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                         //if the user has not entered specific groups to analyze then do them all
  32                         if (globaldata->Groups.size() != 0) {
  33                                 //check that groups are valid
  34                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
  35                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
  36                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
  37                                                 // erase the invalid group from globaldata->Groups
  38                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
  39                                         }
  40                                 }
  41
  42                                 //if the user only entered invalid groups
  43                                 if (globaldata->Groups.size() == 0) {
  44                                         cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl;
  45                                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
  46                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
  47                                         }
  48                                 }
  49                         }else {
  50                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
  51                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
  52                                 }
  53                         }
  54
  55                 }else { //user wants random distribution
  56                         savetmap = globaldata->gTreemap;
  57                         getUserInput();
  58                         parsFile = randomtree + ".rd_parsimony";
  59                         openOutputFile(parsFile, out);
  60                 }
  61
  62                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  63                 pars = new Parsimony(tmap);
  64
  65         }
  66         catch(exception& e) {
  67                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  68                 exit(1);
  69         }
  70         catch(...) {
  71                 cout << "An unknown error has occurred in the ParsimonyCommand class function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  72                 exit(1);
  73         }
  74 }
  75 /***********************************************************/
  76 int ParsimonyCommand::execute() {
  77         try {
  78
  79                 //get pscore for users tree
  80                 userData.resize(1,0);  //data[0] = pscore.
  81                 randomData.resize(1,0);  //data[0] = pscore.
  82
  83                 //format output
  84                 outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
  85                 outDist << "RandomTree#" << '\t' << "ParsScore" << endl;
  86
  87                 if (randomtree == "") {
  88                         //get pscores for users trees
  89                         for (int i = 0; i < T.size(); i++) {
  90                                 cout << "Processing tree " << i+1 << endl;
  91                                 userData = pars->getValues(T[i]);  //userData[0] = pscore
  92                                 //update uscoreFreq
  93                                 it = uscoreFreq.find(userData[0]);
  94                                 if (it == uscoreFreq.end()) {//new score
  95                                         uscoreFreq[userData[0]] = 1;
  96                                 }else{ uscoreFreq[userData[0]]++; }
  97
  98                                 //add users score to valid scores
  99                                 validScores[userData[0]] = userData[0];
 100
 101                                 //save score for summary file
 102                                 userTreeScores.push_back(userData[0]);
 103
 104                         }
 105
 106                         //get pscores for random trees
 107                         for (int j = 0; j < iters; j++) {
 108                                 //create new tree with same num nodes and leaves as users
 109                                 randT = new Tree();
 110                                 //create random relationships between nodes
 111                                 randT->assembleRandomTree();
 112                                 //get pscore of random tree
 113                                 randomData = pars->getValues(randT);
 114
 115                                 //add trees pscore to map of scores
 116                                 it2 = rscoreFreq.find(randomData[0]);
 117                                 if (it2 != rscoreFreq.end()) {//already have that score
 118                                         rscoreFreq[randomData[0]]++;
 119                                 }else{//first time we have seen this score
 120                                         rscoreFreq[randomData[0]] = 1;
 121                                 }
 122
 123                                 //add randoms score to validscores
 124                                 validScores[randomData[0]] = randomData[0];
 125
 126                                 //output info to pdistrib file
 127                                 outDist << j+1 << '\t'<< '\t' << randomData[0] << endl;
 128
 129                                 delete randT;
 130                         }
 131                 }else {
 132                         //get pscores for random trees
 133                         for (int j = 0; j < iters; j++) {
 134                                 //create new tree with same num nodes and leaves as users
 135                                 randT = new Tree();
 136                                 //create random relationships between nodes
 137                                 randT->assembleRandomTree();
 138                                 //get pscore of random tree
 139                                 randomData = pars->getValues(randT);
 140
 141                                 //add trees pscore to map of scores
 142                                 it2 = rscoreFreq.find(randomData[0]);
 143                                 if (it2 != rscoreFreq.end()) {//already have that score
 144                                         rscoreFreq[randomData[0]]++;
 145                                 }else{//first time we have seen this score
 146                                         rscoreFreq[randomData[0]] = 1;
 147                                 }
 148
 149                                 //add randoms score to validscores
 150                                 validScores[randomData[0]] = randomData[0];
 151
 152                                 delete randT;
 153                         }
 154                 }
 155
 156                 float rcumul = 0.0000;
 157                 float ucumul = 0.0000;
 158
 159                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 160                 for (it = validScores.begin(); it != validScores.end(); it++) {
 161                         if (randomtree == "") {
 162                                 it2 = uscoreFreq.find(it->first);
 163                                 //user data has that score
 164                                 if (it2 != uscoreFreq.end()) { uscoreFreq[it->first] /= T.size(); ucumul+= it2->second;  }
 165                                 else { uscoreFreq[it->first] = 0.0000; } //no user trees with that score
 166                                 //make uCumul map
 167                                 uCumul[it->first] = ucumul;
 168                         }
 169
 170                         //make rscoreFreq map and rCumul
 171                         it2 = rscoreFreq.find(it->first);
 172                         //get percentage of random trees with that info
 173                         if (it2 != rscoreFreq.end()) {  rscoreFreq[it->first] /= iters; rcumul+= it2->second;  }
 174                         else { rscoreFreq[it->first] = 0.0000; } //no random trees with that score
 175                         rCumul[it->first] = rcumul;
 176                 }
 177
 178                 //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 179                 for (int h = 0; h < userTreeScores.size(); h++) {
 180                         UScoreSig.push_back(rCumul[userTreeScores[h]]);
 181                 }
 182
 183                 printParsimonyFile();
 184                 printUSummaryFile();
 185
 186                 //reset globaldata's treemap if you just did random distrib
 187                 if (randomtree != "") { globaldata->gTreemap = savetmap; }
 188
 189                 //reset randomTree parameter to ""
 190                 globaldata->setRandomTree("");
 191                 //reset groups parameter
 192                 globaldata->Groups.clear();
 193
 194                 return 0;
 195
 196         }
 197         catch(exception& e) {
 198                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 199                 exit(1);
 200         }
 201         catch(...) {
 202                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 203                 exit(1);
 204         }
 205 }
 206
 207 /***********************************************************/
 208 void ParsimonyCommand::printParsimonyFile() {
 209         try {
 210                 //column headers
 211                 if (randomtree == "") {
 212                         out << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 213                 }else {
 214                         out << "Score" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 215                 }
 216
 217                 //format output
 218                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 219
 220                 //print each line
 221                 for (it = validScores.begin(); it != validScores.end(); it++) {
 222                         if (randomtree == "") {
 223                                 out << setprecision(6) << it->first << '\t' << '\t' << uscoreFreq[it->first] << '\t' << uCumul[it->first] << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 224                         }else{
 225                                 out << setprecision(6) << it->first << '\t' << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 226                         }
 227                 }
 228
 229                 out.close();
 230
 231         }
 232         catch(exception& e) {
 233                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 234                 exit(1);
 235         }
 236         catch(...) {
 237                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 238                 exit(1);
 239         }
 240 }
 241 /***********************************************************/
 242 void ParsimonyCommand::printUSummaryFile() {
 243         try {
 244                 //column headers
 245                 outSum << "Tree#" << '\t'  <<  "ParsScore" << '\t' << '\t' << "ParsSig" <<  endl;
 246
 247                 //format output
 248                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 249
 250                 //print each line
 251                 for (int i = 0; i< T.size(); i++) {
 252                         outSum << setprecision(6) << i+1 << '\t' << '\t' << userTreeScores[i] << '\t' << UScoreSig[i] << endl;
 253                 }
 254
 255                 outSum.close();
 256         }
 257         catch(exception& e) {
 258                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 259                 exit(1);
 260         }
 261         catch(...) {
 262                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 263                 exit(1);
 264         }
 265 }
 266
 267 /***********************************************************/
 268 void ParsimonyCommand::getUserInput() {
 269         try {
 270
 271                 //create treemap
 272                 tmap = new TreeMap();
 273
 274                 cout << "Please enter the number of groups you would like to analyze: ";
 275                 cin >> numGroups;
 276
 277                 int num, count;
 278                 count = 1;
 279                 numEachGroup.resize(numGroups, 0);
 280
 281                 for (int i = 1; i <= numGroups; i++) {
 282                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 283                         cin >> num;
 284
 285                         //set tmaps seqsPerGroup
 286                         tmap->seqsPerGroup[toString(i)] = num;
 287
 288                         //set tmaps namesOfSeqs
 289                         for (int j = 0; j < num; j++) {
 290                                 tmap->namesOfSeqs.push_back(toString(count));
 291                                 tmap->treemap[toString(count)].groupname = toString(i);
 292                                 count++;
 293                         }
 294                 }
 295
 296                 //clears buffer so next command doesn't have error
 297                 string s;
 298                 getline(cin, s);
 299
 300                 //save tmap for later
 301                 globaldata->gTreemap = tmap;
 302
 303         }
 304         catch(exception& e) {
 305                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 306                 exit(1);
 307         }
 308         catch(...) {
 309                 cout << "An unknown error has occurred in the ParsimonyCommand class function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 310                 exit(1);
 311         }
 312 }
 313 /***********************************************************/
 314