bootstrapsharedcommand.cpp

   1 /*
   2  *  bootstrapsharedcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/16/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "bootstrapsharedcommand.h"
  11 #include "sharedjabund.h"
  12 #include "sharedsorabund.h"
  13 #include "sharedjclass.h"
  14 #include "sharedsorclass.h"
  15 #include "sharedjest.h"
  16 #include "sharedsorest.h"
  17 #include "sharedthetayc.h"
  18 #include "sharedthetan.h"
  19 #include "sharedmorisitahorn.h"
  20 #include "sharedbraycurtis.h"
  21
  22 //**********************************************************************************************************************
  23 vector<string> BootSharedCommand::getValidParameters(){
  24         try {
  25                 string AlignArray[] =  {"label","calc","groups","iters","outputdir","inputdir"};
  26                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "BootSharedCommand", "getValidParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 BootSharedCommand::BootSharedCommand(){
  36         try {
  37                 abort = true; calledHelp = true;
  38                 vector<string> tempOutNames;
  39                 outputTypes["tree"] = tempOutNames;
  40         }
  41         catch(exception& e) {
  42                 m->errorOut(e, "BootSharedCommand", "BootSharedCommand");
  43                 exit(1);
  44         }
  45 }
  46 //**********************************************************************************************************************
  47 vector<string> BootSharedCommand::getRequiredParameters(){
  48         try {
  49                 vector<string> myArray;
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "BootSharedCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> BootSharedCommand::getRequiredFiles(){
  59         try {
  60                 string AlignArray[] =  {"shared"};
  61                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  62                 return myArray;
  63         }
  64         catch(exception& e) {
  65                 m->errorOut(e, "BootSharedCommand", "getRequiredFiles");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70 BootSharedCommand::BootSharedCommand(string option) {
  71         try {
  72                 globaldata = GlobalData::getInstance();
  73                 abort = false; calledHelp = false;
  74                 allLines = 1;
  75                 labels.clear();
  76                 Groups.clear();
  77                 Estimators.clear();
  78
  79                 //allow user to run help
  80                 if(option == "help") { help(); abort = true; calledHelp = true; }
  81
  82                 else {
  83                         //valid paramters for this command
  84                         string Array[] =  {"label","calc","groups","iters","outputdir","inputdir"};
  85                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  86
  87                         OptionParser parser(option);
  88                         map<string,string> parameters = parser.getParameters();
  89
  90                         ValidParameters validParameter;
  91
  92                         //check to make sure all parameters are valid for command
  93                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  94                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  95                         }
  96
  97                         //initialize outputTypes
  98                         vector<string> tempOutNames;
  99                         outputTypes["tree"] = tempOutNames;
 100
 101                         //if the user changes the output directory command factory will send this info to us in the output parameter
 102                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 103                                 outputDir = "";
 104                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 105                         }
 106
 107
 108                         //make sure the user has already run the read.otu command
 109                         if (globaldata->getSharedFile() == "") {
 110                                 if (globaldata->getListFile() == "") { m->mothurOut("You must read a list and a group, or a shared before you can use the bootstrap.shared command."); m->mothurOutEndLine(); abort = true; }
 111                                 else if (globaldata->getGroupFile() == "") { m->mothurOut("You must read a list and a group, or a shared before you can use the bootstrap.shared command."); m->mothurOutEndLine(); abort = true; }
 112                         }
 113
 114                         //check for optional parameter and set defaults
 115                         // ...at some point should added some additional type checking...
 116                         label = validParameter.validFile(parameters, "label", false);
 117                         if (label == "not found") { label = ""; }
 118                         else {
 119                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 120                                 else { allLines = 1;  }
 121                         }
 122
 123                         //if the user has not specified any labels use the ones from read.otu
 124                         if(label == "") {
 125                                 allLines = globaldata->allLines;
 126                                 labels = globaldata->labels;
 127                         }
 128
 129                         groups = validParameter.validFile(parameters, "groups", false);
 130                         if (groups == "not found") { groups = ""; }
 131                         else {
 132                                 m->splitAtDash(groups, Groups);
 133                                 globaldata->Groups = Groups;
 134                         }
 135
 136                         calc = validParameter.validFile(parameters, "calc", false);
 137                         if (calc == "not found") { calc = "jclass-thetayc";  }
 138                         else {
 139                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 140                         }
 141                         m->splitAtDash(calc, Estimators);
 142
 143                         string temp;
 144                         temp = validParameter.validFile(parameters, "iters", false);  if (temp == "not found") { temp = "1000"; }
 145                         convert(temp, iters);
 146
 147                         if (abort == false) {
 148
 149                                 //used in tree constructor
 150                                 globaldata->runParse = false;
 151
 152                                 validCalculator = new ValidCalculators();
 153
 154                                 int i;
 155                                 for (i=0; i<Estimators.size(); i++) {
 156                                         if (validCalculator->isValidCalculator("boot", Estimators[i]) == true) {
 157                                                 if (Estimators[i] == "jabund") {
 158                                                         treeCalculators.push_back(new JAbund());
 159                                                 }else if (Estimators[i] == "sorabund") {
 160                                                         treeCalculators.push_back(new SorAbund());
 161                                                 }else if (Estimators[i] == "jclass") {
 162                                                         treeCalculators.push_back(new Jclass());
 163                                                 }else if (Estimators[i] == "sorclass") {
 164                                                         treeCalculators.push_back(new SorClass());
 165                                                 }else if (Estimators[i] == "jest") {
 166                                                         treeCalculators.push_back(new Jest());
 167                                                 }else if (Estimators[i] == "sorest") {
 168                                                         treeCalculators.push_back(new SorEst());
 169                                                 }else if (Estimators[i] == "thetayc") {
 170                                                         treeCalculators.push_back(new ThetaYC());
 171                                                 }else if (Estimators[i] == "thetan") {
 172                                                         treeCalculators.push_back(new ThetaN());
 173                                                 }else if (Estimators[i] == "morisitahorn") {
 174                                                         treeCalculators.push_back(new MorHorn());
 175                                                 }else if (Estimators[i] == "braycurtis") {
 176                                                         treeCalculators.push_back(new BrayCurtis());
 177                                                 }
 178                                         }
 179                                 }
 180
 181                                 delete validCalculator;
 182
 183                                 ofstream* tempo;
 184                                 for (int i=0; i < treeCalculators.size(); i++) {
 185                                         tempo = new ofstream;
 186                                         out.push_back(tempo);
 187                                 }
 188
 189                                 //make a vector of tree* for each calculator
 190                                 trees.resize(treeCalculators.size());
 191                         }
 192                 }
 193
 194         }
 195         catch(exception& e) {
 196                 m->errorOut(e, "BootSharedCommand", "BootSharedCommand");
 197                 exit(1);
 198         }
 199 }
 200
 201 //**********************************************************************************************************************
 202
 203 void BootSharedCommand::help(){
 204         try {
 205                 m->mothurOut("The bootstrap.shared command can only be executed after a successful read.otu command.\n");
 206                 m->mothurOut("The bootstrap.shared command parameters are groups, calc, iters and label.\n");
 207                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n");
 208                 m->mothurOut("The group names are separated by dashes. The label parameter allows you to select what distance levels you would like trees created for, and is also separated by dashes.\n");
 209                 m->mothurOut("The bootstrap.shared command should be in the following format: bootstrap.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels, iters=yourIters).\n");
 210                 m->mothurOut("Example bootstrap.shared(groups=A-B-C, calc=jabund-sorabund, iters=100).\n");
 211                 m->mothurOut("The default value for groups is all the groups in your groupfile.\n");
 212                 m->mothurOut("The default value for calc is jclass-thetayc. The default for iters is 1000.\n");
 213         }
 214         catch(exception& e) {
 215                 m->errorOut(e, "BootSharedCommand", "help");
 216                 exit(1);
 217         }
 218 }
 219
 220 //**********************************************************************************************************************
 221
 222 BootSharedCommand::~BootSharedCommand(){
 223         //made new in execute
 224         if (abort == false) {
 225                 delete input; globaldata->ginput = NULL;
 226                 delete read;
 227                 delete util;
 228                 globaldata->gorder = NULL;
 229         }
 230 }
 231
 232 //**********************************************************************************************************************
 233
 234 int BootSharedCommand::execute(){
 235         try {
 236
 237                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 238
 239                 util = new SharedUtil();
 240
 241                 //read first line
 242                 read = new ReadOTUFile(globaldata->inputFileName);
 243                 read->read(&*globaldata);
 244                 input = globaldata->ginput;
 245                 order = input->getSharedOrderVector();
 246                 string lastLabel = order->getLabel();
 247
 248                 //if the users entered no valid calculators don't execute command
 249                 if (treeCalculators.size() == 0) { return 0; }
 250
 251                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 252                 set<string> processedLabels;
 253                 set<string> userLabels = labels;
 254
 255                 //set users groups
 256                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 257                 numGroups = globaldata->Groups.size();
 258
 259                 //clear globaldatas old tree names if any
 260                 globaldata->Treenames.clear();
 261
 262                 //fills globaldatas tree names
 263                 globaldata->Treenames = globaldata->Groups;
 264
 265                 //create treemap class from groupmap for tree class to use
 266                 tmap = new TreeMap();
 267                 tmap->makeSim(globaldata->gGroupmap);
 268                 globaldata->gTreemap = tmap;
 269
 270                 while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 271                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 272
 273                         if(allLines == 1 || labels.count(order->getLabel()) == 1){
 274
 275                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 276                                 int error = process(order);
 277                                 if (error == 1) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 278
 279                                 processedLabels.insert(order->getLabel());
 280                                 userLabels.erase(order->getLabel());
 281                         }
 282
 283                         //you have a label the user want that is smaller than this label and the last label has not already been processed
 284                         if ((m->anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 285                                 string saveLabel = order->getLabel();
 286
 287                                 delete order;
 288                                 order = input->getSharedOrderVector(lastLabel);
 289                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 290                                 int error = process(order);
 291                                 if (error == 1) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 292
 293                                 processedLabels.insert(order->getLabel());
 294                                 userLabels.erase(order->getLabel());
 295
 296                                 //restore real lastlabel to save below
 297                                 order->setLabel(saveLabel);
 298                         }
 299
 300
 301                         lastLabel = order->getLabel();
 302
 303                         //get next line to process
 304                         delete order;
 305                         order = input->getSharedOrderVector();
 306                 }
 307
 308
 309                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 310
 311                 //output error messages about any remaining user labels
 312                 set<string>::iterator it;
 313                 bool needToRun = false;
 314                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 315                         m->mothurOut("Your file does not include the label " + *it);
 316                         if (processedLabels.count(lastLabel) != 1) {
 317                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 318                                 needToRun = true;
 319                         }else {
 320                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 321                         }
 322                 }
 323
 324                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 325
 326                 //run last line if you need to
 327                 if (needToRun == true)  {
 328                                 if (order != NULL) {    delete order;   }
 329                                 order = input->getSharedOrderVector(lastLabel);
 330                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 331                                 int error = process(order);
 332                                 if (error == 1) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 333
 334                                 delete order;
 335
 336                 }
 337
 338                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } globaldata->Groups.clear(); return 0;        }
 339
 340                 //reset groups parameter
 341                 globaldata->Groups.clear();
 342
 343                 //set first tree file as new current treefile
 344                 string currentTree = "";
 345                 itTypes = outputTypes.find("tree");
 346                 if (itTypes != outputTypes.end()) {
 347                         if ((itTypes->second).size() != 0) { currentTree = (itTypes->second)[0]; m->setTreeFile(currentTree); }
 348                 }
 349
 350                 m->mothurOutEndLine();
 351                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 352                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 353                 m->mothurOutEndLine();
 354
 355
 356                 return 0;
 357         }
 358         catch(exception& e) {
 359                 m->errorOut(e, "BootSharedCommand", "execute");
 360                 exit(1);
 361         }
 362 }
 363 //**********************************************************************************************************************
 364
 365 int BootSharedCommand::createTree(ostream* out, Tree* t){
 366         try {
 367
 368                 //do merges and create tree structure by setting parents and children
 369                 //there are numGroups - 1 merges to do
 370                 for (int i = 0; i < (numGroups - 1); i++) {
 371
 372                         if (m->control_pressed) {  return 1; }
 373
 374                         float largest = -1000.0;
 375                         int row, column;
 376                         //find largest value in sims matrix by searching lower triangle
 377                         for (int j = 1; j < simMatrix.size(); j++) {
 378                                 for (int k = 0; k < j; k++) {
 379                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 380                                 }
 381                         }
 382
 383                         //set non-leaf node info and update leaves to know their parents
 384                         //non-leaf
 385                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 386
 387                         //parents
 388                         t->tree[index[row]].setParent(numGroups + i);
 389                         t->tree[index[column]].setParent(numGroups + i);
 390
 391                         //blength = distance / 2;
 392                         float blength = ((1.0 - largest) / 2);
 393
 394                         //branchlengths
 395                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 396                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 397
 398                         //set your length to leaves to your childs length plus branchlength
 399                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 400
 401
 402                         //update index
 403                         index[row] = numGroups+i;
 404                         index[column] = numGroups+i;
 405
 406                         //zero out highest value that caused the merge.
 407                         simMatrix[row][column] = -1000.0;
 408                         simMatrix[column][row] = -1000.0;
 409
 410                         //merge values in simsMatrix
 411                         for (int n = 0; n < simMatrix.size(); n++)      {
 412                                 //row becomes merge of 2 groups
 413                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 414                                 simMatrix[n][row] = simMatrix[row][n];
 415                                 //delete column
 416                                 simMatrix[column][n] = -1000.0;
 417                                 simMatrix[n][column] = -1000.0;
 418                         }
 419                 }
 420
 421                 //adjust tree to make sure root to tip length is .5
 422                 int root = t->findRoot();
 423                 t->tree[root].setBranchLength((0.5 - t->tree[root].getLengthToLeaves()));
 424
 425                 //assemble tree
 426                 t->assembleTree();
 427
 428                 if (m->control_pressed) { return 1; }
 429
 430                 //print newick file
 431                 t->print(*out);
 432
 433                 return 0;
 434
 435         }
 436         catch(exception& e) {
 437                 m->errorOut(e, "BootSharedCommand", "createTree");
 438                 exit(1);
 439         }
 440 }
 441 /***********************************************************/
 442 void BootSharedCommand::printSims() {
 443         try {
 444                 m->mothurOut("simsMatrix"); m->mothurOutEndLine();
 445                 for (int k = 0; k < simMatrix.size(); k++)      {
 446                         for (int n = 0; n < simMatrix.size(); n++)      {
 447                                 m->mothurOut(toString(simMatrix[k][n]));  m->mothurOut("\t");
 448                         }
 449                         m->mothurOutEndLine();
 450                 }
 451
 452         }
 453         catch(exception& e) {
 454                 m->errorOut(e, "BootSharedCommand", "printSims");
 455                 exit(1);
 456         }
 457 }
 458 /***********************************************************/
 459 int BootSharedCommand::process(SharedOrderVector* order) {
 460         try{
 461                                 EstOutput data;
 462                                 vector<SharedRAbundVector*> subset;
 463
 464                                 //open an ostream for each calc to print to
 465                                 for (int z = 0; z < treeCalculators.size(); z++) {
 466                                         //create a new filename
 467                                         outputFile = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 468                                         m->openOutputFile(outputFile, *(out[z]));
 469                                         outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
 470                                 }
 471
 472                                 m->mothurOut("Generating bootstrap trees..."); cout.flush();
 473
 474                                 //create a file for each calculator with the 1000 trees in it.
 475                                 for (int p = 0; p < iters; p++) {
 476
 477                                         if (m->control_pressed) {  return 1; }
 478
 479                                         util->getSharedVectorswithReplacement(globaldata->Groups, lookup, order);  //fills group vectors from order vector.
 480
 481
 482                                         //for each calculator
 483                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 484
 485                                                 if (m->control_pressed) {  return 1; }
 486
 487                                                 //initialize simMatrix
 488                                                 simMatrix.clear();
 489                                                 simMatrix.resize(numGroups);
 490                                                 for (int o = 0; o < simMatrix.size(); o++)      {
 491                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 492                                                                 simMatrix[o].push_back(0.0);
 493                                                         }
 494                                                 }
 495
 496                                                 //initialize index
 497                                                 index.clear();
 498                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 499
 500                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 501                                                         for (int l = k; l < lookup.size(); l++) {
 502                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 503                                                                         subset.clear(); //clear out old pair of sharedrabunds
 504                                                                         //add new pair of sharedrabunds
 505                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 506
 507                                                                         //get estimated similarity between 2 groups
 508                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 509                                                                         //save values in similarity matrix
 510                                                                         simMatrix[k][l] = data[0];
 511                                                                         simMatrix[l][k] = data[0];
 512                                                                 }
 513                                                         }
 514                                                 }
 515
 516                                                 tempTree = new Tree();
 517
 518                                                 if (m->control_pressed) {   delete tempTree; return 1; }
 519
 520                                                 //creates tree from similarity matrix and write out file
 521                                                 createTree(out[i], tempTree);
 522
 523                                                 if (m->control_pressed) {   delete tempTree; return 1; }
 524
 525                                                 //save trees for consensus command.
 526                                                 trees[i].push_back(tempTree);
 527                                         }
 528                                 }
 529
 530                                 m->mothurOut("\tDone."); m->mothurOutEndLine();
 531                                 //delete globaldata's tree
 532                                 //for (int m = 0; m < globaldata->gTree.size(); m++) {  delete globaldata->gTree[m];  }
 533                                 //globaldata->gTree.clear();
 534
 535
 536                                 //create consensus trees for each bootstrapped tree set
 537                                 for (int k = 0; k < trees.size(); k++) {
 538
 539                                         m->mothurOut("Generating consensus tree for " + treeCalculators[k]->getName()); m->mothurOutEndLine();
 540
 541                                         if (m->control_pressed) {  return 1; }
 542
 543                                         //set global data to calc trees
 544                                         globaldata->gTree = trees[k];
 545
 546                                         string filename = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + treeCalculators[k]->getName() + ".boot" + order->getLabel();
 547                                         consensus = new ConcensusCommand(filename);
 548                                         consensus->execute();
 549                                         delete consensus;
 550
 551                                         outputNames.push_back(filename + ".cons.pairs");
 552                                         outputNames.push_back(filename + ".cons.tre");
 553
 554                                         //delete globaldata's tree
 555                                         //for (int m = 0; m < globaldata->gTree.size(); m++) {  delete globaldata->gTree[m];  }
 556                                         //globaldata->gTree.clear();
 557
 558                                 }
 559
 560
 561
 562                                 //close ostream for each calc
 563                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 564
 565                                 return 0;
 566
 567         }
 568         catch(exception& e) {
 569                 m->errorOut(e, "BootSharedCommand", "process");
 570                 exit(1);
 571         }
 572 }
 573 /***********************************************************/
 574
 575
 576