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                 lines.clear();
  31                 labels.clear();
  32                 Groups.clear();
  33                 Estimators.clear();
  34
  35                 //allow user to run help
  36                 if(option == "help") { help(); abort = true; }
  37
  38                 else {
  39                         //valid paramters for this command
  40                         string Array[] =  {"line","label","calc","groups","iters"};
  41                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  42
  43                         parser = new OptionParser();
  44                         parser->parse(option, parameters);  delete parser;
  45
  46                         ValidParameters* validParameter = new ValidParameters();
  47
  48                         //check to make sure all parameters are valid for command
  49                         for (it = parameters.begin(); it != parameters.end(); it++) {
  50                                 if (validParameter->isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  51                         }
  52
  53                         //make sure the user has already run the read.otu command
  54                         if (globaldata->getSharedFile() == "") {
  55                                 if (globaldata->getListFile() == "") { cout << "You must read a list and a group, or a shared before you can use the bootstrap.shared command." << endl; abort = true; }
  56                                 else if (globaldata->getGroupFile() == "") { cout << "You must read a list and a group, or a shared before you can use the bootstrap.shared command." << endl; abort = true; }
  57                         }
  58
  59                         //check for optional parameter and set defaults
  60                         // ...at some point should added some additional type checking...
  61                         line = validParameter->validFile(parameters, "line", false);
  62                         if (line == "not found") { line = "";  }
  63                         else {
  64                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  65                                 else { allLines = 1;  }
  66                         }
  67
  68                         label = validParameter->validFile(parameters, "label", false);
  69                         if (label == "not found") { label = ""; }
  70                         else {
  71                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  72                                 else { allLines = 1;  }
  73                         }
  74
  75                         //make sure user did not use both the line and label parameters
  76                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  77                         //if the user has not specified any line or labels use the ones from read.otu
  78                         else if((line == "") && (label == "")) {
  79                                 allLines = globaldata->allLines;
  80                                 labels = globaldata->labels;
  81                                 lines = globaldata->lines;
  82                         }
  83
  84                         groups = validParameter->validFile(parameters, "groups", false);
  85                         if (groups == "not found") { groups = ""; }
  86                         else {
  87                                 splitAtDash(groups, Groups);
  88                                 globaldata->Groups = Groups;
  89                         }
  90
  91                         calc = validParameter->validFile(parameters, "calc", false);
  92                         if (calc == "not found") { calc = "jclass-thetayc";  }
  93                         else {
  94                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
  95                         }
  96                         splitAtDash(calc, Estimators);
  97
  98                         string temp;
  99                         temp = validParameter->validFile(parameters, "iters", false);                   if (temp == "not found") { temp = "1000"; }
 100                         convert(temp, iters);
 101
 102                         delete validParameter;
 103
 104                         if (abort == false) {
 105
 106                                 validCalculator = new ValidCalculators();
 107
 108                                 int i;
 109                                 for (i=0; i<Estimators.size(); i++) {
 110                                         if (validCalculator->isValidCalculator("boot", Estimators[i]) == true) {
 111                                                 if (Estimators[i] == "jabund") {
 112                                                         treeCalculators.push_back(new JAbund());
 113                                                 }else if (Estimators[i] == "sorabund") {
 114                                                         treeCalculators.push_back(new SorAbund());
 115                                                 }else if (Estimators[i] == "jclass") {
 116                                                         treeCalculators.push_back(new Jclass());
 117                                                 }else if (Estimators[i] == "sorclass") {
 118                                                         treeCalculators.push_back(new SorClass());
 119                                                 }else if (Estimators[i] == "jest") {
 120                                                         treeCalculators.push_back(new Jest());
 121                                                 }else if (Estimators[i] == "sorest") {
 122                                                         treeCalculators.push_back(new SorEst());
 123                                                 }else if (Estimators[i] == "thetayc") {
 124                                                         treeCalculators.push_back(new ThetaYC());
 125                                                 }else if (Estimators[i] == "thetan") {
 126                                                         treeCalculators.push_back(new ThetaN());
 127                                                 }else if (Estimators[i] == "morisitahorn") {
 128                                                         treeCalculators.push_back(new MorHorn());
 129                                                 }else if (Estimators[i] == "braycurtis") {
 130                                                         treeCalculators.push_back(new BrayCurtis());
 131                                                 }
 132                                         }
 133                                 }
 134
 135                                 delete validCalculator;
 136
 137                                 ofstream* tempo;
 138                                 for (int i=0; i < treeCalculators.size(); i++) {
 139                                         tempo = new ofstream;
 140                                         out.push_back(tempo);
 141                                 }
 142                         }
 143                 }
 144
 145         }
 146         catch(exception& e) {
 147                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 148                 exit(1);
 149         }
 150         catch(...) {
 151                 cout << "An unknown error has occurred in the BootSharedCommand class function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 152                 exit(1);
 153         }
 154 }
 155
 156 //**********************************************************************************************************************
 157
 158 void BootSharedCommand::help(){
 159         try {
 160                 cout << "The bootstrap.shared command can only be executed after a successful read.otu command." << "\n";
 161                 cout << "The bootstrap.shared command parameters are groups, calc, iters, line and label.  You may not use line and label at the same time." << "\n";
 162                 cout << "The groups parameter allows you to specify which of the groups in your groupfile you would like included used." << "\n";
 163                 cout << "The group names are separated by dashes. The line and label allow you to select what distance levels you would like trees created for, and are also separated by dashes." << "\n";
 164                 cout << "The bootstrap.shared command should be in the following format: bootstrap.shared(groups=yourGroups, calc=yourCalcs, line=yourLines, label=yourLabels, iters=yourIters)." << "\n";
 165                 cout << "Example bootstrap.shared(groups=A-B-C, line=1-3-5, calc=jabund-sorabund, iters=100)." << "\n";
 166                 cout << "The default value for groups is all the groups in your groupfile." << "\n";
 167                 cout << "The default value for calc is jclass-thetayc. The default for iters is 1000." << "\n";
 168         }
 169         catch(exception& e) {
 170                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 171                 exit(1);
 172         }
 173         catch(...) {
 174                 cout << "An unknown error has occurred in the BootSharedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 175                 exit(1);
 176         }
 177 }
 178
 179 //**********************************************************************************************************************
 180
 181 BootSharedCommand::~BootSharedCommand(){
 182         //made new in execute
 183         if (abort == false) {
 184                 delete input;
 185                 delete read;
 186                 delete util;
 187         }
 188 }
 189
 190 //**********************************************************************************************************************
 191
 192 int BootSharedCommand::execute(){
 193         try {
 194
 195                 if (abort == true) {    return 0;       }
 196
 197                 int count = 1;
 198                 util = new SharedUtil();
 199
 200                 //read first line
 201                 read = new ReadOTUFile(globaldata->inputFileName);
 202                 read->read(&*globaldata);
 203                 input = globaldata->ginput;
 204                 order = input->getSharedOrderVector();
 205                 SharedOrderVector* lastOrder = order;
 206
 207                 //if the users entered no valid calculators don't execute command
 208                 if (treeCalculators.size() == 0) { return 0; }
 209
 210                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 211                 set<string> processedLabels;
 212                 set<string> userLabels = labels;
 213                 set<int> userLines = lines;
 214
 215                 //set users groups
 216                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 217                 numGroups = globaldata->Groups.size();
 218
 219                 //clear globaldatas old tree names if any
 220                 globaldata->Treenames.clear();
 221
 222                 //fills globaldatas tree names
 223                 globaldata->Treenames = globaldata->Groups;
 224
 225                 //create treemap class from groupmap for tree class to use
 226                 tmap = new TreeMap();
 227                 tmap->makeSim(globaldata->gGroupmap);
 228                 globaldata->gTreemap = tmap;
 229
 230                 while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 231
 232                         if(allLines == 1 || lines.count(count) == 1 || labels.count(order->getLabel()) == 1){
 233
 234                                 cout << order->getLabel() << '\t' << count << endl;
 235                                 process(order);
 236
 237                                 processedLabels.insert(order->getLabel());
 238                                 userLabels.erase(order->getLabel());
 239                                 userLines.erase(count);
 240                         }
 241
 242                         //you have a label the user want that is smaller than this line and the last line has not already been processed
 243                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastOrder->getLabel()) != 1)) {
 244
 245                                 cout << lastOrder->getLabel() << '\t' << count << endl;
 246                                 process(lastOrder);
 247
 248                                 processedLabels.insert(lastOrder->getLabel());
 249                                 userLabels.erase(lastOrder->getLabel());
 250                         }
 251
 252                         if (count != 1) { delete lastOrder; }
 253                         lastOrder = order;
 254
 255                         //get next line to process
 256                         order = input->getSharedOrderVector();
 257                         count++;
 258                 }
 259
 260                 //output error messages about any remaining user labels
 261                 set<string>::iterator it;
 262                 bool needToRun = false;
 263                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 264                         cout << "Your file does not include the label "<< *it;
 265                         if (processedLabels.count(lastOrder->getLabel()) != 1) {
 266                                 cout << ". I will use " << lastOrder->getLabel() << "." << endl;
 267                                 needToRun = true;
 268                         }else {
 269                                 cout << ". Please refer to " << lastOrder->getLabel() << "." << endl;
 270                         }
 271                 }
 272
 273                 //run last line if you need to
 274                 if (needToRun == true)  {
 275                         process(lastOrder);
 276                         cout << lastOrder->getLabel() << '\t' << count << endl;
 277                 }
 278
 279                 delete lastOrder;
 280
 281                 //reset groups parameter
 282                 globaldata->Groups.clear();
 283
 284                 return 0;
 285         }
 286         catch(exception& e) {
 287                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 288                 exit(1);
 289         }
 290         catch(...) {
 291                 cout << "An unknown error has occurred in the BootSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 292                 exit(1);
 293         }
 294 }
 295 //**********************************************************************************************************************
 296
 297 void BootSharedCommand::createTree(ostream* out){
 298         try {
 299                 //create tree
 300                 t = new Tree();
 301
 302                 //do merges and create tree structure by setting parents and children
 303                 //there are numGroups - 1 merges to do
 304                 for (int i = 0; i < (numGroups - 1); i++) {
 305
 306                         float largest = -1.0;
 307                         int row, column;
 308                         //find largest value in sims matrix by searching lower triangle
 309                         for (int j = 1; j < simMatrix.size(); j++) {
 310                                 for (int k = 0; k < j; k++) {
 311                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 312                                 }
 313                         }
 314
 315                         //set non-leaf node info and update leaves to know their parents
 316                         //non-leaf
 317                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 318
 319                         //parents
 320                         t->tree[index[row]].setParent(numGroups + i);
 321                         t->tree[index[column]].setParent(numGroups + i);
 322
 323                         //blength = distance / 2;
 324                         float blength = ((1.0 - largest) / 2);
 325
 326                         //branchlengths
 327                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 328                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 329
 330                         //set your length to leaves to your childs length plus branchlength
 331                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 332
 333
 334                         //update index
 335                         index[row] = numGroups+i;
 336                         index[column] = numGroups+i;
 337
 338                         //zero out highest value that caused the merge.
 339                         simMatrix[row][column] = -1.0;
 340                         simMatrix[column][row] = -1.0;
 341
 342                         //merge values in simsMatrix
 343                         for (int n = 0; n < simMatrix.size(); n++)      {
 344                                 //row becomes merge of 2 groups
 345                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 346                                 simMatrix[n][row] = simMatrix[row][n];
 347                                 //delete column
 348                                 simMatrix[column][n] = -1.0;
 349                                 simMatrix[n][column] = -1.0;
 350                         }
 351                 }
 352
 353                 //assemble tree
 354                 t->assembleTree();
 355
 356                 //print newick file
 357                 t->print(*out);
 358
 359                 //delete tree
 360                 delete t;
 361
 362         }
 363         catch(exception& e) {
 364                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 365                 exit(1);
 366         }
 367         catch(...) {
 368                 cout << "An unknown error has occurred in the BootSharedCommand class function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 369                 exit(1);
 370         }
 371 }
 372 /***********************************************************/
 373 void BootSharedCommand::printSims() {
 374         try {
 375                 cout << "simsMatrix" << endl;
 376                 for (int m = 0; m < simMatrix.size(); m++)      {
 377                         for (int n = 0; n < simMatrix.size(); n++)      {
 378                                 cout << simMatrix[m][n] << '\t';
 379                         }
 380                         cout << endl;
 381                 }
 382
 383         }
 384         catch(exception& e) {
 385                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 386                 exit(1);
 387         }
 388         catch(...) {
 389                 cout << "An unknown error has occurred in the BootSharedCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 390                 exit(1);
 391         }
 392 }
 393 /***********************************************************/
 394 void BootSharedCommand::process(SharedOrderVector* order) {
 395         try{
 396                                 EstOutput data;
 397                                 vector<SharedRAbundVector*> subset;
 398
 399                                 //open an ostream for each calc to print to
 400                                 for (int z = 0; z < treeCalculators.size(); z++) {
 401                                         //create a new filename
 402                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 403                                         openOutputFile(outputFile, *(out[z]));
 404                                 }
 405
 406                                 //create a file for each calculator with the 1000 trees in it.
 407                                 for (int p = 0; p < iters; p++) {
 408
 409                                         util->getSharedVectorswithReplacement(Groups, lookup, order);  //fills group vectors from order vector.
 410
 411                                         //for each calculator
 412                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 413
 414                                                 //initialize simMatrix
 415                                                 simMatrix.clear();
 416                                                 simMatrix.resize(numGroups);
 417                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 418                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 419                                                                 simMatrix[m].push_back(0.0);
 420                                                         }
 421                                                 }
 422
 423                                                 //initialize index
 424                                                 index.clear();
 425                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 426
 427                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 428                                                         for (int l = k; l < lookup.size(); l++) {
 429                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 430                                                                         subset.clear(); //clear out old pair of sharedrabunds
 431                                                                         //add new pair of sharedrabunds
 432                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 433
 434                                                                         //get estimated similarity between 2 groups
 435                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 436                                                                         //save values in similarity matrix
 437                                                                         simMatrix[k][l] = data[0];
 438                                                                         simMatrix[l][k] = data[0];
 439                                                                 }
 440                                                         }
 441                                                 }
 442
 443                                                 //creates tree from similarity matrix and write out file
 444                                                 createTree(out[i]);
 445                                         }
 446                                 }
 447                                 //close ostream for each calc
 448                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 449
 450         }
 451         catch(exception& e) {
 452                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 453                 exit(1);
 454         }
 455         catch(...) {
 456                 cout << "An unknown error has occurred in the BootSharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 457                 exit(1);
 458         }
 459 }
 460 /***********************************************************/
 461
 462
 463