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