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