summarysharedcommand.cpp

   1 /*
   2  *  summarysharedcommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/2/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "summarysharedcommand.h"
  11 #include "sharedsobscollectsummary.h"
  12 #include "sharedchao1.h"
  13 #include "sharedace.h"
  14 #include "sharednseqs.h"
  15 #include "sharedjabund.h"
  16 #include "sharedsorabund.h"
  17 #include "sharedjclass.h"
  18 #include "sharedsorclass.h"
  19 #include "sharedjest.h"
  20 #include "sharedsorest.h"
  21 #include "sharedthetayc.h"
  22 #include "sharedthetan.h"
  23 #include "sharedkstest.h"
  24 #include "whittaker.h"
  25 #include "sharedochiai.h"
  26 #include "sharedanderbergs.h"
  27 #include "sharedkulczynski.h"
  28 #include "sharedkulczynskicody.h"
  29 #include "sharedlennon.h"
  30 #include "sharedmorisitahorn.h"
  31 #include "sharedbraycurtis.h"
  32 #include "sharedjackknife.h"
  33 #include "whittaker.h"
  34
  35
  36 //**********************************************************************************************************************
  37
  38 SummarySharedCommand::SummarySharedCommand(){
  39         try {
  40                 globaldata = GlobalData::getInstance();
  41                 outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
  42                 openOutputFile(outputFileName, outputFileHandle);
  43                 format = globaldata->getFormat();
  44                 validCalculator = new ValidCalculators();
  45                 mult = false;
  46
  47                 int i;
  48                 for (i=0; i<globaldata->Estimators.size(); i++) {
  49                         if (validCalculator->isValidCalculator("sharedsummary", globaldata->Estimators[i]) == true) {
  50                                 if (globaldata->Estimators[i] == "sharedsobs") {
  51                                         sumCalculators.push_back(new SharedSobsCS());
  52                                 }else if (globaldata->Estimators[i] == "sharedchao") {
  53                                         sumCalculators.push_back(new SharedChao1());
  54                                 }else if (globaldata->Estimators[i] == "sharedace") {
  55                                         sumCalculators.push_back(new SharedAce());
  56                                 }else if (globaldata->Estimators[i] == "jabund") {
  57                                         sumCalculators.push_back(new JAbund());
  58                                 }else if (globaldata->Estimators[i] == "sorabund") {
  59                                         sumCalculators.push_back(new SorAbund());
  60                                 }else if (globaldata->Estimators[i] == "jclass") {
  61                                         sumCalculators.push_back(new Jclass());
  62                                 }else if (globaldata->Estimators[i] == "sorclass") {
  63                                         sumCalculators.push_back(new SorClass());
  64                                 }else if (globaldata->Estimators[i] == "jest") {
  65                                         sumCalculators.push_back(new Jest());
  66                                 }else if (globaldata->Estimators[i] == "sorest") {
  67                                         sumCalculators.push_back(new SorEst());
  68                                 }else if (globaldata->Estimators[i] == "thetayc") {
  69                                         sumCalculators.push_back(new ThetaYC());
  70                                 }else if (globaldata->Estimators[i] == "thetan") {
  71                                         sumCalculators.push_back(new ThetaN());
  72                                 }else if (globaldata->Estimators[i] == "kstest") {
  73                                         sumCalculators.push_back(new KSTest());
  74                                 }else if (globaldata->Estimators[i] == "sharednseqs") {
  75                                         sumCalculators.push_back(new SharedNSeqs());
  76                                 }else if (globaldata->Estimators[i] == "ochiai") {
  77                                         sumCalculators.push_back(new Ochiai());
  78                                 }else if (globaldata->Estimators[i] == "anderberg") {
  79                                         sumCalculators.push_back(new Anderberg());
  80                                 }else if (globaldata->Estimators[i] == "kulczynski") {
  81                                         sumCalculators.push_back(new Kulczynski());
  82                                 }else if (globaldata->Estimators[i] == "kulczynskicody") {
  83                                         sumCalculators.push_back(new KulczynskiCody());
  84                                 }else if (globaldata->Estimators[i] == "lennon") {
  85                                         sumCalculators.push_back(new Lennon());
  86                                 }else if (globaldata->Estimators[i] == "morisitahorn") {
  87                                         sumCalculators.push_back(new MorHorn());
  88                                 }else if (globaldata->Estimators[i] == "braycurtis") {
  89                                         sumCalculators.push_back(new BrayCurtis());
  90                                 }else if (globaldata->Estimators[i] == "whittaker") {
  91                                         sumCalculators.push_back(new Whittaker());
  92                                 }
  93                         }
  94                 }
  95                 //reset calc for next command
  96                 globaldata->setCalc("");
  97
  98         }
  99         catch(exception& e) {
 100                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 101                 exit(1);
 102         }
 103         catch(...) {
 104                 cout << "An unknown error has occurred in the SummarySharedCommand class function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 105                 exit(1);
 106         }
 107 }
 108 //**********************************************************************************************************************
 109
 110 SummarySharedCommand::~SummarySharedCommand(){
 111         delete input;
 112         delete read;
 113 }
 114
 115 //**********************************************************************************************************************
 116
 117 int SummarySharedCommand::execute(){
 118         try {
 119                 int count = 1;
 120
 121                 //if the users entered no valid calculators don't execute command
 122                 if (sumCalculators.size() == 0) { return 0; }
 123                 //check if any calcs can do multiples
 124                 else{
 125                         for (int i = 0; i < sumCalculators.size(); i++) {
 126                                 if (sumCalculators[i]->getMultiple() == true) { mult = true; }
 127                         }
 128                 }
 129
 130                 //read first line
 131                 read = new ReadOTUFile(globaldata->inputFileName);
 132                 read->read(&*globaldata);
 133
 134                 input = globaldata->ginput;
 135                 lookup = input->getSharedRAbundVectors();
 136
 137                 //output estimator names as column headers
 138                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 139                 for(int i=0;i<sumCalculators.size();i++){
 140                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 141                 }
 142                 outputFileHandle << endl;
 143
 144                 //create file and put column headers for multiple groups file
 145                 if (mult == true) {
 146                         outAllFileName = ((getRootName(globaldata->inputFileName)) + "sharedmultiple.summary");
 147                         openOutputFile(outAllFileName, outAll);
 148
 149                         outAll << "label" <<'\t' << "comparison" << '\t';
 150                         for(int i=0;i<sumCalculators.size();i++){
 151                                 if (sumCalculators[i]->getMultiple() == true) {
 152                                         outAll << '\t' << sumCalculators[i]->getName();
 153                                 }
 154                         }
 155                         outAll << endl;
 156                 }
 157
 158                 if (lookup.size() < 2) {
 159                         cout << "I cannot run the command without at least 2 valid groups.";
 160                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 161
 162                         //close files and clean up
 163                         outputFileHandle.close();  remove(outputFileName.c_str());
 164                         if (mult == true) {  outAll.close();  remove(outAllFileName.c_str());  }
 165                         return 0;
 166                 }
 167
 168
 169                 while(lookup[0] != NULL){
 170
 171                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(lookup[0]->getLabel()) == 1){
 172
 173                                 cout << lookup[0]->getLabel() << '\t' << count << endl;
 174
 175                                 //loop through calculators and add to file all for all calcs that can do mutiple groups
 176                                 if (mult == true) {
 177                                         //output label
 178                                         outAll << lookup[0]->getLabel() << '\t';
 179
 180                                         //output groups names
 181                                         string outNames = "";
 182                                         for (int j = 0; j < lookup.size(); j++) {
 183                                                 outNames += lookup[j]->getGroup() +  "-";
 184                                         }
 185                                         outNames = outNames.substr(0, outNames.length()-1); //rip off extra '-';
 186                                         outAll << outNames << '\t';
 187
 188                                         for(int i=0;i<sumCalculators.size();i++){
 189                                                 if (sumCalculators[i]->getMultiple() == true) {
 190                                                         sumCalculators[i]->getValues(lookup);
 191                                                         outAll << '\t';
 192                                                         sumCalculators[i]->print(outAll);
 193                                                 }
 194                                         }
 195                                         outAll << endl;
 196                                 }
 197
 198                                 int n = 1;
 199                                 vector<SharedRAbundVector*> subset;
 200                                 for (int k = 0; k < (lookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 201                                         for (int l = n; l < lookup.size(); l++) {
 202
 203                                                 outputFileHandle << lookup[0]->getLabel() << '\t';
 204
 205                                                 subset.clear(); //clear out old pair of sharedrabunds
 206                                                 //add new pair of sharedrabunds
 207                                                 subset.push_back(lookup[k]); subset.push_back(lookup[l]);
 208
 209                                                 //sort groups to be alphanumeric
 210                                                 if (lookup[k]->getGroup() > lookup[l]->getGroup()) {
 211                                                         outputFileHandle << (lookup[l]->getGroup() +'\t' + lookup[k]->getGroup()) << '\t'; //print out groups
 212                                                 }else{
 213                                                         outputFileHandle << (lookup[k]->getGroup() +'\t' + lookup[l]->getGroup()) << '\t'; //print out groups
 214                                                 }
 215
 216                                                 for(int i=0;i<sumCalculators.size();i++) {
 217
 218                                                         sumCalculators[i]->getValues(subset); //saves the calculator outputs
 219                                                         outputFileHandle << '\t';
 220                                                         sumCalculators[i]->print(outputFileHandle);
 221                                                 }
 222                                                 outputFileHandle << endl;
 223                                         }
 224                                         n++;
 225                                 }
 226                         }
 227
 228                         //prevent memory leak
 229                         for (int i = 0; i < lookup.size(); i++) {       delete lookup[i];       }
 230
 231                         //get next line to process
 232                         lookup = input->getSharedRAbundVectors();
 233                         count++;
 234                 }
 235
 236                 //reset groups parameter
 237                 globaldata->Groups.clear();  globaldata->setGroups("");
 238
 239                 //close files
 240                 outputFileHandle.close();
 241                 if (mult == true) {  outAll.close();  }
 242
 243                 return 0;
 244         }
 245         catch(exception& e) {
 246                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 247                 exit(1);
 248         }
 249         catch(...) {
 250                 cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 251                 exit(1);
 252         }
 253 }
 254
 255 /***********************************************************/