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
  55                 if (randomtree == "") {
  56                         //get pscore for users tree
  57                         userData.resize(numComp,0);  //data = AB, AC, BC, ABC.
  58                         randomData.resize(numComp,0);  //data = AB, AC, BC, ABC.
  59                         rscoreFreq.resize(numComp);
  60                         uscoreFreq.resize(numComp);
  61                         rCumul.resize(numComp);
  62                         uCumul.resize(numComp);
  63                         validScores.resize(numComp);
  64                         userTreeScores.resize(numComp);
  65                         UScoreSig.resize(numComp);
  66
  67                         //get pscores for users trees
  68                         for (int i = 0; i < T.size(); i++) {
  69                                 cout << "Processing tree " << i+1 << endl;
  70                                 userData = pars->getValues(T[i]);  //data = AB, AC, BC, ABC.
  71
  72                                 //output scores for each combination
  73                                 for(int k = 0; k < numComp; k++) {
  74                                         cout << "Tree " << i+1 << " Combination " << groupComb[k] << " parsimony score = " << userData[k] << endl;
  75                                         //update uscoreFreq
  76                                         it = uscoreFreq[k].find(userData[k]);
  77                                         if (it == uscoreFreq[k].end()) {//new score
  78                                                 uscoreFreq[k][userData[k]] = 1;
  79                                         }else{ uscoreFreq[k][userData[k]]++; }
  80
  81                                         //add users score to valid scores
  82                                         validScores[k][userData[k]] = userData[k];
  83
  84                                         //save score for summary file
  85                                         userTreeScores[k].push_back(userData[k]);
  86                                 }
  87                         }
  88
  89                         //get pscores for random trees
  90                         for (int j = 0; j < iters; j++) {
  91                                 //create new tree with same num nodes and leaves as users
  92                                 randT = new Tree();
  93                                 //create random relationships between nodes
  94                                 randT->assembleRandomTree();
  95                                 //get pscore of random tree
  96                                 randomData = pars->getValues(randT);
  97
  98                                 for(int r = 0; r < numComp; r++) {
  99                                         //add trees pscore to map of scores
 100                                         it2 = rscoreFreq[r].find(randomData[r]);
 101                                         if (it2 != rscoreFreq[r].end()) {//already have that score
 102                                                 rscoreFreq[r][randomData[r]]++;
 103                                         }else{//first time we have seen this score
 104                                                 rscoreFreq[r][randomData[r]] = 1;
 105                                         }
 106
 107                                         //add randoms score to validscores
 108                                         validScores[r][randomData[r]] = randomData[r];
 109                                 }
 110
 111                                 delete randT;
 112                         }
 113                 }else {
 114                         //get pscores for random trees
 115                         for (int j = 0; j < iters; j++) {
 116                                 //create new tree with same num nodes and leaves as users
 117                                 randT = new Tree();
 118                                 //create random relationships between nodes
 119                                 randT->assembleRandomTree();
 120                                 //get pscore of random tree
 121                                 randomData = pars->getValues(randT);
 122
 123                                 for(int r = 0; r < numComp; r++) {
 124                                         //add trees pscore to map of scores
 125                                         it2 = rscoreFreq[r].find(randomData[r]);
 126                                         if (it2 != rscoreFreq[r].end()) {//already have that score
 127                                                 rscoreFreq[r][randomData[r]]++;
 128                                         }else{//first time we have seen this score
 129                                                 rscoreFreq[r][randomData[r]] = 1;
 130                                         }
 131
 132                                         //add randoms score to validscores
 133                                         validScores[r][randomData[r]] = randomData[r];
 134                                 }
 135
 136                                 delete randT;
 137                         }
 138                 }
 139
 140                 float rcumul = 0.0000;
 141                 float ucumul = 0.0000;
 142
 143                 for(int a = 0; a < numComp; a++) {
 144                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 145                         for (it = validScores[a].begin(); it != validScores[a].end(); it++) {
 146                                 if (randomtree == "") {
 147                                         it2 = uscoreFreq[a].find(it->first);
 148                                         //user data has that score
 149                                         if (it2 != uscoreFreq[a].end()) { uscoreFreq[a][it->first] /= T.size(); ucumul+= it2->second;  }
 150                                         else { uscoreFreq[a][it->first] = 0.0000; } //no user trees with that score
 151                                         //make uCumul map
 152                                         uCumul[a][it->first] = ucumul-a;
 153                                 }
 154
 155                                 //make rscoreFreq map and rCumul
 156                                 it2 = rscoreFreq[a].find(it->first);
 157                                 //get percentage of random trees with that info
 158                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul+= it2->second;  }
 159                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 160                                 rCumul[a][it->first] = rcumul-a;
 161                         }
 162
 163                         //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 164                         for (int h = 0; h < userTreeScores[a].size(); h++) {
 165                                 UScoreSig[a].push_back(rCumul[a][userTreeScores[a][h]]);
 166                         }
 167                 }
 168
 169                 printParsimonyFile();
 170                 printUSummaryFile();
 171
 172                 //reset globaldata's treemap if you just did random distrib
 173                 if (randomtree != "") { globaldata->gTreemap = savetmap; }
 174
 175                 //reset randomTree parameter to ""
 176                 globaldata->setRandomTree("");
 177                 //reset groups parameter
 178                 globaldata->Groups.clear();  globaldata->setGroups("");
 179
 180                 return 0;
 181
 182         }
 183         catch(exception& e) {
 184                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187         catch(...) {
 188                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 189                 exit(1);
 190         }
 191 }
 192
 193 /***********************************************************/
 194 void ParsimonyCommand::printParsimonyFile() {
 195         try {
 196                 //column headers
 197                 if (randomtree == "") {
 198                         out << "Comb" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 199                 }else {
 200                         out << "Comb" << '\t' << "Score" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 201                 }
 202
 203                 //format output
 204                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 205
 206                 for(int a = 0; a < numComp; a++) {
 207                         //print each line
 208                         for (it = validScores[a].begin(); it != validScores[a].end(); it++) {
 209                                 if (randomtree == "") {
 210                                         out << setprecision(6)  << groupComb[a] << '\t' << it->first << '\t' << '\t'<< uscoreFreq[a][it->first] << '\t' << uCumul[a][it->first] << '\t' << rscoreFreq[a][it->first] << '\t' << rCumul[a][it->first] << endl;
 211                                 }else{
 212                                         out << setprecision(6) << groupComb[a] << '\t' << it->first << '\t' << '\t' << rscoreFreq[a][it->first] << '\t' << rCumul[a][it->first] << endl;
 213                                 }
 214                         }
 215                 }
 216                 out.close();
 217
 218         }
 219         catch(exception& e) {
 220                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 221                 exit(1);
 222         }
 223         catch(...) {
 224                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 225                 exit(1);
 226         }
 227 }
 228 /***********************************************************/
 229 void ParsimonyCommand::printUSummaryFile() {
 230         try {
 231                 //column headers
 232                 outSum << "Tree#" << '\t' << "Comb" << '\t'  <<  "ParsScore" << '\t' << '\t' << "ParsSig" <<  endl;
 233
 234                 //format output
 235                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 236
 237
 238                 //print each line
 239                 for (int i = 0; i< T.size(); i++) {
 240                         for(int a = 0; a < numComp; a++) {
 241                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << '\t' << userTreeScores[a][i] << '\t' << UScoreSig[a][i] << endl;
 242                         }
 243                 }
 244
 245                 outSum.close();
 246         }
 247         catch(exception& e) {
 248                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 249                 exit(1);
 250         }
 251         catch(...) {
 252                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 253                 exit(1);
 254         }
 255 }
 256
 257 /***********************************************************/
 258 void ParsimonyCommand::getUserInput() {
 259         try {
 260
 261                 //create treemap
 262                 tmap = new TreeMap();
 263
 264                 cout << "Please enter the number of groups you would like to analyze: ";
 265                 cin >> numGroups;
 266
 267                 int num, count;
 268                 count = 1;
 269                 numEachGroup.resize(numGroups, 0);
 270
 271                 for (int i = 1; i <= numGroups; i++) {
 272                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 273                         cin >> num;
 274
 275                         //set tmaps seqsPerGroup
 276                         tmap->seqsPerGroup[toString(i)] = num;
 277
 278                         //set tmaps namesOfSeqs
 279                         for (int j = 0; j < num; j++) {
 280                                 tmap->namesOfSeqs.push_back(toString(count));
 281                                 tmap->treemap[toString(count)].groupname = toString(i);
 282                                 count++;
 283                         }
 284                 }
 285
 286                 //clears buffer so next command doesn't have error
 287                 string s;
 288                 getline(cin, s);
 289
 290                 //save tmap for later
 291                 globaldata->gTreemap = tmap;
 292
 293         }
 294         catch(exception& e) {
 295                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 296                 exit(1);
 297         }
 298         catch(...) {
 299                 cout << "An unknown error has occurred in the ParsimonyCommand class function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 300                 exit(1);
 301         }
 302 }
 303 /***********************************************************/
 304
 305 void ParsimonyCommand::setGroups() {
 306         try {
 307                 string allGroups = "";
 308                 numGroups = 0;
 309                 //if the user has not entered specific groups to analyze then do them all
 310                 if (globaldata->Groups.size() != 0) {
 311                         if (globaldata->Groups[0] != "all") {
 312                                 //check that groups are valid
 313                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 314                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 315                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 316                                                 // erase the invalid group from globaldata->Groups
 317                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 318                                         }
 319                                 }
 320
 321                                 //if the user only entered invalid groups
 322                                 if (globaldata->Groups.size() == 0) {
 323                                         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;
 324                                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 325                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 326                                                 numGroups++;
 327                                                 allGroups += tmap->namesOfGroups[i];
 328                                         }
 329                                 }else {
 330                                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 331                                                 allGroups += tmap->namesOfGroups[i];
 332                                                 numGroups++;
 333                                         }
 334                                 }
 335                         }else{//user has enter "all" and wants the default groups
 336                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 337                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 338                                         numGroups++;
 339                                         allGroups += tmap->namesOfGroups[i];
 340                                 }
 341                                 globaldata->setGroups("");
 342                         }
 343                 }else {
 344                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 345                                 allGroups += tmap->namesOfGroups[i];
 346                         }
 347                         numGroups = 1;
 348                 }
 349
 350                 //calculate number of comparsions
 351                 numComp = 0;
 352                 for (int r=0; r<numGroups; r++) {
 353                         for (int l = r+1; l < numGroups; l++) {
 354                                 groupComb.push_back(globaldata->Groups[r]+globaldata->Groups[l]);
 355                                 numComp++;
 356                         }
 357                 }
 358
 359                 //ABC
 360                 groupComb.push_back(allGroups);
 361                 numComp++;
 362
 363         }
 364         catch(exception& e) {
 365                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 366                 exit(1);
 367         }
 368         catch(...) {
 369                 cout << "An unknown error has occurred in the ParsimonyCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 370                 exit(1);
 371         }
 372
 373 }
 374 /*****************************************************************/
 375
 376