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                 EstOutput data;
  94                 vector<SharedRAbundVector*> subset;
  95
  96                 //if the users entered no valid calculators don't execute command
  97                 if (treeCalculators.size() == 0) { return 0; }
  98
  99                 if (format == "sharedfile") {
 100                         read = new ReadOTUFile(globaldata->inputFileName);
 101                         read->read(&*globaldata);
 102
 103                         input = globaldata->ginput;
 104                         order = input->getSharedOrderVector();
 105                 }else {
 106                         //you are using a list and a groupfile
 107                         read = new ReadOTUFile(globaldata->inputFileName);
 108                         read->read(&*globaldata);
 109
 110                         input = globaldata->ginput;
 111                         SharedList = globaldata->gSharedList;
 112                         order = SharedList->getSharedOrderVector();
 113                 }
 114
 115                 //set users groups
 116                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 117                 numGroups = globaldata->Groups.size();
 118
 119                 //clear globaldatas old tree names if any
 120                 globaldata->Treenames.clear();
 121
 122                 //fills globaldatas tree names
 123                 globaldata->Treenames = globaldata->Groups;
 124
 125                 //create treemap class from groupmap for tree class to use
 126                 tmap = new TreeMap();
 127                 tmap->makeSim(globaldata->gGroupmap);
 128                 globaldata->gTreemap = tmap;
 129
 130                 while(order != NULL){
 131
 132                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 133
 134                                 cout << order->getLabel() << '\t' << count << endl;
 135
 136                                 //open an ostream for each calc to print to
 137                                 for (int z = 0; z < treeCalculators.size(); z++) {
 138                                         //create a new filename
 139                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 140                                         openOutputFile(outputFile, *(out[z]));
 141                                 }
 142
 143                                 //create a file for each calculator with the 1000 trees in it.
 144                                 for (int p = 0; p < iters; p++) {
 145
 146                                         util->getSharedVectorswithReplacement(globaldata->Groups, lookup, order);  //fills group vectors from order vector.
 147
 148                                         //for each calculator
 149                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 150
 151                                                 //initialize simMatrix
 152                                                 simMatrix.clear();
 153                                                 simMatrix.resize(numGroups);
 154                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 155                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 156                                                                 simMatrix[m].push_back(0.0);
 157                                                         }
 158                                                 }
 159
 160                                                 //initialize index
 161                                                 index.clear();
 162                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 163
 164                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 165                                                         for (int l = k; l < lookup.size(); l++) {
 166                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 167                                                                         subset.clear(); //clear out old pair of sharedrabunds
 168                                                                         //add new pair of sharedrabunds
 169                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 170
 171                                                                         //get estimated similarity between 2 groups
 172                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 173                                                                         //save values in similarity matrix
 174                                                                         simMatrix[k][l] = data[0];
 175                                                                         simMatrix[l][k] = data[0];
 176                                                                 }
 177                                                         }
 178                                                 }
 179
 180                                                 //creates tree from similarity matrix and write out file
 181                                                 createTree(out[i]);
 182                                         }
 183                                 }
 184                                 //close ostream for each calc
 185                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 186                         }
 187
 188                         //get next line to process
 189                         if (format == "sharedfile") {
 190                                 order = input->getSharedOrderVector();
 191                         }else {
 192                                 //you are using a list and a groupfile
 193                                 SharedList = input->getSharedListVector(); //get new list vector to process
 194                                 if (SharedList != NULL) {
 195                                         order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
 196                                 }else {
 197                                         break;
 198                                 }
 199                         }
 200                         count++;
 201                 }
 202
 203                 //reset groups parameter
 204                 globaldata->Groups.clear();  globaldata->setGroups("");
 205
 206                 return 0;
 207         }
 208         catch(exception& e) {
 209                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 210                 exit(1);
 211         }
 212         catch(...) {
 213                 cout << "An unknown error has occurred in the BootSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 214                 exit(1);
 215         }
 216 }
 217 //**********************************************************************************************************************
 218
 219 void BootSharedCommand::createTree(ostream* out){
 220         try {
 221                 //create tree
 222                 t = new Tree();
 223
 224                 //do merges and create tree structure by setting parents and children
 225                 //there are numGroups - 1 merges to do
 226                 for (int i = 0; i < (numGroups - 1); i++) {
 227
 228                         float largest = -1.0;
 229                         int row, column;
 230                         //find largest value in sims matrix by searching lower triangle
 231                         for (int j = 1; j < simMatrix.size(); j++) {
 232                                 for (int k = 0; k < j; k++) {
 233                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 234                                 }
 235                         }
 236
 237                         //set non-leaf node info and update leaves to know their parents
 238                         //non-leaf
 239                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 240
 241                         //parents
 242                         t->tree[index[row]].setParent(numGroups + i);
 243                         t->tree[index[column]].setParent(numGroups + i);
 244
 245                         //blength = distance / 2;
 246                         float blength = ((1.0 - largest) / 2);
 247
 248                         //branchlengths
 249                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 250                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 251
 252                         //set your length to leaves to your childs length plus branchlength
 253                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 254
 255
 256                         //update index
 257                         index[row] = numGroups+i;
 258                         index[column] = numGroups+i;
 259
 260                         //zero out highest value that caused the merge.
 261                         simMatrix[row][column] = -1.0;
 262                         simMatrix[column][row] = -1.0;
 263
 264                         //merge values in simsMatrix
 265                         for (int n = 0; n < simMatrix.size(); n++)      {
 266                                 //row becomes merge of 2 groups
 267                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 268                                 simMatrix[n][row] = simMatrix[row][n];
 269                                 //delete column
 270                                 simMatrix[column][n] = -1.0;
 271                                 simMatrix[n][column] = -1.0;
 272                         }
 273                 }
 274
 275                 //assemble tree
 276                 t->assembleTree();
 277
 278                 //print newick file
 279                 t->print(*out);
 280
 281                 //delete tree
 282                 delete t;
 283
 284         }
 285         catch(exception& e) {
 286                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 287                 exit(1);
 288         }
 289         catch(...) {
 290                 cout << "An unknown error has occurred in the BootSharedCommand class function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 291                 exit(1);
 292         }
 293 }
 294 /***********************************************************/
 295 void BootSharedCommand::printSims() {
 296         try {
 297                 cout << "simsMatrix" << endl;
 298                 for (int m = 0; m < simMatrix.size(); m++)      {
 299                         for (int n = 0; n < simMatrix.size(); n++)      {
 300                                 cout << simMatrix[m][n] << '\t';
 301                         }
 302                         cout << endl;
 303                 }
 304
 305         }
 306         catch(exception& e) {
 307                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 308                 exit(1);
 309         }
 310         catch(...) {
 311                 cout << "An unknown error has occurred in the BootSharedCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 312                 exit(1);
 313         }
 314 }
 315 /***********************************************************/
 316
 317