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(string option) {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 Groups.clear();
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"random","groups","iters"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string, string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         randomtree = validParameter.validFile(parameters, "random", false);             if (randomtree == "not found") { randomtree = ""; }
  38
  39                         //are you trying to use parsimony without reading a tree or saying you want random distribution
  40                         if (randomtree == "")  {
  41                                 if (globaldata->gTree.size() == 0) {
  42                                         cout << "You must read a treefile and a groupfile or set the randomtree parameter to the output filename you wish, before you may execute the parsimony command." << endl; abort = true;  }
  43                         }
  44
  45                         //check for optional parameter and set defaults
  46                         // ...at some point should added some additional type checking...
  47                         groups = validParameter.validFile(parameters, "groups", false);
  48                         if (groups == "not found") { groups = ""; }
  49                         else {
  50                                 splitAtDash(groups, Groups);
  51                                 globaldata->Groups = Groups;
  52                         }
  53
  54                         itersString = validParameter.validFile(parameters, "iters", false);                     if (itersString == "not found") { itersString = "1000"; }
  55                         convert(itersString, iters);
  56
  57                         if (abort == false) {
  58                                 //randomtree will tell us if user had their own treefile or if they just want the random distribution
  59                                 //user has entered their own tree
  60                                 if (randomtree == "") {
  61                                         T = globaldata->gTree;
  62                                         tmap = globaldata->gTreemap;
  63                                         output = new ColumnFile(globaldata->getTreeFile()  +  ".parsimony", itersString);
  64                                         sumFile = globaldata->getTreeFile() + ".psummary";
  65                                         openOutputFile(sumFile, outSum);
  66                                 }else { //user wants random distribution
  67                                         savetmap = globaldata->gTreemap;
  68                                         getUserInput();
  69                                         output = new ColumnFile(randomtree, itersString);
  70                                 }
  71
  72                                 //set users groups to analyze
  73                                 util = new SharedUtil();
  74                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  75                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  76
  77                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
  78
  79                                 pars = new Parsimony(tmap);
  80                                 counter = 0;
  81
  82                         }
  83
  84                 }
  85
  86         }
  87         catch(exception& e) {
  88                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  89                 exit(1);
  90         }
  91         catch(...) {
  92                 cout << "An unknown error has occurred in the ParsimonyCommand class function ParsimonyCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  93                 exit(1);
  94         }
  95 }
  96
  97 //**********************************************************************************************************************
  98
  99 void ParsimonyCommand::help(){
 100         try {
 101                 cout << "The parsimony command can only be executed after a successful read.tree command, unless you use the random parameter." << "\n";
 102                 cout << "The parsimony command parameters are random, groups and iters.  No parameters are required." << "\n";
 103                 cout << "The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 1 valid group." << "\n";
 104                 cout << "The group names are separated by dashes.  The iters parameter allows you to specify how many random trees you would like compared to your tree." << "\n";
 105                 cout << "The parsimony command should be in the following format: parsimony(random=yourOutputFilename, groups=yourGroups, iters=yourIters)." << "\n";
 106                 cout << "Example parsimony(random=out, iters=500)." << "\n";
 107                 cout << "The default value for random is "" (meaning you want to use the trees in your inputfile, randomtree=out means you just want the random distribution of trees outputted to out.rd_parsimony)," << "\n";
 108                 cout << "and iters is 1000.  The parsimony command output two files: .parsimony and .psummary their descriptions are in the manual." << "\n";
 109                 cout << "Note: No spaces between parameter labels (i.e. random), '=' and parameters (i.e.yourOutputFilename)." << "\n" << "\n";
 110         }
 111         catch(exception& e) {
 112                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 113                 exit(1);
 114         }
 115         catch(...) {
 116                 cout << "An unknown error has occurred in the ParsimonyCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 117                 exit(1);
 118         }
 119 }
 120
 121
 122 /***********************************************************/
 123 int ParsimonyCommand::execute() {
 124         try {
 125
 126                 if (abort == true) { return 0; }
 127
 128                 Progress* reading;
 129                 reading = new Progress("Comparing to random:", iters);
 130
 131                 //get pscore for users tree
 132                 userData.resize(numComp,0);  //data = AB, AC, BC, ABC.
 133                 randomData.resize(numComp,0);  //data = AB, AC, BC, ABC.
 134                 rscoreFreq.resize(numComp);
 135                 uscoreFreq.resize(numComp);
 136                 rCumul.resize(numComp);
 137                 uCumul.resize(numComp);
 138                 userTreeScores.resize(numComp);
 139                 UScoreSig.resize(numComp);
 140
 141                 if (randomtree == "") {
 142                         //get pscores for users trees
 143                         for (int i = 0; i < T.size(); i++) {
 144                                 userData = pars->getValues(T[i]);  //data = AB, AC, BC, ABC.
 145
 146                                 //output scores for each combination
 147                                 for(int k = 0; k < numComp; k++) {
 148
 149                                         //update uscoreFreq
 150                                         map<int,double>::iterator it = uscoreFreq[k].find(userData[k]);
 151                                         if (it == uscoreFreq[k].end()) {//new score
 152                                                 uscoreFreq[k][userData[k]] = 1;
 153                                         }else{ uscoreFreq[k][userData[k]]++; }
 154
 155                                         //add users score to valid scores
 156                                         validScores[userData[k]] = userData[k];
 157
 158                                         //save score for summary file
 159                                         userTreeScores[k].push_back(userData[k]);
 160                                 }
 161                         }
 162
 163                         //get pscores for random trees
 164                         for (int j = 0; j < iters; j++) {
 165                                 //create new tree with same num nodes and leaves as users
 166                                 randT = new Tree();
 167
 168                                 //create random relationships between nodes
 169                                 randT->assembleRandomTree();
 170
 171                                 //get pscore of random tree
 172                                 randomData = pars->getValues(randT);
 173
 174                                 for(int r = 0; r < numComp; r++) {
 175                                         //add trees pscore to map of scores
 176                                         map<int,double>::iterator it = rscoreFreq[r].find(randomData[r]);
 177                                         if (it != rscoreFreq[r].end()) {//already have that score
 178                                                 rscoreFreq[r][randomData[r]]++;
 179                                         }else{//first time we have seen this score
 180                                                 rscoreFreq[r][randomData[r]] = 1;
 181                                         }
 182
 183                                         //add randoms score to validscores
 184                                         validScores[randomData[r]] = randomData[r];
 185                                 }
 186
 187                                 //update progress bar
 188                                 reading->update(j);
 189
 190                                 delete randT;
 191                         }
 192
 193                 }else {
 194                         //get pscores for random trees
 195                         for (int j = 0; j < iters; j++) {
 196                                 //create new tree with same num nodes and leaves as users
 197                                 randT = new Tree();
 198                                 //create random relationships between nodes
 199
 200                                 randT->assembleRandomTree();
 201
 202                                 //get pscore of random tree
 203                                 randomData = pars->getValues(randT);
 204
 205                                 for(int r = 0; r < numComp; r++) {
 206                                         //add trees pscore to map of scores
 207                                         map<int,double>::iterator it = rscoreFreq[r].find(randomData[r]);
 208                                         if (it != rscoreFreq[r].end()) {//already have that score
 209                                                 rscoreFreq[r][randomData[r]]++;
 210                                         }else{//first time we have seen this score
 211                                                 rscoreFreq[r][randomData[r]] = 1;
 212                                         }
 213
 214                                         //add randoms score to validscores
 215                                         validScores[randomData[r]] = randomData[r];
 216                                 }
 217
 218                                 //update progress bar
 219                                 reading->update(j);
 220
 221                                 delete randT;
 222                         }
 223                 }
 224
 225                 for(int a = 0; a < numComp; a++) {
 226                         float rcumul = 0.0000;
 227                         float ucumul = 0.0000;
 228                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 229                         for (map<int,double>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 230                                 if (randomtree == "") {
 231                                         map<int,double>::iterator it2 = uscoreFreq[a].find(it->first);
 232                                         //user data has that score
 233                                         if (it2 != uscoreFreq[a].end()) { uscoreFreq[a][it->first] /= T.size(); ucumul+= it2->second;  }
 234                                         else { uscoreFreq[a][it->first] = 0.0000; } //no user trees with that score
 235                                         //make uCumul map
 236                                         uCumul[a][it->first] = ucumul;
 237                                 }
 238
 239                                 //make rscoreFreq map and rCumul
 240                                 map<int,double>::iterator it2 = rscoreFreq[a].find(it->first);
 241                                 //get percentage of random trees with that info
 242                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul+= it2->second;  }
 243                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 244                                 rCumul[a][it->first] = rcumul;
 245                         }
 246
 247                         //find the signifigance of each user trees score when compared to the random trees and save for printing the summary file
 248                         for (int h = 0; h < userTreeScores[a].size(); h++) {
 249                                 UScoreSig[a].push_back(rCumul[a][userTreeScores[a][h]]);
 250                         }
 251                 }
 252
 253                 //finish progress bar
 254                 reading->finish();
 255                 delete reading;
 256
 257
 258                 printParsimonyFile();
 259                 if (randomtree == "") { printUSummaryFile(); }
 260
 261                 //reset globaldata's treemap if you just did random distrib
 262                 if (randomtree != "") {
 263                         //memory leak prevention
 264                         //if (globaldata->gTreemap != NULL) { delete globaldata->gTreemap;  }
 265                         globaldata->gTreemap = savetmap;
 266                 }
 267
 268                 //reset groups parameter
 269                 globaldata->Groups.clear();
 270
 271                 return 0;
 272
 273         }
 274         catch(exception& e) {
 275                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 276                 exit(1);
 277         }
 278         catch(...) {
 279                 cout << "An unknown error has occurred in the ParsimonyCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 280                 exit(1);
 281         }
 282 }
 283
 284 /***********************************************************/
 285 void ParsimonyCommand::printParsimonyFile() {
 286         try {
 287                 vector<double> data;
 288                 vector<string> tags;
 289
 290                 if (randomtree == "") {
 291                         tags.push_back("Score"); tags.push_back("UserFreq"); tags.push_back("UserCumul"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 292                 }else {
 293                         tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 294                 }
 295
 296                 for(int a = 0; a < numComp; a++) {
 297                         output->initFile(groupComb[a], tags);
 298                         //print each line
 299                         for (map<int,double>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 300                                 if (randomtree == "") {
 301                                         data.push_back(it->first);  data.push_back(uscoreFreq[a][it->first]); data.push_back(uCumul[a][it->first]); data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 302                                 }else{
 303                                         data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 304                                 }
 305                                 output->output(data);
 306                                 data.clear();
 307                         }
 308                         output->resetFile();
 309                 }
 310         }
 311         catch(exception& e) {
 312                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 313                 exit(1);
 314         }
 315         catch(...) {
 316                 cout << "An unknown error has occurred in the ParsimonyCommand class function printParsimonyFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 317                 exit(1);
 318         }
 319 }
 320 /***********************************************************/
 321 void ParsimonyCommand::printUSummaryFile() {
 322         try {
 323                 //column headers
 324                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "ParsScore" << '\t' << "ParsSig" <<  endl;
 325                 cout << "Tree#" << '\t' << "Groups" << '\t'  <<  "ParsScore" << '\t' << "ParsSig" <<  endl;
 326
 327                 //format output
 328                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 329
 330
 331                 //print each line
 332                 for (int i = 0; i< T.size(); i++) {
 333                         for(int a = 0; a < numComp; a++) {
 334                                 if (UScoreSig[a][i] > (1/(float)iters)) {
 335                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[a]  << '\t' << userTreeScores[a][i] << setprecision(itersString.length()) << '\t' << UScoreSig[a][i] << endl;
 336                                         cout << setprecision(6) << i+1 << '\t' << groupComb[a]  << '\t' << userTreeScores[a][i] << setprecision(itersString.length()) << '\t' << UScoreSig[a][i] << endl;
 337                                 }else {
 338                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << userTreeScores[a][i] << setprecision(itersString.length())  << '\t' << "<" << (1/float(iters)) << endl;
 339                                         cout << setprecision(6) << i+1 << '\t' << groupComb[a] << '\t' << userTreeScores[a][i] << setprecision(itersString.length()) << '\t' << "<" << (1/float(iters)) << endl;
 340                                 }
 341                         }
 342                 }
 343
 344                 outSum.close();
 345         }
 346         catch(exception& e) {
 347                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 348                 exit(1);
 349         }
 350         catch(...) {
 351                 cout << "An unknown error has occurred in the ParsimonyCommand class function printUSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 352                 exit(1);
 353         }
 354 }
 355
 356 /***********************************************************/
 357 void ParsimonyCommand::getUserInput() {
 358         try {
 359
 360                 //create treemap
 361                 tmap = new TreeMap();
 362
 363                 cout << "Please enter the number of groups you would like to analyze: ";
 364                 cin >> numGroups;
 365
 366                 int num, count;
 367                 count = 1;
 368                 numEachGroup.resize(numGroups, 0);
 369
 370                 for (int i = 1; i <= numGroups; i++) {
 371                         cout << "Please enter the number of sequences in group " << i <<  ": ";
 372                         cin >> num;
 373
 374                         //set tmaps seqsPerGroup
 375                         tmap->seqsPerGroup[toString(i)] = num;
 376                         tmap->namesOfGroups.push_back(toString(i));
 377
 378                         //set tmaps namesOfSeqs
 379                         for (int j = 0; j < num; j++) {
 380                                 tmap->namesOfSeqs.push_back(toString(count));
 381                                 tmap->treemap[toString(count)].groupname = toString(i);
 382                                 count++;
 383                         }
 384                 }
 385
 386                 //clears buffer so next command doesn't have error
 387                 string s;
 388                 getline(cin, s);
 389
 390                 //save tmap for later
 391                 //memory leak prevention
 392                 //if (globaldata->gTreemap != NULL) { delete globaldata->gTreemap;  }
 393                 globaldata->gTreemap = tmap;
 394                 globaldata->Treenames = tmap->namesOfSeqs;
 395
 396         }
 397         catch(exception& e) {
 398                 cout << "Standard Error: " << e.what() << " has occurred in the ParsimonyCommand class Function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 399                 exit(1);
 400         }
 401         catch(...) {
 402                 cout << "An unknown error has occurred in the ParsimonyCommand class function getUserInput. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 403                 exit(1);
 404         }
 405 }
 406
 407 /***********************************************************/
 408
 409