tmaps seqsPerGroup
 260                         tmap->seqsPerGroup[toString(i)] = num;
 261
 262                         //set tmaps namesOfSeqs
 263                         for (int j = 0; j < num; j++) {
 264                                 tmap->namesOfSeqs.push_back(toString(count));
 265                                 tmap->treemap[toString(count)].groupname = toString(i);
 266                                 count++;
 267                         }
 268                 }
 269
 270                 //clears buffer so next command doesn't have error
 271                 string s;
 272                 getline(cin, s);
 273
 274                 //save tmap for later
 275                 globaldata->gTreemap = tmap;
 276
 277         }
 278         catch(exception& e) {
 279                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 280                 exit(1);
 281         }
 282         catch(...) {
 283                 cout << "An unknown error has occurred in the ParsimonyCommand class function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 284                 exit(1);
 285         }
 286 }
 287 /***********************************************************/
 288