consensuscommand.cpp

   1 /*
   2  *  consensuscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/29/09.
   6  *  Copyright 2009 Schloss Lab UMASS AMherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "consensuscommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> ConcensusCommand::getValidParameters(){
  14         try {
  15                 vector<string> myArray;
  16                 return myArray;
  17         }
  18         catch(exception& e) {
  19                 m->errorOut(e, "ConcensusCommand", "getValidParameters");
  20                 exit(1);
  21         }
  22 }
  23 //**********************************************************************************************************************
  24 vector<string> ConcensusCommand::getRequiredParameters(){
  25         try {
  26                 vector<string> myArray;
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ConcensusCommand", "getRequiredParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 vector<string> ConcensusCommand::getRequiredFiles(){
  36         try {
  37                 string AlignArray[] =  {"tree","group"};
  38                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  39                 return myArray;
  40         }
  41         catch(exception& e) {
  42                 m->errorOut(e, "ConcensusCommand", "getRequiredFiles");
  43                 exit(1);
  44         }
  45 }
  46 //**********************************************************************************************************************
  47 ConcensusCommand::ConcensusCommand(){
  48         try {
  49                 abort = true; calledHelp = true;
  50                 vector<string> tempOutNames;
  51                 outputTypes["tree"] = tempOutNames;
  52                 outputTypes["nodepairs"] = tempOutNames;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "ConcensusCommand", "ConcensusCommand");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60
  61 ConcensusCommand::ConcensusCommand(string fileroot)  {
  62         try {
  63                 globaldata = GlobalData::getInstance();
  64                 abort = false; calledHelp = false;
  65
  66                 //initialize outputTypes
  67                 vector<string> tempOutNames;
  68                 outputTypes["tree"] = tempOutNames;
  69                 outputTypes["nodepairs"] = tempOutNames;
  70
  71                 filename = fileroot;
  72
  73                 t = globaldata->gTree;
  74
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "ConcensusCommand", "ConcensusCommand");
  78                 exit(1);
  79         }
  80 }
  81
  82 //**********************************************************************************************************************
  83
  84 void ConcensusCommand::help(){
  85         try {
  86                 m->mothurOut("The consensus command can only be executed after a successful read.tree command.\n");
  87                 m->mothurOut("The consensus command has no parameters.\n");
  88                 m->mothurOut("The consensus command should be in the following format: consensus().\n");
  89                 m->mothurOut("The consensus command output two files: .consensus.tre and .consensuspairs.\n");
  90                 m->mothurOut("The .consensus.tre file contains the consensus tree of the trees in your input file.\n");
  91                 m->mothurOut("The branch lengths are the percentage of trees in your input file that had the given pair.\n");
  92                 m->mothurOut("The .consensuspairs file contains a list of the internal nodes in your tree.  For each node, the pair that was used in the consensus tree \n");
  93                 m->mothurOut("is reported with its percentage, as well as the other pairs that were seen for that node but not used and their percentages.\n\n");
  94         }
  95         catch(exception& e) {
  96                 m->errorOut(e, "ConcensusCommand", "help");
  97                 exit(1);
  98         }
  99 }
 100
 101 //**********************************************************************************************************************
 102
 103 ConcensusCommand::~ConcensusCommand(){}
 104
 105 //**********************************************************************************************************************
 106
 107 int ConcensusCommand::execute(){
 108         try {
 109
 110                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 111                 else {
 112                         numNodes = t[0]->getNumNodes();
 113                         numLeaves = t[0]->getNumLeaves();
 114                 }
 115
 116                 //get the possible pairings
 117                 getSets();
 118
 119                 if (m->control_pressed) { return 0; }
 120
 121                 //open file for pairing not included in the tree
 122                 notIncluded = filename + ".cons.pairs"; outputNames.push_back(notIncluded);  outputTypes["nodepairs"].push_back(notIncluded);
 123                 m->openOutputFile(notIncluded, out2);
 124
 125                 consensusTree = new Tree();
 126
 127                 it2 = nodePairs.find(treeSet);
 128
 129                 nodePairsInTree[treeSet] = it2->second;
 130
 131                 //erase treeset because you are adding it
 132                 nodePairs.erase(treeSet);
 133
 134                 //set count to numLeaves;
 135                 count = numLeaves;
 136
 137                 buildConcensusTree(treeSet);
 138
 139                 if (m->control_pressed) { delete consensusTree; return 0; }
 140
 141                 consensusTree->assembleTree();
 142
 143                 if (m->control_pressed) { delete consensusTree; return 0; }
 144
 145                 //output species in order
 146                 out2 << "Species in Order: " << endl << endl;
 147                 for (int i = 0; i < treeSet.size(); i++) {  out2 << i+1 << ".  " << treeSet[i] << endl; }
 148
 149                 //output sets included
 150                 out2 << endl << "Sets included in the consensus tree:" << endl << endl;
 151
 152                 if (m->control_pressed) { delete consensusTree; return 0; }
 153
 154                 vector<string> temp;
 155                 for (it2 = nodePairsInTree.begin(); it2 != nodePairsInTree.end(); it2++) {
 156
 157                         if (m->control_pressed) { delete consensusTree; return 0; }
 158
 159                         //only output pairs not leaves
 160                         if (it2->first.size() > 1) {
 161                                 temp.clear();
 162                                 //initialize temp to all "."
 163                                 temp.resize(treeSet.size(), ".");
 164
 165                                 //set the spot in temp that represents it2->first[i] to a "*"
 166                                 for (int i = 0; i < it2->first.size(); i++) {
 167                                         //find spot
 168                                         int index = findSpot(it2->first[i]);
 169                                         temp[index] = "*";
 170                                         //temp[index] = it2->first[i] + "  ";
 171                                 }
 172
 173                                 //output temp
 174                                 for (int j = 0; j < temp.size(); j++) {
 175                                         out2 << temp[j];
 176                                 }
 177                                 out2 << '\t' << it2->second << endl;
 178                         }
 179                 }
 180
 181                 //output sets not included
 182                 out2 << endl << "Sets NOT included in the consensus tree:" << endl << endl;
 183                 for (it2 = nodePairs.begin(); it2 != nodePairs.end(); it2++) {
 184
 185                         if (m->control_pressed) { delete consensusTree; return 0; }
 186
 187                         temp.clear();
 188                         //initialize temp to all "."
 189                         temp.resize(treeSet.size(), ".");
 190
 191                         //set the spot in temp that represents it2->first[i] to a "*"
 192                         for (int i = 0; i < it2->first.size(); i++) {
 193                                 //find spot
 194                                 int index = findSpot(it2->first[i]);
 195                                 temp[index] = "*";
 196                         }
 197
 198                         //output temp
 199                         for (int j = 0; j < temp.size(); j++) {
 200                                 out2 << temp[j];
 201                         }
 202                         out2 << '\t' << it2->second << endl;
 203                 }
 204
 205                 outputFile = filename + ".cons.tre";  outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 206                 m->openOutputFile(outputFile, out);
 207
 208                 consensusTree->print(out, "boot");
 209
 210                 out.close(); out2.close();
 211
 212                 delete consensusTree;
 213
 214                 return 0;
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "ConcensusCommand", "execute");
 218                 exit(1);
 219         }
 220 }
 221
 222 //**********************************************************************************************************************
 223 int ConcensusCommand::buildConcensusTree(vector<string> nodeSet) {
 224         try {
 225                 vector<string> leftChildSet;
 226                 vector<string> rightChildSet;
 227
 228                 if (m->control_pressed) { return 1; }
 229
 230                 //if you are at a leaf
 231                 if (nodeSet.size() == 1) {
 232                         //return the vector index of the leaf you are at
 233                         return consensusTree->getIndex(nodeSet[0]);
 234                 //terminate recursion
 235                 }else if (count == numNodes) { return 0; }
 236                 else {
 237                         //finds best child pair
 238                         leftChildSet = getNextAvailableSet(nodeSet, rightChildSet);
 239                         int left = buildConcensusTree(leftChildSet);
 240                         int right = buildConcensusTree(rightChildSet);
 241                         consensusTree->tree[count].setChildren(left, right);
 242                         consensusTree->tree[count].setLabel(nodePairsInTree[nodeSet]);
 243                         consensusTree->tree[left].setParent(count);
 244                         consensusTree->tree[right].setParent(count);
 245                         count++;
 246                         return (count-1);
 247                 }
 248
 249         }
 250         catch(exception& e) {
 251                 m->errorOut(e, "ConcensusCommand", "buildConcensusTree");
 252                 exit(1);
 253         }
 254 }
 255
 256 //**********************************************************************************************************************
 257 int ConcensusCommand::getSets() {
 258         try {
 259                 vector<string> temp;
 260                 treeSet.clear();
 261
 262                 //for each tree add the possible pairs you find
 263                 for (int i = 0; i < t.size(); i++) {
 264
 265                         //for each non-leaf node get descendant info.
 266                         for (int j = numLeaves; j < numNodes; j++) {
 267
 268                                 if (m->control_pressed) { return 1; }
 269
 270                                 temp.clear();
 271                                 //go through pcounts and pull out descendants
 272                                 for (it = t[i]->tree[j].pcount.begin(); it != t[i]->tree[j].pcount.end(); it++) {
 273                                         temp.push_back(it->first);
 274                                 }
 275
 276                                 //sort temp
 277                                 sort(temp.begin(), temp.end());
 278
 279                                 it2 = nodePairs.find(temp);
 280                                 if (it2 != nodePairs.end()) {
 281                                         nodePairs[temp]++;
 282                                 }else{
 283                                         nodePairs[temp] = 1;
 284                                 }
 285                         }
 286                 }
 287
 288
 289                 //add each leaf to terminate recursion in consensus
 290                 //you want the leaves in there but with insignifigant sightings value so it is added last
 291                 //for each leaf node get descendant info.
 292                 for (int j = 0; j < numLeaves; j++) {
 293
 294                         if (m->control_pressed) { return 1; }
 295
 296                         //only need the first one since leaves have no descendants but themselves
 297                         it = t[0]->tree[j].pcount.begin();
 298                         temp.clear();  temp.push_back(it->first);
 299
 300                         //fill treeSet
 301                         treeSet.push_back(it->first);
 302
 303                         //add leaf to list but with sighting value less then all non leaf pairs
 304                         nodePairs[temp] = 0;
 305                 }
 306
 307                 sort(treeSet.begin(), treeSet.end());
 308
 309
 310                 map< vector<string>, int> nodePairsCopy = nodePairs;
 311
 312
 313                 //set initial rating on pairs to sightings + subgroup sightings
 314                 while (nodePairsCopy.size() != 0) {
 315                         if (m->control_pressed) { return 1; }
 316
 317                         vector<string> smallOne = getSmallest(nodePairsCopy);
 318
 319                         int subgrouprate = getSubgroupRating(smallOne);
 320
 321                         nodePairsInitialRate[smallOne] = nodePairs[smallOne] + subgrouprate;
 322
 323                         nodePairsCopy.erase(smallOne);
 324                 }
 325
 326                 return 0;
 327         }
 328         catch(exception& e) {
 329                 m->errorOut(e, "ConcensusCommand", "getSets");
 330                 exit(1);
 331         }
 332 }
 333 //**********************************************************************************************************************
 334 vector<string> ConcensusCommand::getSmallest(map< vector<string>, int> nodes) {
 335         try{
 336                 vector<string> smallest = nodes.begin()->first;
 337                 int smallsize = smallest.size();
 338
 339                 for(it2 = nodes.begin(); it2 != nodes.end(); it2++) {
 340                         if(it2->first.size() < smallsize) { smallsize = it2->first.size();  smallest = it2->first;  }
 341                 }
 342
 343                 return smallest;
 344         }
 345         catch(exception& e) {
 346                 m->errorOut(e, "ConcensusCommand", "getSmallest");
 347                 exit(1);
 348         }
 349 }
 350
 351 //**********************************************************************************************************************
 352 vector<string> ConcensusCommand::getNextAvailableSet(vector<string> bigset, vector<string>& rest) {
 353         try {
 354 //cout << "new call " << endl << endl << endl;
 355                 vector<string> largest; largest.clear();
 356                 rest.clear();
 357
 358                 //if you are just 2 groups
 359                 if (bigset.size() == 2)  {
 360                         rest.push_back(bigset[0]);
 361                         largest.push_back(bigset[1]);
 362                 }else{
 363                         rest = bestSplit[bigset][0];
 364                         largest = bestSplit[bigset][1];
 365                 }
 366
 367
 368                 //save for printing out later and for branch lengths
 369                 nodePairsInTree[rest] = nodePairs[rest];
 370
 371                 //delete whatever set you return because it is no longer available
 372                 nodePairs.erase(rest);
 373
 374                 //save for printing out later and for branch lengths
 375                 nodePairsInTree[largest] = nodePairs[largest];
 376
 377                 //delete whatever set you return because it is no longer available
 378                 nodePairs.erase(largest);
 379
 380                 return largest;
 381
 382         }
 383         catch(exception& e) {
 384                 m->errorOut(e, "ConcensusCommand", "getNextAvailableSet");
 385                 exit(1);
 386         }
 387 }
 388
 389 /**********************************************************************************************************************/
 390 int ConcensusCommand::getSubgroupRating(vector<string> group) {
 391         try {
 392                 map< vector<string>, int>::iterator ittemp;
 393                 map< vector< vector<string> > , int >::iterator it3;
 394                 int rate = 0;
 395
 396                 // ***********************************************************************************//
 397                 //1. this function must be called passing it littlest sets to biggest
 398                 //              since it the rating is made from your sighting plus you best splits rating
 399                 //2. it saves the top pair to use later
 400                 // ***********************************************************************************//
 401
 402
 403                 if (group.size() < 3) {  return rate;  }
 404
 405                 map< vector<string>, int> possiblePairing;  //this is all the subsets of group
 406
 407                 //go through the sets
 408                 for (it2 = nodePairs.begin(); it2 != nodePairs.end(); it2++) {
 409                         //are you a subset of bigset, then save in possiblePairings
 410                         if (isSubset(group, it2->first) == true) {  possiblePairing[it2->first] = it2->second;  }
 411                 }
 412
 413                 map< vector< vector<string> > , int > rating;
 414
 415                 while (possiblePairing.size() != 0) {
 416
 417                         it2 = possiblePairing.begin();
 418                         vector<string> temprest = getRestSet(group, it2->first);
 419
 420                         //is the rest a set available in possiblePairings
 421                         ittemp = possiblePairing.find(temprest);
 422                         if (ittemp != possiblePairing.end()) {  //if the rest is in the possible pairings then add this pair to rating map
 423                                 vector< vector<string> > temprate;
 424                                 temprate.push_back(it2->first);  temprate.push_back(temprest);
 425
 426                                 rating[temprate] = (nodePairsInitialRate[it2->first] + nodePairsInitialRate[temprest]);
 427
 428                                 //erase so you dont add 1,2 and 2,1.
 429                                 possiblePairing.erase(temprest);
 430                         }
 431
 432                         possiblePairing.erase(it2);
 433                 }
 434
 435
 436                 it3 = rating.begin();
 437                 rate = it3->second;
 438                 vector< vector<string> > topPair = it3->first;
 439
 440                 //choose the split with the best rating
 441                 for (it3 = rating.begin(); it3 != rating.end(); it3++) {
 442
 443                         if (it3->second > rate) {  rate = it3->second;  topPair = it3->first;  }
 444                 }
 445
 446                 bestSplit[group] = topPair;
 447
 448                 return rate;
 449         }
 450         catch(exception& e) {
 451                 m->errorOut(e, "ConcensusCommand", "getSubgroupRating");
 452                 exit(1);
 453         }
 454 }
 455
 456 //**********************************************************************************************************************
 457 vector<string> ConcensusCommand::getRestSet(vector<string> bigset, vector<string> subset) {
 458         try {
 459                 vector<string> rest;
 460
 461                 for (int i = 0; i < bigset.size(); i++) {
 462                         bool inSubset = false;
 463                         for (int j = 0; j < subset.size(); j++) {
 464                                 if (bigset[i] == subset[j]) { inSubset = true; break; }
 465                         }
 466
 467                         //its not in the subset so put it in the rest
 468                         if (inSubset == false) { rest.push_back(bigset[i]); }
 469                 }
 470
 471                 return rest;
 472
 473         }
 474         catch(exception& e) {
 475                 m->errorOut(e, "ConcensusCommand", "getRestSet");
 476                 exit(1);
 477         }
 478 }
 479
 480 //**********************************************************************************************************************
 481 bool ConcensusCommand::isSubset(vector<string> bigset, vector<string> subset) {
 482         try {
 483
 484
 485                 if (subset.size() > bigset.size()) { return false;  }
 486
 487                 //check if each guy in suset is also in bigset
 488                 for (int i = 0; i < subset.size(); i++) {
 489                         bool match = false;
 490                         for (int j = 0; j < bigset.size(); j++) {
 491                                 if (subset[i] == bigset[j]) { match = true; break; }
 492                         }
 493
 494                         //you have a guy in subset that had no match in bigset
 495                         if (match == false) { return false; }
 496                 }
 497
 498                 return true;
 499
 500         }
 501         catch(exception& e) {
 502                 m->errorOut(e, "ConcensusCommand", "isSubset");
 503                 exit(1);
 504         }
 505 }
 506 //**********************************************************************************************************************
 507 int ConcensusCommand::findSpot(string node) {
 508         try {
 509                 int spot;
 510
 511                 //check if each guy in suset is also in bigset
 512                 for (int i = 0; i < treeSet.size(); i++) {
 513                         if (treeSet[i] == node) { spot = i; break; }
 514                 }
 515
 516                 return spot;
 517
 518         }
 519         catch(exception& e) {
 520                 m->errorOut(e, "ConcensusCommand", "findSpot");
 521                 exit(1);
 522         }
 523 }
 524 //**********************************************************************************************************************
 525
 526
 527
 528