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                         OptionParser parser(option);
  44                         map<string,string> parameters = parser.getParameters();
  45
  46                         ValidParameters validParameter;
  47
  48                         //check to make sure all parameters are valid for command
  49                         for (map<string,string>::iterator 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                         if (abort == false) {
 103
 104                                 validCalculator = new ValidCalculators();
 105
 106                                 int i;
 107                                 for (i=0; i<Estimators.size(); i++) {
 108                                         if (validCalculator->isValidCalculator("boot", Estimators[i]) == true) {
 109                                                 if (Estimators[i] == "jabund") {
 110                                                         treeCalculators.push_back(new JAbund());
 111                                                 }else if (Estimators[i] == "sorabund") {
 112                                                         treeCalculators.push_back(new SorAbund());
 113                                                 }else if (Estimators[i] == "jclass") {
 114                                                         treeCalculators.push_back(new Jclass());
 115                                                 }else if (Estimators[i] == "sorclass") {
 116                                                         treeCalculators.push_back(new SorClass());
 117                                                 }else if (Estimators[i] == "jest") {
 118                                                         treeCalculators.push_back(new Jest());
 119                                                 }else if (Estimators[i] == "sorest") {
 120                                                         treeCalculators.push_back(new SorEst());
 121                                                 }else if (Estimators[i] == "thetayc") {
 122                                                         treeCalculators.push_back(new ThetaYC());
 123                                                 }else if (Estimators[i] == "thetan") {
 124                                                         treeCalculators.push_back(new ThetaN());
 125                                                 }else if (Estimators[i] == "morisitahorn") {
 126                                                         treeCalculators.push_back(new MorHorn());
 127                                                 }else if (Estimators[i] == "braycurtis") {
 128                                                         treeCalculators.push_back(new BrayCurtis());
 129                                                 }
 130                                         }
 131                                 }
 132
 133                                 delete validCalculator;
 134
 135                                 ofstream* tempo;
 136                                 for (int i=0; i < treeCalculators.size(); i++) {
 137                                         tempo = new ofstream;
 138                                         out.push_back(tempo);
 139                                 }
 140                         }
 141                 }
 142
 143         }
 144         catch(exception& e) {
 145                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 146                 exit(1);
 147         }
 148         catch(...) {
 149                 cout << "An unknown error has occurred in the BootSharedCommand class function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 150                 exit(1);
 151         }
 152 }
 153
 154 //**********************************************************************************************************************
 155
 156 void BootSharedCommand::help(){
 157         try {
 158                 cout << "The bootstrap.shared command can only be executed after a successful read.otu command." << "\n";
 159                 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";
 160                 cout << "The groups parameter allows you to specify which of the groups in your groupfile you would like included used." << "\n";
 161                 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";
 162                 cout << "The bootstrap.shared command should be in the following format: bootstrap.shared(groups=yourGroups, calc=yourCalcs, line=yourLines, label=yourLabels, iters=yourIters)." << "\n";
 163                 cout << "Example bootstrap.shared(groups=A-B-C, line=1-3-5, calc=jabund-sorabund, iters=100)." << "\n";
 164                 cout << "The default value for groups is all the groups in your groupfile." << "\n";
 165                 cout << "The default value for calc is jclass-thetayc. The default for iters is 1000." << "\n";
 166         }
 167         catch(exception& e) {
 168                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 169                 exit(1);
 170         }
 171         catch(...) {
 172                 cout << "An unknown error has occurred in the BootSharedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 173                 exit(1);
 174         }
 175 }
 176
 177 //**********************************************************************************************************************
 178
 179 BootSharedCommand::~BootSharedCommand(){
 180         //made new in execute
 181         if (abort == false) {
 182                 delete input; globaldata->ginput = NULL;
 183                 delete read;
 184                 delete util;
 185                 globaldata->gorder = NULL;
 186         }
 187 }
 188
 189 //**********************************************************************************************************************
 190
 191 int BootSharedCommand::execute(){
 192         try {
 193
 194                 if (abort == true) {    return 0;       }
 195
 196                 int count = 1;
 197                 util = new SharedUtil();
 198
 199                 //read first line
 200                 read = new ReadOTUFile(globaldata->inputFileName);
 201                 read->read(&*globaldata);
 202                 input = globaldata->ginput;
 203                 order = input->getSharedOrderVector();
 204                 string lastLabel = order->getLabel();
 205
 206                 //if the users entered no valid calculators don't execute command
 207                 if (treeCalculators.size() == 0) { return 0; }
 208
 209                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 210                 set<string> processedLabels;
 211                 set<string> userLabels = labels;
 212                 set<int> userLines = lines;
 213
 214                 //set users groups
 215                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 216                 numGroups = globaldata->Groups.size();
 217
 218                 //clear globaldatas old tree names if any
 219                 globaldata->Treenames.clear();
 220
 221                 //fills globaldatas tree names
 222                 globaldata->Treenames = globaldata->Groups;
 223
 224                 //create treemap class from groupmap for tree class to use
 225                 tmap = new TreeMap();
 226                 tmap->makeSim(globaldata->gGroupmap);
 227                 globaldata->gTreemap = tmap;
 228
 229                 while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 230
 231                         if(allLines == 1 || lines.count(count) == 1 || labels.count(order->getLabel()) == 1){
 232
 233                                 cout << order->getLabel() << '\t' << count << endl;
 234                                 process(order);
 235
 236                                 processedLabels.insert(order->getLabel());
 237                                 userLabels.erase(order->getLabel());
 238                                 userLines.erase(count);
 239                         }
 240
 241                         //you have a label the user want that is smaller than this line and the last line has not already been processed
 242                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 243
 244                                 delete order;
 245                                 order = input->getSharedOrderVector(lastLabel);
 246                                 cout << order->getLabel() << '\t' << count << endl;
 247                                 process(order);
 248
 249                                 processedLabels.insert(order->getLabel());
 250                                 userLabels.erase(order->getLabel());
 251                         }
 252
 253
 254                         lastLabel = order->getLabel();
 255
 256                         //get next line to process
 257                         delete order;
 258                         order = input->getSharedOrderVector();
 259                         count++;
 260                 }
 261
 262                 //output error messages about any remaining user labels
 263                 set<string>::iterator it;
 264                 bool needToRun = false;
 265                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 266                         cout << "Your file does not include the label "<< *it;
 267                         if (processedLabels.count(lastLabel) != 1) {
 268                                 cout << ". I will use " << lastLabel << "." << endl;
 269                                 needToRun = true;
 270                         }else {
 271                                 cout << ". Please refer to " << lastLabel << "." << endl;
 272                         }
 273                 }
 274
 275                 //run last line if you need to
 276                 if (needToRun == true)  {
 277                                 delete order;
 278                                 order = input->getSharedOrderVector(lastLabel);
 279                                 cout << order->getLabel() << '\t' << count << endl;
 280                                 process(order);
 281                                 delete order;
 282
 283                 }
 284
 285                 //reset groups parameter
 286                 globaldata->Groups.clear();
 287
 288                 return 0;
 289         }
 290         catch(exception& e) {
 291                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 292                 exit(1);
 293         }
 294         catch(...) {
 295                 cout << "An unknown error has occurred in the BootSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 296                 exit(1);
 297         }
 298 }
 299 //**********************************************************************************************************************
 300
 301 void BootSharedCommand::createTree(ostream* out){
 302         try {
 303                 //create tree
 304                 t = new Tree();
 305
 306                 //do merges and create tree structure by setting parents and children
 307                 //there are numGroups - 1 merges to do
 308                 for (int i = 0; i < (numGroups - 1); i++) {
 309
 310                         float largest = -1.0;
 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] = -1.0;
 344                         simMatrix[column][row] = -1.0;
 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] = -1.0;
 353                                 simMatrix[n][column] = -1.0;
 354                         }
 355                 }
 356
 357                 //assemble tree
 358                 t->assembleTree();
 359
 360                 //print newick file
 361                 t->print(*out);
 362
 363                 //delete tree
 364                 delete t;
 365
 366         }
 367         catch(exception& e) {
 368                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 369                 exit(1);
 370         }
 371         catch(...) {
 372                 cout << "An unknown error has occurred in the BootSharedCommand class function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 373                 exit(1);
 374         }
 375 }
 376 /***********************************************************/
 377 void BootSharedCommand::printSims() {
 378         try {
 379                 cout << "simsMatrix" << endl;
 380                 for (int m = 0; m < simMatrix.size(); m++)      {
 381                         for (int n = 0; n < simMatrix.size(); n++)      {
 382                                 cout << simMatrix[m][n] << '\t';
 383                         }
 384                         cout << endl;
 385                 }
 386
 387         }
 388         catch(exception& e) {
 389                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 390                 exit(1);
 391         }
 392         catch(...) {
 393                 cout << "An unknown error has occurred in the BootSharedCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 394                 exit(1);
 395         }
 396 }
 397 /***********************************************************/
 398 void BootSharedCommand::process(SharedOrderVector* order) {
 399         try{
 400                                 EstOutput data;
 401                                 vector<SharedRAbundVector*> subset;
 402
 403                                 //open an ostream for each calc to print to
 404                                 for (int z = 0; z < treeCalculators.size(); z++) {
 405                                         //create a new filename
 406                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 407                                         openOutputFile(outputFile, *(out[z]));
 408                                 }
 409
 410                                 //create a file for each calculator with the 1000 trees in it.
 411                                 for (int p = 0; p < iters; p++) {
 412
 413                                         util->getSharedVectorswithReplacement(Groups, lookup, order);  //fills group vectors from order vector.
 414
 415                                         //for each calculator
 416                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 417
 418                                                 //initialize simMatrix
 419                                                 simMatrix.clear();
 420                                                 simMatrix.resize(numGroups);
 421                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 422                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 423                                                                 simMatrix[m].push_back(0.0);
 424                                                         }
 425                                                 }
 426
 427                                                 //initialize index
 428                                                 index.clear();
 429                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 430
 431                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 432                                                         for (int l = k; l < lookup.size(); l++) {
 433                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 434                                                                         subset.clear(); //clear out old pair of sharedrabunds
 435                                                                         //add new pair of sharedrabunds
 436                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 437
 438                                                                         //get estimated similarity between 2 groups
 439                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 440                                                                         //save values in similarity matrix
 441                                                                         simMatrix[k][l] = data[0];
 442                                                                         simMatrix[l][k] = data[0];
 443                                                                 }
 444                                                         }
 445                                                 }
 446
 447                                                 //creates tree from similarity matrix and write out file
 448                                                 createTree(out[i]);
 449                                         }
 450                                 }
 451                                 //close ostream for each calc
 452                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 453
 454         }
 455         catch(exception& e) {
 456                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 457                 exit(1);
 458         }
 459         catch(...) {
 460                 cout << "An unknown error has occurred in the BootSharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 461                 exit(1);
 462         }
 463 }
 464 /***********************************************************/
 465
 466
 467