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                                 process(order);
 227
 228                                 processedLabels.insert(order->getLabel());
 229                                 userLabels.erase(order->getLabel());
 230                         }
 231
 232                         //you have a label the user want that is smaller than this label and the last label has not already been processed
 233                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 234                                 string saveLabel = order->getLabel();
 235
 236                                 delete order;
 237                                 order = input->getSharedOrderVector(lastLabel);
 238                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 239                                 process(order);
 240
 241                                 processedLabels.insert(order->getLabel());
 242                                 userLabels.erase(order->getLabel());
 243
 244                                 //restore real lastlabel to save below
 245                                 order->setLabel(saveLabel);
 246                         }
 247
 248
 249                         lastLabel = order->getLabel();
 250
 251                         //get next line to process
 252                         delete order;
 253                         order = input->getSharedOrderVector();
 254                 }
 255
 256                 //output error messages about any remaining user labels
 257                 set<string>::iterator it;
 258                 bool needToRun = false;
 259                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 260                         m->mothurOut("Your file does not include the label " + *it);
 261                         if (processedLabels.count(lastLabel) != 1) {
 262                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 263                                 needToRun = true;
 264                         }else {
 265                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 266                         }
 267                 }
 268
 269                 //run last line if you need to
 270                 if (needToRun == true)  {
 271                                 if (order != NULL) {    delete order;   }
 272                                 order = input->getSharedOrderVector(lastLabel);
 273                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 274                                 process(order);
 275                                 delete order;
 276
 277                 }
 278
 279                 //reset groups parameter
 280                 globaldata->Groups.clear();
 281
 282                 m->mothurOutEndLine();
 283                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 284                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 285                 m->mothurOutEndLine();
 286
 287
 288                 return 0;
 289         }
 290         catch(exception& e) {
 291                 m->errorOut(e, "BootSharedCommand", "execute");
 292                 exit(1);
 293         }
 294 }
 295 //**********************************************************************************************************************
 296
 297 void BootSharedCommand::createTree(ostream* out, Tree* t){
 298         try {
 299
 300                 //do merges and create tree structure by setting parents and children
 301                 //there are numGroups - 1 merges to do
 302                 for (int i = 0; i < (numGroups - 1); i++) {
 303
 304                         float largest = -1000.0;
 305                         int row, column;
 306                         //find largest value in sims matrix by searching lower triangle
 307                         for (int j = 1; j < simMatrix.size(); j++) {
 308                                 for (int k = 0; k < j; k++) {
 309                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 310                                 }
 311                         }
 312
 313                         //set non-leaf node info and update leaves to know their parents
 314                         //non-leaf
 315                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 316
 317                         //parents
 318                         t->tree[index[row]].setParent(numGroups + i);
 319                         t->tree[index[column]].setParent(numGroups + i);
 320
 321                         //blength = distance / 2;
 322                         float blength = ((1.0 - largest) / 2);
 323
 324                         //branchlengths
 325                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 326                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 327
 328                         //set your length to leaves to your childs length plus branchlength
 329                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 330
 331
 332                         //update index
 333                         index[row] = numGroups+i;
 334                         index[column] = numGroups+i;
 335
 336                         //zero out highest value that caused the merge.
 337                         simMatrix[row][column] = -1000.0;
 338                         simMatrix[column][row] = -1000.0;
 339
 340                         //merge values in simsMatrix
 341                         for (int n = 0; n < simMatrix.size(); n++)      {
 342                                 //row becomes merge of 2 groups
 343                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 344                                 simMatrix[n][row] = simMatrix[row][n];
 345                                 //delete column
 346                                 simMatrix[column][n] = -1000.0;
 347                                 simMatrix[n][column] = -1000.0;
 348                         }
 349                 }
 350
 351                 //adjust tree to make sure root to tip length is .5
 352                 int root = t->findRoot();
 353                 t->tree[root].setBranchLength((0.5 - t->tree[root].getLengthToLeaves()));
 354
 355                 //assemble tree
 356                 t->assembleTree();
 357
 358                 //print newick file
 359                 t->print(*out);
 360
 361         }
 362         catch(exception& e) {
 363                 m->errorOut(e, "BootSharedCommand", "createTree");
 364                 exit(1);
 365         }
 366 }
 367 /***********************************************************/
 368 void BootSharedCommand::printSims() {
 369         try {
 370                 m->mothurOut("simsMatrix"); m->mothurOutEndLine();
 371                 for (int k = 0; k < simMatrix.size(); k++)      {
 372                         for (int n = 0; n < simMatrix.size(); n++)      {
 373                                 m->mothurOut(toString(simMatrix[k][n]));  m->mothurOut("\t");
 374                         }
 375                         m->mothurOutEndLine();
 376                 }
 377
 378         }
 379         catch(exception& e) {
 380                 m->errorOut(e, "BootSharedCommand", "printSims");
 381                 exit(1);
 382         }
 383 }
 384 /***********************************************************/
 385 void BootSharedCommand::process(SharedOrderVector* order) {
 386         try{
 387                                 EstOutput data;
 388                                 vector<SharedRAbundVector*> subset;
 389
 390                                 //open an ostream for each calc to print to
 391                                 for (int z = 0; z < treeCalculators.size(); z++) {
 392                                         //create a new filename
 393                                         outputFile = outputDir + getRootName(getSimpleName(globaldata->inputFileName)) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 394                                         openOutputFile(outputFile, *(out[z]));
 395                                         outputNames.push_back(outputFile);
 396                                 }
 397
 398                                 m->mothurOut("Generating bootstrap trees..."); cout.flush();
 399
 400                                 //create a file for each calculator with the 1000 trees in it.
 401                                 for (int p = 0; p < iters; p++) {
 402
 403                                         util->getSharedVectorswithReplacement(globaldata->Groups, lookup, order);  //fills group vectors from order vector.
 404
 405
 406                                         //for each calculator
 407                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 408
 409                                                 //initialize simMatrix
 410                                                 simMatrix.clear();
 411                                                 simMatrix.resize(numGroups);
 412                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 413                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 414                                                                 simMatrix[m].push_back(0.0);
 415                                                         }
 416                                                 }
 417
 418                                                 //initialize index
 419                                                 index.clear();
 420                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 421
 422                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 423                                                         for (int l = k; l < lookup.size(); l++) {
 424                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 425                                                                         subset.clear(); //clear out old pair of sharedrabunds
 426                                                                         //add new pair of sharedrabunds
 427                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 428
 429                                                                         //get estimated similarity between 2 groups
 430                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 431                                                                         //save values in similarity matrix
 432                                                                         simMatrix[k][l] = data[0];
 433                                                                         simMatrix[l][k] = data[0];
 434                                                                 }
 435                                                         }
 436                                                 }
 437
 438                                                 tempTree = new Tree();
 439
 440                                                 //creates tree from similarity matrix and write out file
 441                                                 createTree(out[i], tempTree);
 442
 443                                                 //save trees for consensus command.
 444                                                 trees[i].push_back(tempTree);
 445                                         }
 446                                 }
 447
 448                                 m->mothurOut("\tDone."); m->mothurOutEndLine();
 449                                 //delete globaldata's tree
 450                                 //for (int m = 0; m < globaldata->gTree.size(); m++) {  delete globaldata->gTree[m];  }
 451                                 //globaldata->gTree.clear();
 452
 453
 454                                 //create consensus trees for each bootstrapped tree set
 455                                 for (int k = 0; k < trees.size(); k++) {
 456
 457                                         m->mothurOut("Generating consensus tree for " + treeCalculators[k]->getName()); m->mothurOutEndLine();
 458
 459                                         //set global data to calc trees
 460                                         globaldata->gTree = trees[k];
 461
 462                                         string filename = outputDir + getRootName(getSimpleName(globaldata->inputFileName)) + treeCalculators[k]->getName() + ".boot" + order->getLabel();
 463                                         consensus = new ConcensusCommand(filename);
 464                                         consensus->execute();
 465                                         delete consensus;
 466
 467                                         outputNames.push_back(filename + ".cons.pairs");
 468                                         outputNames.push_back(filename + ".cons.tre");
 469
 470                                         //delete globaldata's tree
 471                                         //for (int m = 0; m < globaldata->gTree.size(); m++) {  delete globaldata->gTree[m];  }
 472                                         //globaldata->gTree.clear();
 473
 474                                 }
 475
 476
 477
 478                                 //close ostream for each calc
 479                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 480
 481         }
 482         catch(exception& e) {
 483                 m->errorOut(e, "BootSharedCommand", "process");
 484                 exit(1);
 485         }
 486 }
 487 /***********************************************************/
 488
 489
 490