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(){
  26         try {
  27                 globaldata = GlobalData::getInstance();
  28                 format = globaldata->getFormat();
  29                 convert(globaldata->getIters(), iters);
  30                 validCalculator = new ValidCalculators();
  31                 util = new SharedUtil();
  32
  33
  34                 int i;
  35                 for (i=0; i<globaldata->Estimators.size(); i++) {
  36                         if (validCalculator->isValidCalculator("boot", globaldata->Estimators[i]) == true) {
  37                                 if (globaldata->Estimators[i] == "jabund") {
  38                                         treeCalculators.push_back(new JAbund());
  39                                 }else if (globaldata->Estimators[i] == "sorabund") {
  40                                         treeCalculators.push_back(new SorAbund());
  41                                 }else if (globaldata->Estimators[i] == "jclass") {
  42                                         treeCalculators.push_back(new Jclass());
  43                                 }else if (globaldata->Estimators[i] == "sorclass") {
  44                                         treeCalculators.push_back(new SorClass());
  45                                 }else if (globaldata->Estimators[i] == "jest") {
  46                                         treeCalculators.push_back(new Jest());
  47                                 }else if (globaldata->Estimators[i] == "sorest") {
  48                                         treeCalculators.push_back(new SorEst());
  49                                 }else if (globaldata->Estimators[i] == "thetayc") {
  50                                         treeCalculators.push_back(new ThetaYC());
  51                                 }else if (globaldata->Estimators[i] == "thetan") {
  52                                         treeCalculators.push_back(new ThetaN());
  53                                 }else if (globaldata->Estimators[i] == "morisitahorn") {
  54                                         treeCalculators.push_back(new MorHorn());
  55                                 }else if (globaldata->Estimators[i] == "braycurtis") {
  56                                         treeCalculators.push_back(new BrayCurtis());
  57                                 }
  58                         }
  59                 }
  60
  61                 ofstream* temp;
  62                 for (int i=0; i < treeCalculators.size(); i++) {
  63                         temp = new ofstream;
  64                         out.push_back(temp);
  65                 }
  66
  67                 //reset calc for next command
  68                 globaldata->setCalc("");
  69
  70         }
  71         catch(exception& e) {
  72                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  73                 exit(1);
  74         }
  75         catch(...) {
  76                 cout << "An unknown error has occurred in the BootSharedCommand class function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81
  82 BootSharedCommand::~BootSharedCommand(){
  83         delete input;
  84         delete read;
  85         delete util;
  86 }
  87
  88 //**********************************************************************************************************************
  89
  90 int BootSharedCommand::execute(){
  91         try {
  92                 int count = 1;
  93
  94                 //if the users entered no valid calculators don't execute command
  95                 if (treeCalculators.size() == 0) { return 0; }
  96
  97                 //read first line
  98                 read = new ReadOTUFile(globaldata->inputFileName);
  99                 read->read(&*globaldata);
 100                 input = globaldata->ginput;
 101                 order = input->getSharedOrderVector();
 102                 SharedOrderVector* lastOrder = order;
 103
 104                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 105                 set<string> processedLabels;
 106                 set<string> userLabels = globaldata->labels;
 107
 108                 //set users groups
 109                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 110                 numGroups = globaldata->Groups.size();
 111
 112                 //clear globaldatas old tree names if any
 113                 globaldata->Treenames.clear();
 114
 115                 //fills globaldatas tree names
 116                 globaldata->Treenames = globaldata->Groups;
 117
 118                 //create treemap class from groupmap for tree class to use
 119                 tmap = new TreeMap();
 120                 tmap->makeSim(globaldata->gGroupmap);
 121                 globaldata->gTreemap = tmap;
 122
 123                 while((order != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0))) {
 124
 125                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 126
 127                                 cout << order->getLabel() << '\t' << count << endl;
 128                                 process(order);
 129
 130                                 processedLabels.insert(order->getLabel());
 131                                 userLabels.erase(order->getLabel());
 132
 133                         //you have a label the user want that is smaller than this line and the last line has not already been processed
 134                         }
 135
 136                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastOrder->getLabel()) != 1)) {
 137
 138                                 cout << lastOrder->getLabel() << '\t' << count << endl;
 139                                 process(lastOrder);
 140
 141                                 processedLabels.insert(lastOrder->getLabel());
 142                                 userLabels.erase(lastOrder->getLabel());
 143                         }
 144
 145                         if (count != 1) { delete lastOrder; }
 146                         lastOrder = order;
 147
 148                         //get next line to process
 149                         order = input->getSharedOrderVector();
 150                         count++;
 151                 }
 152
 153                 //output error messages about any remaining user labels
 154                 set<string>::iterator it;
 155                 bool needToRun = false;
 156                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 157                         cout << "Your file does not include the label "<< *it;
 158                         if (processedLabels.count(lastOrder->getLabel()) != 1) {
 159                                 cout << ". I will use " << lastOrder->getLabel() << "." << endl;
 160                                 needToRun = true;
 161                         }else {
 162                                 cout << ". Please refer to " << lastOrder->getLabel() << "." << endl;
 163                         }
 164                 }
 165
 166                 //run last line if you need to
 167                 if (needToRun == true)  {
 168                         process(lastOrder);
 169                         cout << lastOrder->getLabel() << '\t' << count << endl;
 170                 }
 171
 172                 delete lastOrder;
 173
 174                 //reset groups parameter
 175                 globaldata->Groups.clear();  globaldata->setGroups("");
 176
 177                 return 0;
 178         }
 179         catch(exception& e) {
 180                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 181                 exit(1);
 182         }
 183         catch(...) {
 184                 cout << "An unknown error has occurred in the BootSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187 }
 188 //**********************************************************************************************************************
 189
 190 void BootSharedCommand::createTree(ostream* out){
 191         try {
 192                 //create tree
 193                 t = new Tree();
 194
 195                 //do merges and create tree structure by setting parents and children
 196                 //there are numGroups - 1 merges to do
 197                 for (int i = 0; i < (numGroups - 1); i++) {
 198
 199                         float largest = -1.0;
 200                         int row, column;
 201                         //find largest value in sims matrix by searching lower triangle
 202                         for (int j = 1; j < simMatrix.size(); j++) {
 203                                 for (int k = 0; k < j; k++) {
 204                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 205                                 }
 206                         }
 207
 208                         //set non-leaf node info and update leaves to know their parents
 209                         //non-leaf
 210                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 211
 212                         //parents
 213                         t->tree[index[row]].setParent(numGroups + i);
 214                         t->tree[index[column]].setParent(numGroups + i);
 215
 216                         //blength = distance / 2;
 217                         float blength = ((1.0 - largest) / 2);
 218
 219                         //branchlengths
 220                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 221                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 222
 223                         //set your length to leaves to your childs length plus branchlength
 224                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 225
 226
 227                         //update index
 228                         index[row] = numGroups+i;
 229                         index[column] = numGroups+i;
 230
 231                         //zero out highest value that caused the merge.
 232                         simMatrix[row][column] = -1.0;
 233                         simMatrix[column][row] = -1.0;
 234
 235                         //merge values in simsMatrix
 236                         for (int n = 0; n < simMatrix.size(); n++)      {
 237                                 //row becomes merge of 2 groups
 238                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 239                                 simMatrix[n][row] = simMatrix[row][n];
 240                                 //delete column
 241                                 simMatrix[column][n] = -1.0;
 242                                 simMatrix[n][column] = -1.0;
 243                         }
 244                 }
 245
 246                 //assemble tree
 247                 t->assembleTree();
 248
 249                 //print newick file
 250                 t->print(*out);
 251
 252                 //delete tree
 253                 delete t;
 254
 255         }
 256         catch(exception& e) {
 257                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 258                 exit(1);
 259         }
 260         catch(...) {
 261                 cout << "An unknown error has occurred in the BootSharedCommand class function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 262                 exit(1);
 263         }
 264 }
 265 /***********************************************************/
 266 void BootSharedCommand::printSims() {
 267         try {
 268                 cout << "simsMatrix" << endl;
 269                 for (int m = 0; m < simMatrix.size(); m++)      {
 270                         for (int n = 0; n < simMatrix.size(); n++)      {
 271                                 cout << simMatrix[m][n] << '\t';
 272                         }
 273                         cout << endl;
 274                 }
 275
 276         }
 277         catch(exception& e) {
 278                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 279                 exit(1);
 280         }
 281         catch(...) {
 282                 cout << "An unknown error has occurred in the BootSharedCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 283                 exit(1);
 284         }
 285 }
 286 /***********************************************************/
 287 void BootSharedCommand::process(SharedOrderVector* order) {
 288         try{
 289                                 EstOutput data;
 290                                 vector<SharedRAbundVector*> subset;
 291
 292                                 //open an ostream for each calc to print to
 293                                 for (int z = 0; z < treeCalculators.size(); z++) {
 294                                         //create a new filename
 295                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 296                                         openOutputFile(outputFile, *(out[z]));
 297                                 }
 298
 299                                 //create a file for each calculator with the 1000 trees in it.
 300                                 for (int p = 0; p < iters; p++) {
 301
 302                                         util->getSharedVectorswithReplacement(globaldata->Groups, lookup, order);  //fills group vectors from order vector.
 303
 304                                         //for each calculator
 305                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 306
 307                                                 //initialize simMatrix
 308                                                 simMatrix.clear();
 309                                                 simMatrix.resize(numGroups);
 310                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 311                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 312                                                                 simMatrix[m].push_back(0.0);
 313                                                         }
 314                                                 }
 315
 316                                                 //initialize index
 317                                                 index.clear();
 318                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 319
 320                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 321                                                         for (int l = k; l < lookup.size(); l++) {
 322                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 323                                                                         subset.clear(); //clear out old pair of sharedrabunds
 324                                                                         //add new pair of sharedrabunds
 325                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 326
 327                                                                         //get estimated similarity between 2 groups
 328                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 329                                                                         //save values in similarity matrix
 330                                                                         simMatrix[k][l] = data[0];
 331                                                                         simMatrix[l][k] = data[0];
 332                                                                 }
 333                                                         }
 334                                                 }
 335
 336                                                 //creates tree from similarity matrix and write out file
 337                                                 createTree(out[i]);
 338                                         }
 339                                 }
 340                                 //close ostream for each calc
 341                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 342
 343         }
 344         catch(exception& e) {
 345                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 346                 exit(1);
 347         }
 348         catch(...) {
 349                 cout << "An unknown error has occurred in the BootSharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 350                 exit(1);
 351         }
 352 }
 353 /***********************************************************/
 354
 355
 356