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