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                 vector<SharedRAbundVector*> lastLookup = lookup;
 137
 138                 //output estimator names as column headers
 139                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 140                 for(int i=0;i<sumCalculators.size();i++){
 141                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 142                 }
 143                 outputFileHandle << endl;
 144
 145                 //create file and put column headers for multiple groups file
 146                 if (mult == true) {
 147                         outAllFileName = ((getRootName(globaldata->inputFileName)) + "sharedmultiple.summary");
 148                         openOutputFile(outAllFileName, outAll);
 149
 150                         outAll << "label" <<'\t' << "comparison" << '\t';
 151                         for(int i=0;i<sumCalculators.size();i++){
 152                                 if (sumCalculators[i]->getMultiple() == true) {
 153                                         outAll << '\t' << sumCalculators[i]->getName();
 154                                 }
 155                         }
 156                         outAll << endl;
 157                 }
 158
 159                 if (lookup.size() < 2) {
 160                         cout << "I cannot run the command without at least 2 valid groups.";
 161                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 162
 163                         //close files and clean up
 164                         outputFileHandle.close();  remove(outputFileName.c_str());
 165                         if (mult == true) {  outAll.close();  remove(outAllFileName.c_str());  }
 166                         return 0;
 167                 }
 168
 169                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 170                 set<string> processedLabels;
 171                 set<string> userLabels = globaldata->labels;
 172                 set<int> userLines = globaldata->lines;
 173
 174                 //as long as you are not at the end of the file or done wih the lines you want
 175                 while((lookup[0] != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 176
 177                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(lookup[0]->getLabel()) == 1){
 178                                 cout << lookup[0]->getLabel() << '\t' << count << endl;
 179                                 process(lookup);
 180
 181                                 processedLabels.insert(lookup[0]->getLabel());
 182                                 userLabels.erase(lookup[0]->getLabel());
 183                                 userLines.erase(count);
 184                         }
 185
 186                         if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLookup[0]->getLabel()) != 1)) {
 187                                         cout << lastLookup[0]->getLabel() << '\t' << count << endl;
 188                                         process(lastLookup);
 189
 190                                         processedLabels.insert(lastLookup[0]->getLabel());
 191                                         userLabels.erase(lastLookup[0]->getLabel());
 192                         }
 193
 194
 195                         //prevent memory leak
 196                         if (count != 1) { for (int i = 0; i < lastLookup.size(); i++) {  delete lastLookup[i];  } }
 197                         lastLookup = lookup;
 198
 199                         //get next line to process
 200                         lookup = input->getSharedRAbundVectors();
 201                         count++;
 202                 }
 203
 204                 //output error messages about any remaining user labels
 205                 set<string>::iterator it;
 206                 bool needToRun = false;
 207                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 208                         cout << "Your file does not include the label "<< *it;
 209                         if (processedLabels.count(lastLookup[0]->getLabel()) != 1) {
 210                                 cout << ". I will use " << lastLookup[0]->getLabel() << "." << endl;
 211                                 needToRun = true;
 212                         }else {
 213                                 cout << ". Please refer to " << lastLookup[0]->getLabel() << "." << endl;
 214                         }
 215                 }
 216
 217                 //run last line if you need to
 218                 if (needToRun == true)  {
 219                         cout << lastLookup[0]->getLabel() << '\t' << count << endl;
 220                         process(lastLookup);
 221                 }
 222
 223                 for (int i = 0; i < lastLookup.size(); i++) {  delete lastLookup[i];  }
 224
 225                 //reset groups parameter
 226                 globaldata->Groups.clear();  globaldata->setGroups("");
 227
 228                 //close files
 229                 outputFileHandle.close();
 230                 if (mult == true) {  outAll.close();  }
 231
 232                 return 0;
 233         }
 234         catch(exception& e) {
 235                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 236                 exit(1);
 237         }
 238         catch(...) {
 239                 cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 240                 exit(1);
 241         }
 242 }
 243
 244 /***********************************************************/
 245 void SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup) {
 246         try {
 247                                 //loop through calculators and add to file all for all calcs that can do mutiple groups
 248                                 if (mult == true) {
 249                                         //output label
 250                                         outAll << thisLookup[0]->getLabel() << '\t';
 251
 252                                         //output groups names
 253                                         string outNames = "";
 254                                         for (int j = 0; j < thisLookup.size(); j++) {
 255                                                 outNames += thisLookup[j]->getGroup() +  "-";
 256                                         }
 257                                         outNames = outNames.substr(0, outNames.length()-1); //rip off extra '-';
 258                                         outAll << outNames << '\t';
 259
 260                                         for(int i=0;i<sumCalculators.size();i++){
 261                                                 if (sumCalculators[i]->getMultiple() == true) {
 262                                                         sumCalculators[i]->getValues(thisLookup);
 263                                                         outAll << '\t';
 264                                                         sumCalculators[i]->print(outAll);
 265                                                 }
 266                                         }
 267                                         outAll << endl;
 268                                 }
 269
 270                                 int n = 1;
 271                                 vector<SharedRAbundVector*> subset;
 272                                 for (int k = 0; k < (thisLookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 273                                         for (int l = n; l < thisLookup.size(); l++) {
 274
 275                                                 outputFileHandle << thisLookup[0]->getLabel() << '\t';
 276
 277                                                 subset.clear(); //clear out old pair of sharedrabunds
 278                                                 //add new pair of sharedrabunds
 279                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 280
 281                                                 //sort groups to be alphanumeric
 282                                                 if (thisLookup[k]->getGroup() > thisLookup[l]->getGroup()) {
 283                                                         outputFileHandle << (thisLookup[l]->getGroup() +'\t' + thisLookup[k]->getGroup()) << '\t'; //print out groups
 284                                                 }else{
 285                                                         outputFileHandle << (thisLookup[k]->getGroup() +'\t' + thisLookup[l]->getGroup()) << '\t'; //print out groups
 286                                                 }
 287
 288                                                 for(int i=0;i<sumCalculators.size();i++) {
 289
 290                                                         sumCalculators[i]->getValues(subset); //saves the calculator outputs
 291                                                         outputFileHandle << '\t';
 292                                                         sumCalculators[i]->print(outputFileHandle);
 293                                                 }
 294                                                 outputFileHandle << endl;
 295                                         }
 296                                         n++;
 297                                 }
 298
 299         }
 300         catch(exception& e) {
 301                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 302                 exit(1);
 303         }
 304         catch(...) {
 305                 cout << "An unknown error has occurred in the SummarySharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 306                 exit(1);
 307         }
 308 }
 309
 310 /***********************************************************/