treegroupscommand.cpp

   1 /*
   2  *  treegroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "treegroupscommand.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 TreeGroupCommand::TreeGroupCommand(){
  26         try {
  27                 globaldata = GlobalData::getInstance();
  28                 format = globaldata->getFormat();
  29                 validCalculator = new ValidCalculators();
  30
  31                 int i;
  32                 for (i=0; i<globaldata->Estimators.size(); i++) {
  33                         if (validCalculator->isValidCalculator("treegroup", globaldata->Estimators[i]) == true) {
  34                                 if (globaldata->Estimators[i] == "jabund") {
  35                                         treeCalculators.push_back(new JAbund());
  36                                 }else if (globaldata->Estimators[i] == "sorabund") {
  37                                         treeCalculators.push_back(new SorAbund());
  38                                 }else if (globaldata->Estimators[i] == "jclass") {
  39                                         treeCalculators.push_back(new Jclass());
  40                                 }else if (globaldata->Estimators[i] == "sorclass") {
  41                                         treeCalculators.push_back(new SorClass());
  42                                 }else if (globaldata->Estimators[i] == "jest") {
  43                                         treeCalculators.push_back(new Jest());
  44                                 }else if (globaldata->Estimators[i] == "sorest") {
  45                                         treeCalculators.push_back(new SorEst());
  46                                 }else if (globaldata->Estimators[i] == "thetayc") {
  47                                         treeCalculators.push_back(new ThetaYC());
  48                                 }else if (globaldata->Estimators[i] == "thetan") {
  49                                         treeCalculators.push_back(new ThetaN());
  50                                 }else if (globaldata->Estimators[i] == "morisitahorn") {
  51                                         treeCalculators.push_back(new MorHorn());
  52                                 }else if (globaldata->Estimators[i] == "braycurtis") {
  53                                         treeCalculators.push_back(new BrayCurtis());
  54                                 }
  55                         }
  56                 }
  57
  58                 //reset calc for next command
  59                 globaldata->setCalc("");
  60
  61         }
  62         catch(exception& e) {
  63                 cout << "Standard Error: " << e.what() << " has occurred in the TreeGroupCommand class Function TreeGroupCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  64                 exit(1);
  65         }
  66         catch(...) {
  67                 cout << "An unknown error has occurred in the TreeGroupCommand class function TreeGroupCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  68                 exit(1);
  69         }
  70 }
  71 //**********************************************************************************************************************
  72
  73 TreeGroupCommand::~TreeGroupCommand(){
  74         delete input;
  75         delete read;
  76 }
  77
  78 //**********************************************************************************************************************
  79
  80 int TreeGroupCommand::execute(){
  81         try {
  82                 int count = 1;
  83                 EstOutput data;
  84                 vector<SharedRAbundVector*> subset;
  85
  86                 //if the users entered no valid calculators don't execute command
  87                 if (treeCalculators.size() == 0) { return 0; }
  88
  89                 //you have groups
  90                 read = new ReadOTUFile(globaldata->inputFileName);
  91                 read->read(&*globaldata);
  92
  93                 input = globaldata->ginput;
  94                 lookup = input->getSharedRAbundVectors();
  95
  96                 if (lookup.size() < 2) { cout << "You have not provided enough valid groups.  I cannot run the command." << endl; return 0; }
  97
  98                 numGroups = globaldata->Groups.size();
  99                 groupNames = "";
 100                 for (int i = 0; i < numGroups; i++) { groupNames += globaldata->Groups[i]; }
 101
 102                 //clear globaldatas old tree names if any
 103                 globaldata->Treenames.clear();
 104
 105                 //fills globaldatas tree names
 106                 globaldata->Treenames = globaldata->Groups;
 107
 108                 //create treemap class from groupmap for tree class to use
 109                 tmap = new TreeMap();
 110                 tmap->makeSim(globaldata->gGroupmap);
 111                 globaldata->gTreemap = tmap;
 112
 113                 while(lookup[0] != NULL){
 114
 115                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(lookup[0]->getLabel()) == 1){
 116
 117                                 cout << lookup[0]->getLabel() << '\t' << count << endl;
 118
 119                                 //for each calculator
 120                                 for(int i = 0 ; i < treeCalculators.size(); i++) {
 121
 122                                         //initialize simMatrix
 123                                         simMatrix.clear();
 124                                         simMatrix.resize(numGroups);
 125                                         for (int m = 0; m < simMatrix.size(); m++)      {
 126                                                 for (int j = 0; j < simMatrix.size(); j++)      {
 127                                                         simMatrix[m].push_back(0.0);
 128                                                 }
 129                                         }
 130
 131                                         //initialize index
 132                                         index.clear();
 133                                         for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 134
 135                                         //create a new filename
 136                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[i]->getName() + "." + lookup[0]->getLabel() + ".tre";
 137
 138
 139                                         for (int k = 0; k < lookup.size(); k++) {
 140                                                 for (int l = k; l < lookup.size(); l++) {
 141                                                         if (k != l) { //we dont need to similiarity of a groups to itself
 142                                                                 //get estimated similarity between 2 groups
 143
 144                                                                 subset.clear(); //clear out old pair of sharedrabunds
 145                                                                 //add new pair of sharedrabunds
 146                                                                 subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 147
 148                                                                 data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 149                                                                 //save values in similarity matrix
 150                                                                 simMatrix[k][l] = data[0];
 151                                                                 simMatrix[l][k] = data[0];
 152                                                         }
 153                                                 }
 154                                         }
 155
 156                                         //creates tree from similarity matrix and write out file
 157                                         createTree();
 158                                 }
 159                         }
 160
 161                         //prevent memory leak
 162                         for (int i = 0; i < lookup.size(); i++) {       delete lookup[i];       }
 163
 164                         //get next line to process
 165                         lookup = input->getSharedRAbundVectors();
 166                         count++;
 167                 }
 168
 169                 //reset groups parameter
 170                 globaldata->Groups.clear();  globaldata->setGroups("");
 171
 172                 return 0;
 173         }
 174         catch(exception& e) {
 175                 cout << "Standard Error: " << e.what() << " has occurred in the TreeGroupCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 176                 exit(1);
 177         }
 178         catch(...) {
 179                 cout << "An unknown error has occurred in the TreeGroupCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 180                 exit(1);
 181         }
 182 }
 183 //**********************************************************************************************************************
 184
 185 void TreeGroupCommand::createTree(){
 186         try {
 187                 //create tree
 188                 t = new Tree();
 189
 190                 //do merges and create tree structure by setting parents and children
 191                 //there are numGroups - 1 merges to do
 192                 for (int i = 0; i < (numGroups - 1); i++) {
 193
 194                         float largest = 0.0;
 195                         int row, column;
 196                         //find largest value in sims matrix by searching lower triangle
 197                         for (int j = 1; j < simMatrix.size(); j++) {
 198                                 for (int k = 0; k < j; k++) {
 199                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 200                                 }
 201                         }
 202
 203                         //set non-leaf node info and update leaves to know their parents
 204                         //non-leaf
 205                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 206
 207                         //parents
 208                         t->tree[index[row]].setParent(numGroups + i);
 209                         t->tree[index[column]].setParent(numGroups + i);
 210
 211                         //blength = distance / 2;
 212                         float blength = ((1.0 - largest) / 2);
 213
 214                         //branchlengths
 215                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 216                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 217
 218                         //set your length to leaves to your childs length plus branchlength
 219                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 220
 221
 222                         //update index
 223                         index[row] = numGroups+i;
 224                         index[column] = numGroups+i;
 225
 226                         //zero out highest value that caused the merge.
 227                         simMatrix[row][column] = 0.0;
 228                         simMatrix[column][row] = 0.0;
 229
 230                         //merge values in simsMatrix
 231                         for (int n = 0; n < simMatrix.size(); n++)      {
 232                                 //row becomes merge of 2 groups
 233                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 234                                 simMatrix[n][row] = simMatrix[row][n];
 235                                 //delete column
 236                                 simMatrix[column][n] = 0.0;
 237                                 simMatrix[n][column] = 0.0;
 238                         }
 239                 }
 240
 241                 //assemble tree
 242                 t->assembleTree();
 243
 244                 //print newick file
 245                 t->createNewickFile(outputFile);
 246
 247                 //delete tree
 248                 delete t;
 249
 250         }
 251         catch(exception& e) {
 252                 cout << "Standard Error: " << e.what() << " has occurred in the TreeGroupCommand class Function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 253                 exit(1);
 254         }
 255         catch(...) {
 256                 cout << "An unknown error has occurred in the TreeGroupCommand class function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 257                 exit(1);
 258         }
 259 }
 260 /***********************************************************/
 261 void TreeGroupCommand::printSims(ostream& out) {
 262         try {
 263
 264                 //output column headers
 265                 out << '\t';
 266                 for (int i = 0; i < lookup.size(); i++) {       out << lookup[i]->getGroup() << '\t';           }
 267                 out << endl;
 268
 269
 270                 for (int m = 0; m < simMatrix.size(); m++)      {
 271                         out << lookup[m]->getGroup() << '\t';
 272                         for (int n = 0; n < simMatrix.size(); n++)      {
 273                                 out << simMatrix[m][n] << '\t';
 274                         }
 275                         out << endl;
 276                 }
 277
 278         }
 279         catch(exception& e) {
 280                 cout << "Standard Error: " << e.what() << " has occurred in the TreeGroupCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 281                 exit(1);
 282         }
 283         catch(...) {
 284                 cout << "An unknown error has occurred in the TreeGroupCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 285                 exit(1);
 286         }
 287 }
 288 /***********************************************************/
 289
 290