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                         //set users groups to analyze
  29                         setGroups();
  30                 }else { //user wants random distribution
  31                         savetmap = globaldata->gTreemap;
  32                         getUserInput();
  33                         parsFile = randomtree + ".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                 if (randomtree == "") {
  59                         copyUserTree = new Tree();
  60                         //get pscores for users trees
  61                         for (int i = 0; i < T.size(); i++) {
  62                                 //copy users tree so that you can redo pgroups
  63                                 copyUserTree->getCopy(T[i]);
  64                                 cout << "Processing tree " << i+1 << endl;
  65                                 userData = pars->getValues(copyUserTree);  //userData[0] = pscore
  66                                 cout << "Tree " << i+1 << " parsimony score = " << userData[0] << endl;
  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                                 delete randT;
 102                         }
 103                 }else {
 104                         //get pscores for random trees
 105                         for (int j = 0; j < iters; j++) {
 106                                 //create new tree with same num nodes and leaves as users
 107                                 randT = new Tree();
 108                                 //create random relationships between nodes
 109                                 randT->assembleRandomTree();
 110                                 //get pscore of random tree
 111                                 randomData = pars->getValues(randT);
 112
 113                                 //add trees pscore to map of scores
 114                                 it2 = rscoreFreq.find(randomData[0]);
 115                                 if (it2 != rscoreFreq.end()) {//already have that score
 116                                         rscoreFreq[randomData[0]]++;
 117                                 }else{//first time we have seen this score
 118                                         rscoreFreq[randomData[0]] = 1;
 119                                 }
 120
 121                                 //add randoms score to validscores
 122                                 validScores[randomData[0]] = randomData[0];
 123
 124                                 delete randT;
 125                         }
 126                 }
 127
 128                 float rcumul = 0.0000;
 129                 float ucumul = 0.0000;
 130
 131                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 132                 for (it = validScores.begin(); it != validScores.end(); it++) {
 133                         if (randomtree == "") {
 134                                 it2 = uscoreFreq.find(it->first);
 135                                 //user data has that score
 136                                 if (it2 != uscoreFreq.end()) { uscoreFreq[it->first] /= T.size(); ucumul+= it2->second;  }
 137                                 else { uscoreFreq[it->first] = 0.0000; } //no user trees with that score
 138                                 //make uCumul map
 139                                 uCumul[it->first] = ucumul;
 140                         }
 141
 142                         //make rscoreFreq map and rCumul
 143                         it2 = rscoreFreq.find(it->first);
 144                         //get percentage of random trees with that info
 145                         if (it2 != rscoreFreq.end()) {  rscoreFreq[it->first] /= iters; rcumul+= it2->second;  }
 146                         else { rscoreFreq[it->first] = 0.0000; } //no random trees with that score
 147                         rCumul[it->first] = rcumul;
 148                 }
 149
 150                 //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 151                 for (int h = 0; h < userTreeScores.size(); h++) {
 152                         UScoreSig.push_back(rCumul[userTreeScores[h]]);
 153                 }
 154
 155                 printParsimonyFile();
 156                 printUSummaryFile();
 157
 158                 //reset globaldata's treemap if you just did random distrib
 159                 if (randomtree != "") { globaldata->gTreemap = savetmap; }
 160
 161                 //reset randomTree parameter to ""
 162                 globaldata->setRandomTree("");
 163                 //reset groups parameter
 164                 globaldata->Groups.clear();  globaldata->setGroups("");
 165
 166                 return 0;
 167
 168         }
 169         catch(exception& e) {
 170                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 171                 exit(1);
 172         }
 173         catch(...) {
 174                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 175                 exit(1);
 176         }
 177 }
 178
 179 /***********************************************************/
 180 void ParsimonyCommand::printParsimonyFile() {
 181         try {
 182                 //column headers
 183                 if (randomtree == "") {
 184                         out << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 185                 }else {
 186                         out << "Score" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 187                 }
 188
 189                 //format output
 190                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 191
 192                 //print each line
 193                 for (it = validScores.begin(); it != validScores.end(); it++) {
 194                         if (randomtree == "") {
 195                                 out << setprecision(6) << it->first << '\t' << '\t' << uscoreFreq[it->first] << '\t' << uCumul[it->first] << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 196                         }else{
 197                                 out << setprecision(6) << it->first << '\t' << '\t' << rscoreFreq[it->first] << '\t' << rCumul[it->first] << endl;
 198                         }
 199                 }
 200
 201                 out.close();
 202
 203         }
 204         catch(exception& e) {
 205                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 206                 exit(1);
 207         }
 208         catch(...) {
 209                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 210                 exit(1);
 211         }
 212 }
 213 /***********************************************************/
 214 void ParsimonyCommand::printUSummaryFile() {
 215         try {
 216                 //column headers
 217                 outSum << "Tree#" << '\t'  <<  "ParsScore" << '\t' << '\t' << "ParsSig" <<  endl;
 218
 219                 //format output
 220                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 221
 222                 //print each line
 223                 for (int i = 0; i< T.size(); i++) {
 224                         outSum << setprecision(6) << i+1 << '\t' << '\t' << userTreeScores[i] << '\t' << UScoreSig[i] << endl;
 225                 }
 226
 227                 outSum.close();
 228         }
 229         catch(exception& e) {
 230                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 231                 exit(1);
 232         }
 233         catch(...) {
 234                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 235                 exit(1);
 236         }
 237 }
 238
 239 /***********************************************************/
 240 void ParsimonyCommand::getUserInput() {
 241         try {
 242
 243                 //create treemap
 244                 tmap = new TreeMap();
 245
 246                 cout << "Please enter the number of groups you would like to analyze: ";
 247                 cin >> numGroups;
 248
 249                 int num, count;
 250                 count = 1;
 251                 numEachGroup.resize(numGroups, 0);
 252
 253                 for (int i = 1; i <= numGroups; i++) {
 254                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 255                         cin >> num;
 256
 257                         //set tmaps seqsPerGroup
 258                         tmap->seqsPerGroup[toString(i)] = num;
 259
 260                         //set tmaps namesOfSeqs
 261                         for (int j = 0; j < num; j++) {
 262                                 tmap->namesOfSeqs.push_back(toString(count));
 263                                 tmap->treemap[toString(count)].groupname = toString(i);
 264                                 count++;
 265                         }
 266                 }
 267
 268                 //clears buffer so next command doesn't have error
 269                 string s;
 270                 getline(cin, s);
 271
 272                 //save tmap for later
 273                 globaldata->gTreemap = tmap;
 274
 275         }
 276         catch(exception& e) {
 277                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 278                 exit(1);
 279         }
 280         catch(...) {
 281                 cout << "An unknown error has occurred in the ParsimonyCommand class function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 282                 exit(1);
 283         }
 284 }
 285 /***********************************************************/
 286
 287 void ParsimonyCommand::setGroups() {
 288         try {
 289                 //if the user has not entered specific groups to analyze then do them all
 290                 if (globaldata->Groups.size() != 0) {
 291                         if (globaldata->Groups[0] != "all") {
 292                                 //check that groups are valid
 293                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 294                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 295                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 296                                                 // erase the invalid group from globaldata->Groups
 297                                                 globaldata->Groups.erase (globaldata->Groups.begin()+i);
 298                                         }
 299                                 }
 300
 301                                 //if the user only entered invalid groups
 302                                 if (globaldata->Groups.size() == 0) {
 303                                         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;
 304                                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 305                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 306                                         }
 307                                 }
 308                         }else{//user has enter "all" and wants the default groups
 309                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 310                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 311                                 }
 312                                 globaldata->setGroups("");
 313                         }
 314                 }else {
 315                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 316                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 317                         }
 318                 }
 319         }
 320         catch(exception& e) {
 321                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 322                 exit(1);
 323         }
 324         catch(...) {
 325                 cout << "An unknown error has occurred in the ParsimonyCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 326                 exit(1);
 327         }
 328
 329 }
 330 /*****************************************************************/
 331
 332