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
  21
  22 //**********************************************************************************************************************
  23
  24 BootSharedCommand::BootSharedCommand(){
  25         try {
  26                 globaldata = GlobalData::getInstance();
  27                 format = globaldata->getFormat();
  28                 convert(globaldata->getIters(), iters);
  29                 validCalculator = new ValidCalculators();
  30                 util = new SharedUtil();
  31
  32
  33                 int i;
  34                 for (i=0; i<globaldata->Estimators.size(); i++) {
  35                         if (validCalculator->isValidCalculator("boot", globaldata->Estimators[i]) == true) {
  36                                 if (globaldata->Estimators[i] == "jabund") {
  37                                         treeCalculators.push_back(new JAbund());
  38                                 }else if (globaldata->Estimators[i] == "sorensonabund") {
  39                                         treeCalculators.push_back(new SorAbund());
  40                                 }else if (globaldata->Estimators[i] == "jclass") {
  41                                         treeCalculators.push_back(new Jclass());
  42                                 }else if (globaldata->Estimators[i] == "sorclass") {
  43                                         treeCalculators.push_back(new SorClass());
  44                                 }else if (globaldata->Estimators[i] == "jest") {
  45                                         treeCalculators.push_back(new Jest());
  46                                 }else if (globaldata->Estimators[i] == "sorest") {
  47                                         treeCalculators.push_back(new SorEst());
  48                                 }else if (globaldata->Estimators[i] == "thetayc") {
  49                                         treeCalculators.push_back(new ThetaYC());
  50                                 }else if (globaldata->Estimators[i] == "thetan") {
  51                                         treeCalculators.push_back(new ThetaN());
  52                                 }else if (globaldata->Estimators[i] == "morisitahorn") {
  53                                         treeCalculators.push_back(new MorHorn());
  54                                 }
  55                         }
  56                 }
  57
  58                 ofstream* temp;
  59                 for (int i=0; i < treeCalculators.size(); i++) {
  60                         temp = new ofstream;
  61                         out.push_back(temp);
  62                 }
  63
  64                 //reset calc for next command
  65                 globaldata->setCalc("");
  66
  67         }
  68         catch(exception& e) {
  69                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  70                 exit(1);
  71         }
  72         catch(...) {
  73                 cout << "An unknown error has occurred in the BootSharedCommand class function BootSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78
  79 BootSharedCommand::~BootSharedCommand(){
  80         delete input;
  81         delete read;
  82         delete util;
  83 }
  84
  85 //**********************************************************************************************************************
  86
  87 int BootSharedCommand::execute(){
  88         try {
  89                 int count = 1;
  90                 EstOutput data;
  91
  92                 //if the users entered no valid calculators don't execute command
  93                 if (treeCalculators.size() == 0) { return 0; }
  94
  95                 if (format == "sharedfile") {
  96                         read = new ReadOTUFile(globaldata->inputFileName);
  97                         read->read(&*globaldata);
  98
  99                         input = globaldata->ginput;
 100                         order = input->getSharedOrderVector();
 101                 }else {
 102                         //you are using a list and a groupfile
 103                         read = new ReadOTUFile(globaldata->inputFileName);
 104                         read->read(&*globaldata);
 105
 106                         input = globaldata->ginput;
 107                         SharedList = globaldata->gSharedList;
 108                         order = SharedList->getSharedOrderVector();
 109                 }
 110
 111                 //set users groups
 112                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 113                 numGroups = globaldata->Groups.size();
 114
 115                 //clear globaldatas old tree names if any
 116                 globaldata->Treenames.clear();
 117
 118                 //fills globaldatas tree names
 119                 globaldata->Treenames = globaldata->Groups;
 120
 121                 //create treemap class from groupmap for tree class to use
 122                 tmap = new TreeMap();
 123                 tmap->makeSim(globaldata->gGroupmap);
 124                 globaldata->gTreemap = tmap;
 125
 126                 while(order != NULL){
 127
 128                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 129
 130                                 cout << order->getLabel() << '\t' << count << endl;
 131
 132                                 //open an ostream for each calc to print to
 133                                 for (int z = 0; z < treeCalculators.size(); z++) {
 134                                         //create a new filename
 135                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 136                                         openOutputFile(outputFile, *(out[z]));
 137                                 }
 138
 139                                 //create a file for each calculator with the 1000 trees in it.
 140                                 for (int p = 0; p < iters; p++) {
 141
 142                                         util->getSharedVectorswithReplacement(globaldata->Groups, lookup, order);  //fills group vectors from order vector.
 143
 144                                         //for each calculator
 145                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 146
 147                                                 //initialize simMatrix
 148                                                 simMatrix.clear();
 149                                                 simMatrix.resize(numGroups);
 150                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 151                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 152                                                                 simMatrix[m].push_back(0.0);
 153                                                         }
 154                                                 }
 155
 156                                                 //initialize index
 157                                                 index.clear();
 158                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 159
 160                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 161                                                         for (int l = k; l < lookup.size(); l++) {
 162                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 163                                                                         //get estimated similarity between 2 groups
 164                                                                         data = treeCalculators[i]->getValues(lookup[k], lookup[l]); //saves the calculator outputs
 165                                                                         //save values in similarity matrix
 166                                                                         simMatrix[k][l] = data[0];
 167                                                                         simMatrix[l][k] = data[0];
 168                                                                 }
 169                                                         }
 170                                                 }
 171
 172                                                 //creates tree from similarity matrix and write out file
 173                                                 createTree(out[i]);
 174                                         }
 175                                 }
 176                                 //close ostream for each calc
 177                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 178                         }
 179
 180                         //get next line to process
 181                         if (format == "sharedfile") {
 182                                 order = input->getSharedOrderVector();
 183                         }else {
 184                                 //you are using a list and a groupfile
 185                                 SharedList = input->getSharedListVector(); //get new list vector to process
 186                                 if (SharedList != NULL) {
 187                                         order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
 188                                 }else {
 189                                         break;
 190                                 }
 191                         }
 192                         count++;
 193                 }
 194
 195                 //reset groups parameter
 196                 globaldata->Groups.clear();  globaldata->setGroups("");
 197
 198                 return 0;
 199         }
 200         catch(exception& e) {
 201                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 202                 exit(1);
 203         }
 204         catch(...) {
 205                 cout << "An unknown error has occurred in the BootSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 206                 exit(1);
 207         }
 208 }
 209 //**********************************************************************************************************************
 210
 211 void BootSharedCommand::createTree(ostream* out){
 212         try {
 213                 //create tree
 214                 t = new Tree();
 215
 216                 //do merges and create tree structure by setting parents and children
 217                 //there are numGroups - 1 merges to do
 218                 for (int i = 0; i < (numGroups - 1); i++) {
 219
 220                         float largest = 0.0;
 221                         int row, column;
 222                         //find largest value in sims matrix by searching lower triangle
 223                         for (int j = 1; j < simMatrix.size(); j++) {
 224                                 for (int k = 0; k < j; k++) {
 225                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 226                                 }
 227                         }
 228
 229                         //set non-leaf node info and update leaves to know their parents
 230                         //non-leaf
 231                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 232
 233                         //parents
 234                         t->tree[index[row]].setParent(numGroups + i);
 235                         t->tree[index[column]].setParent(numGroups + i);
 236
 237                         //blength = distance / 2;
 238                         float blength = ((1.0 - largest) / 2);
 239
 240                         //branchlengths
 241                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 242                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 243
 244                         //set your length to leaves to your childs length plus branchlength
 245                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 246
 247
 248                         //update index
 249                         index[row] = numGroups+i;
 250                         index[column] = numGroups+i;
 251
 252                         //zero out highest value that caused the merge.
 253                         simMatrix[row][column] = 0.0;
 254                         simMatrix[column][row] = 0.0;
 255
 256                         //merge values in simsMatrix
 257                         for (int n = 0; n < simMatrix.size(); n++)      {
 258                                 //row becomes merge of 2 groups
 259                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 260                                 simMatrix[n][row] = simMatrix[row][n];
 261                                 //delete column
 262                                 simMatrix[column][n] = 0.0;
 263                                 simMatrix[n][column] = 0.0;
 264                         }
 265                 }
 266
 267                 //assemble tree
 268                 t->assembleTree();
 269
 270                 //print newick file
 271                 t->print(*out);
 272
 273                 //delete tree
 274                 delete t;
 275
 276         }
 277         catch(exception& e) {
 278                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function createTree. 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 createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 283                 exit(1);
 284         }
 285 }
 286 /***********************************************************/
 287 void BootSharedCommand::printSims() {
 288         try {
 289                 cout << "simsMatrix" << endl;
 290                 for (int m = 0; m < simMatrix.size(); m++)      {
 291                         for (int n = 0; n < simMatrix.size(); n++)      {
 292                                 cout << simMatrix[m][n] << '\t';
 293                         }
 294                         cout << endl;
 295                 }
 296
 297         }
 298         catch(exception& e) {
 299                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 300                 exit(1);
 301         }
 302         catch(...) {
 303                 cout << "An unknown error has occurred in the BootSharedCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 304                 exit(1);
 305         }
 306 }
 307 /***********************************************************/
 308
 309