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                 }else { //user wants random distribution
  29                         savetmap = globaldata->gTreemap;
  30                         getUserInput();
  31                         parsFile = randomtree + ".rd_parsimony";
  32                         openOutputFile(parsFile, out);
  33                 }
  34
  35                 //set users groups to analyze
  36                 setGroups();
  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(numComp,0);  //data = AB, AC, BC, ABC.
  56                 randomData.resize(numComp,0);  //data = AB, AC, BC, ABC.
  57                 rscoreFreq.resize(numComp);
  58                 uscoreFreq.resize(numComp);
  59                 rCumul.resize(numComp);
  60                 uCumul.resize(numComp);
  61                 validScores.resize(numComp);
  62                 userTreeScores.resize(numComp);
  63                 UScoreSig.resize(numComp);
  64
  65                 if (randomtree == "") {
  66                         //get pscores for users trees
  67                         for (int i = 0; i < T.size(); i++) {
  68                                 cout << "Processing tree " << i+1 << endl;
  69                                 userData = pars->getValues(T[i]);  //data = AB, AC, BC, ABC.
  70
  71                                 //output scores for each combination
  72                                 for(int k = 0; k < numComp; k++) {
  73                                         //update uscoreFreq
  74                                         it = uscoreFreq[k].find(userData[k]);
  75                                         if (it == uscoreFreq[k].end()) {//new score
  76                                                 uscoreFreq[k][userData[k]] = 1;
  77                                         }else{ uscoreFreq[k][userData[k]]++; }
  78
  79                                         //add users score to valid scores
  80                                         validScores[k][userData[k]] = userData[k];
  81
  82                                         //save score for summary file
  83                                         userTreeScores[k].push_back(userData[k]);
  84                                 }
  85                         }
  86
  87                         //get pscores for random trees
  88                         for (int j = 0; j < iters; j++) {
  89                                 //create new tree with same num nodes and leaves as users
  90                                 randT = new Tree();
  91                                 //create random relationships between nodes
  92                                 randT->assembleRandomTree();
  93                                 //get pscore of random tree
  94                                 randomData = pars->getValues(randT);
  95
  96                                 for(int r = 0; r < numComp; r++) {
  97                                         //add trees pscore to map of scores
  98                                         it2 = rscoreFreq[r].find(randomData[r]);
  99                                         if (it2 != rscoreFreq[r].end()) {//already have that score
 100                                                 rscoreFreq[r][randomData[r]]++;
 101                                         }else{//first time we have seen this score
 102                                                 rscoreFreq[r][randomData[r]] = 1;
 103                                         }
 104
 105                                         //add randoms score to validscores
 106                                         validScores[r][randomData[r]] = randomData[r];
 107                                 }
 108
 109                                 delete randT;
 110                         }
 111                 }else {
 112                         //get pscores for random trees
 113                         for (int j = 0; j < iters; j++) {
 114                                 //create new tree with same num nodes and leaves as users
 115                                 randT = new Tree();
 116                                 //create random relationships between nodes
 117                                 randT->assembleRandomTree();
 118                                 //get pscore of random tree
 119                                 randomData = pars->getValues(randT);
 120
 121                                 for(int r = 0; r < numComp; r++) {
 122                                         //add trees pscore to map of scores
 123                                         it2 = rscoreFreq[r].find(randomData[r]);
 124                                         if (it2 != rscoreFreq[r].end()) {//already have that score
 125                                                 rscoreFreq[r][randomData[r]]++;
 126                                         }else{//first time we have seen this score
 127                                                 rscoreFreq[r][randomData[r]] = 1;
 128                                         }
 129
 130                                         //add randoms score to validscores
 131                                         validScores[r][randomData[r]] = randomData[r];
 132                                 }
 133
 134                                 delete randT;
 135                         }
 136                 }
 137
 138                 float rcumul = 0.0000;
 139                 float ucumul = 0.0000;
 140
 141                 for(int a = 0; a < numComp; a++) {
 142                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 143                         for (it = validScores[a].begin(); it != validScores[a].end(); it++) {
 144                                 if (randomtree == "") {
 145                                         it2 = uscoreFreq[a].find(it->first);
 146                                         //user data has that score
 147                                         if (it2 != uscoreFreq[a].end()) { uscoreFreq[a][it->first] /= T.size(); ucumul+= it2->second;  }
 148                                         else { uscoreFreq[a][it->first] = 0.0000; } //no user trees with that score
 149                                         //make uCumul map
 150                                         uCumul[a][it->first] = ucumul-a;
 151                                 }
 152
 153                                 //make rscoreFreq map and rCumul
 154                                 it2 = rscoreFreq[a].find(it->first);
 155                                 //get percentage of random trees with that info
 156                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul+= it2->second;  }
 157                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 158                                 rCumul[a][it->first] = rcumul-a;
 159                         }
 160
 161                         //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 162                         for (int h = 0; h < userTreeScores[a].size(); h++) {
 163                                 UScoreSig[a].push_back(rCumul[a][userTreeScores[a][h]]);
 164                         }
 165                 }
 166
 167                 printParsimonyFile();
 168                 if (randomtree == "") { printUSummaryFile(); }
 169
 170                 //reset globaldata's treemap if you just did random distrib
 171                 if (randomtree != "") { globaldata->gTreemap = savetmap; }
 172
 173                 //reset randomTree parameter to ""
 174                 globaldata->setRandomTree("");
 175                 //reset groups parameter
 176                 globaldata->Groups.clear();  globaldata->setGroups("");
 177
 178                 return 0;
 179
 180         }
 181         catch(exception& e) {
 182                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 183                 exit(1);
 184         }
 185         catch(...) {
 186                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 187                 exit(1);
 188         }
 189 }
 190
 191 /***********************************************************/
 192 void ParsimonyCommand::printParsimonyFile() {
 193         try {
 194                 //column headers
 195                 if (randomtree == "") {
 196                         out << "Comb" << '\t' << "Score" << '\t' << "UserFreq" << '\t' << "UserCumul" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 197                 }else {
 198                         out << "Comb" << '\t' << "Score" << '\t' << "RandFreq" << '\t' << "RandCumul" << endl;
 199                 }
 200
 201                 //format output
 202                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 203
 204                 for(int a = 0; a < numComp; a++) {
 205                         //print each line
 206                         for (it = validScores[a].begin(); it != validScores[a].end(); it++) {
 207                                 if (randomtree == "") {
 208                                         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;
 209                                 }else{
 210                                         out << setprecision(6) << groupComb[a] << '\t' << it->first << '\t' << '\t' << rscoreFreq[a][it->first] << '\t' << rCumul[a][it->first] << endl;
 211                                 }
 212                         }
 213                 }
 214                 out.close();
 215
 216         }
 217         catch(exception& e) {
 218                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 219                 exit(1);
 220         }
 221         catch(...) {
 222                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 223                 exit(1);
 224         }
 225 }
 226 /***********************************************************/
 227 void ParsimonyCommand::printUSummaryFile() {
 228         try {
 229                 //column headers
 230                 outSum << "Tree#" << '\t' << "Comb" << '\t'  <<  "ParsScore" << '\t' << '\t' << "ParsSig" <<  endl;
 231
 232                 //format output
 233                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 234
 235
 236                 //print each line
 237                 for (int i = 0; i< T.size(); i++) {
 238                         for(int a = 0; a < numComp; a++) {
 239                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << '\t' << userTreeScores[a][i] << '\t' << UScoreSig[a][i] << endl;
 240                                 cout << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << '\t' << userTreeScores[a][i] << '\t' << UScoreSig[a][i] << endl;
 241                         }
 242                 }
 243
 244                 outSum.close();
 245         }
 246         catch(exception& e) {
 247                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 248                 exit(1);
 249         }
 250         catch(...) {
 251                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 252                 exit(1);
 253         }
 254 }
 255
 256 /***********************************************************/
 257 void ParsimonyCommand::getUserInput() {
 258         try {
 259
 260                 //create treemap
 261                 tmap = new TreeMap();
 262
 263                 cout << "Please enter the number of groups you would like to analyze: ";
 264                 cin >> numGroups;
 265
 266                 int num, count;
 267                 count = 1;
 268                 numEachGroup.resize(numGroups, 0);
 269
 270                 for (int i = 1; i <= numGroups; i++) {
 271                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 272                         cin >> num;
 273
 274                         //set tmaps seqsPerGroup
 275                         tmap->seqsPerGroup[toString(i)] = num;
 276                         tmap->namesOfGroups.push_back(toString(i));
 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                 if (numComp != 1) {
 361                         groupComb.push_back(allGroups);
 362                         numComp++;
 363                 }
 364
 365         }
 366         catch(exception& e) {
 367                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 368                 exit(1);
 369         }
 370         catch(...) {
 371                 cout << "An unknown error has occurred in the ParsimonyCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 372                 exit(1);
 373         }
 374
 375 }
 376 /*****************************************************************/
 377
 378