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 "sharedochiai.h"
  24 #include "sharedanderberg.h"
  25 #include "sharedkulczynski.h"
  26 #include "sharedkulczynskicody.h"
  27 #include "sharedlennon.h"
  28 #include "sharedmorisitahorn.h"
  29 #include "sharedbraycurtis.h"
  30
  31
  32 //**********************************************************************************************************************
  33
  34 SummarySharedCommand::SummarySharedCommand(){
  35         try {
  36                 globaldata = GlobalData::getInstance();
  37                 outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
  38                 openOutputFile(outputFileName, outputFileHandle);
  39                 format = globaldata->getFormat();
  40                 validCalculator = new ValidCalculators();
  41
  42                 int i;
  43                 for (i=0; i<globaldata->Estimators.size(); i++) {
  44                         if (validCalculator->isValidCalculator("sharedsummary", globaldata->Estimators[i]) == true) {
  45                                 if (globaldata->Estimators[i] == "sharedsobs") {
  46                                         sumCalculators.push_back(new SharedSobsCS());
  47                                 }else if (globaldata->Estimators[i] == "sharedchao") {
  48                                         sumCalculators.push_back(new SharedChao1());
  49                                 }else if (globaldata->Estimators[i] == "sharedace") {
  50                                         sumCalculators.push_back(new SharedAce());
  51                                 }else if (globaldata->Estimators[i] == "sharedjabund") {
  52                                         sumCalculators.push_back(new SharedJAbund());
  53                                 }else if (globaldata->Estimators[i] == "sharedsorensonabund") {
  54                                         sumCalculators.push_back(new SharedSorAbund());
  55                                 }else if (globaldata->Estimators[i] == "sharedjclass") {
  56                                         sumCalculators.push_back(new SharedJclass());
  57                                 }else if (globaldata->Estimators[i] == "sharedsorclass") {
  58                                         sumCalculators.push_back(new SharedSorClass());
  59                                 }else if (globaldata->Estimators[i] == "sharedjest") {
  60                                         sumCalculators.push_back(new SharedJest());
  61                                 }else if (globaldata->Estimators[i] == "sharedsorest") {
  62                                         sumCalculators.push_back(new SharedSorEst());
  63                                 }else if (globaldata->Estimators[i] == "sharedthetayc") {
  64                                         sumCalculators.push_back(new SharedThetaYC());
  65                                 }else if (globaldata->Estimators[i] == "sharedthetan") {
  66                                         sumCalculators.push_back(new SharedThetaN());
  67                                 }else if (globaldata->Estimators[i] == "sharednseqs") {
  68                                         sumCalculators.push_back(new SharedNSeqs());
  69                                 }else if (globaldata->Estimators[i] == "sharedochiai") {
  70                                         sumCalculators.push_back(new SharedOchiai());
  71                                 }else if (globaldata->Estimators[i] == "sharedanderberg") {
  72                                         sumCalculators.push_back(new SharedAnderberg());
  73                                 }else if (globaldata->Estimators[i] == "sharedkulczynski") {
  74                                         sumCalculators.push_back(new SharedKulczynski());
  75                                 }else if (globaldata->Estimators[i] == "sharedkulczynskicody") {
  76                                         sumCalculators.push_back(new SharedKulczynskiCody());
  77                                 }else if (globaldata->Estimators[i] == "sharedlennon") {
  78                                         sumCalculators.push_back(new SharedLennon());
  79                                 }else if (globaldata->Estimators[i] == "sharedmorisitahorn") {
  80                                         sumCalculators.push_back(new SharedMorHorn());
  81                                 }else if (globaldata->Estimators[i] == "sharedbraycurtis") {
  82                                         sumCalculators.push_back(new SharedBrayCurtis());
  83                                 }
  84                         }
  85                 }
  86
  87                 //reset calc for next command
  88                 globaldata->setCalc("");
  89
  90         }
  91         catch(exception& e) {
  92                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  93                 exit(1);
  94         }
  95         catch(...) {
  96                 cout << "An unknown error has occurred in the SummarySharedCommand class function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  97                 exit(1);
  98         }
  99 }
 100 //**********************************************************************************************************************
 101
 102 SummarySharedCommand::~SummarySharedCommand(){
 103         delete input;
 104         delete read;
 105 }
 106
 107 //**********************************************************************************************************************
 108
 109 int SummarySharedCommand::execute(){
 110         try {
 111                 int count = 1;
 112
 113                 //if the users entered no valid calculators don't execute command
 114                 if (sumCalculators.size() == 0) { return 0; }
 115
 116                 if (format == "sharedfile") {
 117                         read = new ReadPhilFile(globaldata->inputFileName);
 118                         read->read(&*globaldata);
 119
 120                         input = globaldata->ginput;
 121                         order = input->getSharedOrderVector();
 122                 }else {
 123                         //you are using a list and a groupfile
 124                         read = new ReadPhilFile(globaldata->inputFileName);
 125                         read->read(&*globaldata);
 126
 127                         input = globaldata->ginput;
 128                         SharedList = globaldata->gSharedList;
 129                         order = SharedList->getSharedOrderVector();
 130                 }
 131
 132                 //set users groups
 133                 setGroups();
 134
 135                 //output estimator names as column headers
 136                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 137                 for(int i=0;i<sumCalculators.size();i++){
 138                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 139                 }
 140                 outputFileHandle << endl;
 141
 142                 while(order != NULL){
 143
 144                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 145
 146                                 cout << order->getLabel() << '\t' << count << endl;
 147                                 getSharedVectors();  //fills group vectors from order vector.
 148
 149                                 //randomize group order
 150                                 if (globaldata->getJumble() == "1") { random_shuffle(lookup.begin(), lookup.end()); }
 151
 152                                 int n = 1;
 153                                 for (int k = 0; k < (lookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 154                                         for (int l = n; l < lookup.size(); l++) {
 155                                                 outputFileHandle << order->getLabel() << '\t';
 156
 157                                                 //sort groups to be alphanumeric
 158                                                 if (lookup[k]->getGroup() > lookup[l]->getGroup()) {
 159                                                         outputFileHandle << (lookup[l]->getGroup() +'\t' + lookup[k]->getGroup()) << '\t'; //print out groups
 160                                                 }else{
 161                                                         outputFileHandle << (lookup[k]->getGroup() +'\t' + lookup[l]->getGroup()) << '\t'; //print out groups
 162                                                 }
 163
 164                                                 for(int i=0;i<sumCalculators.size();i++){
 165                                                         sumCalculators[i]->getValues(lookup[k], lookup[l]); //saves the calculator outputs
 166                                                         outputFileHandle << '\t';
 167                                                         sumCalculators[i]->print(outputFileHandle);
 168                                                 }
 169                                                 outputFileHandle << endl;
 170                                         }
 171                                         n++;
 172                                 }
 173                         }
 174
 175                         //get next line to process
 176                         if (format == "sharedfile") {
 177                                 order = input->getSharedOrderVector();
 178                         }else {
 179                                 //you are using a list and a groupfile
 180                                 SharedList = input->getSharedListVector(); //get new list vector to process
 181                                 if (SharedList != NULL) {
 182                                         order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
 183                                 }else {
 184                                         break;
 185                                 }
 186                         }
 187                         count++;
 188                 }
 189
 190                 //reset groups parameter
 191                 globaldata->Groups.clear();  globaldata->setGroups("");
 192
 193                 return 0;
 194         }
 195         catch(exception& e) {
 196                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 197                 exit(1);
 198         }
 199         catch(...) {
 200                 cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 201                 exit(1);
 202         }
 203 }
 204
 205 //**********************************************************************************************************************
 206
 207 void SummarySharedCommand::getSharedVectors(){
 208 try {
 209                 lookup.clear();
 210                 //create and initialize vector of sharedvectors, one for each group
 211                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 212                         SharedRAbundVector* temp = new SharedRAbundVector(order->getNumBins());
 213                         temp->setLabel(order->getLabel());
 214                         temp->setGroup(globaldata->Groups[i]);
 215                         lookup.push_back(temp);
 216                 }
 217
 218                 int numSeqs = order->size();
 219                 //sample all the members
 220                 for(int i=0;i<numSeqs;i++){
 221                         //get first sample
 222                         individual chosen = order->get(i);
 223                         int abundance;
 224
 225                         //set info for sharedvector in chosens group
 226                         for (int j = 0; j < lookup.size(); j++) {
 227                                 if (chosen.group == lookup[j]->getGroup()) {
 228                                          abundance = lookup[j]->getAbundance(chosen.bin);
 229                                          lookup[j]->set(chosen.bin, (abundance + 1), chosen.group);
 230                                          break;
 231                                 }
 232                         }
 233                 }
 234         }
 235         catch(exception& e) {
 236                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 237                 exit(1);
 238         }
 239         catch(...) {
 240                 cout << "An unknown error has occurred in the SummarySharedCommand class function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 241                 exit(1);
 242         }
 243
 244 }
 245
 246 //**********************************************************************************************************************
 247 void SummarySharedCommand::setGroups() {
 248         try {
 249                 //if the user has not entered specific groups to analyze then do them all
 250                 if (globaldata->Groups.size() != 0) {
 251                         if (globaldata->Groups[0] != "all") {
 252                                 //check that groups are valid
 253                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 254                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 255                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 256                                                 // erase the invalid group from globaldata->Groups
 257                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 258                                         }
 259                                 }
 260
 261                                 //if the user only entered invalid groups
 262                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 263                                         cout << "When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile." << endl;
 264                                         for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 265                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 266                                         }
 267                                 }
 268                         }else{//user has enter "all" and wants the default groups
 269                                 globaldata->Groups.clear();
 270                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 271                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 272                                 }
 273                                 globaldata->setGroups("");
 274                         }
 275                 }else {
 276                         for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 277                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 278                         }
 279                 }
 280
 281         }
 282         catch(exception& e) {
 283                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 284                 exit(1);
 285         }
 286         catch(...) {
 287                 cout << "An unknown error has occurred in the SummarySharedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 288                 exit(1);
 289         }
 290
 291 }
 292 /***********************************************************/