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                 set<int> userLines = globaldata->lines;
 108
 109                 //set users groups
 110                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "treegroup");
 111                 numGroups = globaldata->Groups.size();
 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) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 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                                 process(order);
 130
 131                                 processedLabels.insert(order->getLabel());
 132                                 userLabels.erase(order->getLabel());
 133                                 userLines.erase(count);
 134                         }
 135
 136                         //you have a label the user want that is smaller than this line and the last line has not already been processed
 137                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastOrder->getLabel()) != 1)) {
 138
 139                                 cout << lastOrder->getLabel() << '\t' << count << endl;
 140                                 process(lastOrder);
 141
 142                                 processedLabels.insert(lastOrder->getLabel());
 143                                 userLabels.erase(lastOrder->getLabel());
 144                         }
 145
 146                         if (count != 1) { delete lastOrder; }
 147                         lastOrder = order;
 148
 149                         //get next line to process
 150                         order = input->getSharedOrderVector();
 151                         count++;
 152                 }
 153
 154                 //output error messages about any remaining user labels
 155                 set<string>::iterator it;
 156                 bool needToRun = false;
 157                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 158                         cout << "Your file does not include the label "<< *it;
 159                         if (processedLabels.count(lastOrder->getLabel()) != 1) {
 160                                 cout << ". I will use " << lastOrder->getLabel() << "." << endl;
 161                                 needToRun = true;
 162                         }else {
 163                                 cout << ". Please refer to " << lastOrder->getLabel() << "." << endl;
 164                         }
 165                 }
 166
 167                 //run last line if you need to
 168                 if (needToRun == true)  {
 169                         process(lastOrder);
 170                         cout << lastOrder->getLabel() << '\t' << count << endl;
 171                 }
 172
 173                 delete lastOrder;
 174
 175                 //reset groups parameter
 176                 globaldata->Groups.clear();  globaldata->setGroups("");
 177
 178                 return 0;
 179         }
 180         catch(exception& e) {
 181                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 182                 exit(1);
 183         }
 184         catch(...) {
 185                 cout << "An unknown error has occurred in the BootSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 186                 exit(1);
 187         }
 188 }
 189 //**********************************************************************************************************************
 190
 191 void BootSharedCommand::createTree(ostream* out){
 192         try {
 193                 //create tree
 194                 t = new Tree();
 195
 196                 //do merges and create tree structure by setting parents and children
 197                 //there are numGroups - 1 merges to do
 198                 for (int i = 0; i < (numGroups - 1); i++) {
 199
 200                         float largest = -1.0;
 201                         int row, column;
 202                         //find largest value in sims matrix by searching lower triangle
 203                         for (int j = 1; j < simMatrix.size(); j++) {
 204                                 for (int k = 0; k < j; k++) {
 205                                         if (simMatrix[j][k] > largest) {  largest = simMatrix[j][k]; row = j; column = k;  }
 206                                 }
 207                         }
 208
 209                         //set non-leaf node info and update leaves to know their parents
 210                         //non-leaf
 211                         t->tree[numGroups + i].setChildren(index[row], index[column]);
 212
 213                         //parents
 214                         t->tree[index[row]].setParent(numGroups + i);
 215                         t->tree[index[column]].setParent(numGroups + i);
 216
 217                         //blength = distance / 2;
 218                         float blength = ((1.0 - largest) / 2);
 219
 220                         //branchlengths
 221                         t->tree[index[row]].setBranchLength(blength - t->tree[index[row]].getLengthToLeaves());
 222                         t->tree[index[column]].setBranchLength(blength - t->tree[index[column]].getLengthToLeaves());
 223
 224                         //set your length to leaves to your childs length plus branchlength
 225                         t->tree[numGroups + i].setLengthToLeaves(t->tree[index[row]].getLengthToLeaves() + t->tree[index[row]].getBranchLength());
 226
 227
 228                         //update index
 229                         index[row] = numGroups+i;
 230                         index[column] = numGroups+i;
 231
 232                         //zero out highest value that caused the merge.
 233                         simMatrix[row][column] = -1.0;
 234                         simMatrix[column][row] = -1.0;
 235
 236                         //merge values in simsMatrix
 237                         for (int n = 0; n < simMatrix.size(); n++)      {
 238                                 //row becomes merge of 2 groups
 239                                 simMatrix[row][n] = (simMatrix[row][n] + simMatrix[column][n]) / 2;
 240                                 simMatrix[n][row] = simMatrix[row][n];
 241                                 //delete column
 242                                 simMatrix[column][n] = -1.0;
 243                                 simMatrix[n][column] = -1.0;
 244                         }
 245                 }
 246
 247                 //assemble tree
 248                 t->assembleTree();
 249
 250                 //print newick file
 251                 t->print(*out);
 252
 253                 //delete tree
 254                 delete t;
 255
 256         }
 257         catch(exception& e) {
 258                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 259                 exit(1);
 260         }
 261         catch(...) {
 262                 cout << "An unknown error has occurred in the BootSharedCommand class function createTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 263                 exit(1);
 264         }
 265 }
 266 /***********************************************************/
 267 void BootSharedCommand::printSims() {
 268         try {
 269                 cout << "simsMatrix" << endl;
 270                 for (int m = 0; m < simMatrix.size(); m++)      {
 271                         for (int n = 0; n < simMatrix.size(); n++)      {
 272                                 cout << simMatrix[m][n] << '\t';
 273                         }
 274                         cout << endl;
 275                 }
 276
 277         }
 278         catch(exception& e) {
 279                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 280                 exit(1);
 281         }
 282         catch(...) {
 283                 cout << "An unknown error has occurred in the BootSharedCommand class function printSims. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 284                 exit(1);
 285         }
 286 }
 287 /***********************************************************/
 288 void BootSharedCommand::process(SharedOrderVector* order) {
 289         try{
 290                                 EstOutput data;
 291                                 vector<SharedRAbundVector*> subset;
 292
 293                                 //open an ostream for each calc to print to
 294                                 for (int z = 0; z < treeCalculators.size(); z++) {
 295                                         //create a new filename
 296                                         outputFile = getRootName(globaldata->inputFileName) + treeCalculators[z]->getName() + ".boot" + order->getLabel() + ".tre";
 297                                         openOutputFile(outputFile, *(out[z]));
 298                                 }
 299
 300                                 //create a file for each calculator with the 1000 trees in it.
 301                                 for (int p = 0; p < iters; p++) {
 302
 303                                         util->getSharedVectorswithReplacement(globaldata->Groups, lookup, order);  //fills group vectors from order vector.
 304
 305                                         //for each calculator
 306                                         for(int i = 0 ; i < treeCalculators.size(); i++) {
 307
 308                                                 //initialize simMatrix
 309                                                 simMatrix.clear();
 310                                                 simMatrix.resize(numGroups);
 311                                                 for (int m = 0; m < simMatrix.size(); m++)      {
 312                                                         for (int j = 0; j < simMatrix.size(); j++)      {
 313                                                                 simMatrix[m].push_back(0.0);
 314                                                         }
 315                                                 }
 316
 317                                                 //initialize index
 318                                                 index.clear();
 319                                                 for (int g = 0; g < numGroups; g++) {   index[g] = g;   }
 320
 321                                                 for (int k = 0; k < lookup.size(); k++) { // pass cdd each set of groups to commpare
 322                                                         for (int l = k; l < lookup.size(); l++) {
 323                                                                 if (k != l) { //we dont need to similiarity of a groups to itself
 324                                                                         subset.clear(); //clear out old pair of sharedrabunds
 325                                                                         //add new pair of sharedrabunds
 326                                                                         subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 327
 328                                                                         //get estimated similarity between 2 groups
 329                                                                         data = treeCalculators[i]->getValues(subset); //saves the calculator outputs
 330                                                                         //save values in similarity matrix
 331                                                                         simMatrix[k][l] = data[0];
 332                                                                         simMatrix[l][k] = data[0];
 333                                                                 }
 334                                                         }
 335                                                 }
 336
 337                                                 //creates tree from similarity matrix and write out file
 338                                                 createTree(out[i]);
 339                                         }
 340                                 }
 341                                 //close ostream for each calc
 342                                 for (int z = 0; z < treeCalculators.size(); z++) { out[z]->close(); }
 343
 344         }
 345         catch(exception& e) {
 346                 cout << "Standard Error: " << e.what() << " has occurred in the BootSharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 347                 exit(1);
 348         }
 349         catch(...) {
 350                 cout << "An unknown error has occurred in the BootSharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 351                 exit(1);
 352         }
 353 }
 354 /***********************************************************/
 355
 356
 357