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 "sharedbdiversity.h"
  25 #include "sharedochiai.h"
  26 #include "sharedanderberg.h"
  27 #include "sharedkulczynski.h"
  28 #include "sharedkulczynskicody.h"
  29 #include "sharedlennon.h"
  30 #include "sharedmorisitahorn.h"
  31 #include "sharedbraycurtis.h"
  32
  33
  34 //**********************************************************************************************************************
  35
  36 SummarySharedCommand::SummarySharedCommand(){
  37         try {
  38                 globaldata = GlobalData::getInstance();
  39                 outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
  40                 openOutputFile(outputFileName, outputFileHandle);
  41                 format = globaldata->getFormat();
  42                 validCalculator = new ValidCalculators();
  43
  44                 int i;
  45                 for (i=0; i<globaldata->Estimators.size(); i++) {
  46                         if (validCalculator->isValidCalculator("sharedsummary", globaldata->Estimators[i]) == true) {
  47                                 if (globaldata->Estimators[i] == "sharedsobs") {
  48                                         sumCalculators.push_back(new SharedSobsCS());
  49                                 }else if (globaldata->Estimators[i] == "sharedchao") {
  50                                         sumCalculators.push_back(new SharedChao1());
  51                                 }else if (globaldata->Estimators[i] == "sharedace") {
  52                                         sumCalculators.push_back(new SharedAce());
  53                                 }else if (globaldata->Estimators[i] == "sharedjabund") {
  54                                         sumCalculators.push_back(new SharedJAbund());
  55                                 }else if (globaldata->Estimators[i] == "sharedsorensonabund") {
  56                                         sumCalculators.push_back(new SharedSorAbund());
  57                                 }else if (globaldata->Estimators[i] == "sharedjclass") {
  58                                         sumCalculators.push_back(new SharedJclass());
  59                                 }else if (globaldata->Estimators[i] == "sharedsorclass") {
  60                                         sumCalculators.push_back(new SharedSorClass());
  61                                 }else if (globaldata->Estimators[i] == "sharedjest") {
  62                                         sumCalculators.push_back(new SharedJest());
  63                                 }else if (globaldata->Estimators[i] == "sharedsorest") {
  64                                         sumCalculators.push_back(new SharedSorEst());
  65                                 }else if (globaldata->Estimators[i] == "sharedthetayc") {
  66                                         sumCalculators.push_back(new SharedThetaYC());
  67                                 }else if (globaldata->Estimators[i] == "sharedthetan") {
  68                                         sumCalculators.push_back(new SharedThetaN());
  69                                 }else if (globaldata->Estimators[i] == "sharedkstest") {
  70                                         sumCalculators.push_back(new SharedKSTest());
  71                                 }else if (globaldata->Estimators[i] == "sharednseqs") {
  72                                         sumCalculators.push_back(new SharedNSeqs());
  73                                 }else if (globaldata->Estimators[i] == "sharedochiai") {
  74                                         sumCalculators.push_back(new SharedOchiai());
  75                                 }else if (globaldata->Estimators[i] == "sharedanderberg") {
  76                                         sumCalculators.push_back(new SharedAnderberg());
  77                                 }else if (globaldata->Estimators[i] == "sharedkulczynski") {
  78                                         sumCalculators.push_back(new SharedKulczynski());
  79                                 }else if (globaldata->Estimators[i] == "sharedkulczynskicody") {
  80                                         sumCalculators.push_back(new SharedKulczynskiCody());
  81                                 }else if (globaldata->Estimators[i] == "sharedlennon") {
  82                                         sumCalculators.push_back(new SharedLennon());
  83                                 }else if (globaldata->Estimators[i] == "sharedmorisitahorn") {
  84                                         sumCalculators.push_back(new SharedMorHorn());
  85                                 }else if (globaldata->Estimators[i] == "sharedbraycurtis") {
  86                                         sumCalculators.push_back(new SharedBrayCurtis());
  87                                 }
  88                                 else if (globaldata->Estimators[i] == "sharedbdiversity") {
  89                                         sumCalculators.push_back(new SharedBDiversity());
  90                                 }
  91
  92                         }
  93                 }
  94                 //reset calc for next command
  95                 globaldata->setCalc("");
  96
  97         }
  98         catch(exception& e) {
  99                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 100                 exit(1);
 101         }
 102         catch(...) {
 103                 cout << "An unknown error has occurred in the SummarySharedCommand class function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 104                 exit(1);
 105         }
 106 }
 107 //**********************************************************************************************************************
 108
 109 SummarySharedCommand::~SummarySharedCommand(){
 110         delete input;
 111         delete read;
 112 }
 113
 114 //**********************************************************************************************************************
 115
 116 int SummarySharedCommand::execute(){
 117         try {
 118                 int count = 1;
 119
 120                 //if the users entered no valid calculators don't execute command
 121                 if (sumCalculators.size() == 0) { return 0; }
 122
 123                 if (format == "sharedfile") {
 124                         read = new ReadPhilFile(globaldata->inputFileName);
 125                         read->read(&*globaldata);
 126
 127                         input = globaldata->ginput;
 128                         order = input->getSharedOrderVector();
 129                 }else {
 130                         //you are using a list and a groupfile
 131                         read = new ReadPhilFile(globaldata->inputFileName);
 132                         read->read(&*globaldata);
 133
 134                         input = globaldata->ginput;
 135                         SharedList = globaldata->gSharedList;
 136                         order = SharedList->getSharedOrderVector();
 137                 }
 138
 139                 //set users groups
 140                 setGroups();
 141
 142                 //output estimator names as column headers
 143                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 144                 for(int i=0;i<sumCalculators.size();i++){
 145                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 146                 }
 147                 outputFileHandle << endl;
 148
 149                 while(order != NULL){
 150
 151                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 152
 153                                 cout << order->getLabel() << '\t' << count << endl;
 154                                 getSharedVectors();  //fills group vectors from order vector.
 155
 156                                 //randomize group order
 157                                 if (globaldata->getJumble() == "1") { random_shuffle(lookup.begin(), lookup.end()); }
 158
 159                                 int n = 1;
 160                                 for (int k = 0; k < (lookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 161                                         for (int l = n; l < lookup.size(); l++) {
 162                                                 outputFileHandle << order->getLabel() << '\t' << (lookup[k]->getGroup() + lookup[l]->getGroup()) << '\t' << '\t'; //print out label and group
 163                                                 outputFileHandle << order->getLabel() << '\t';
 164
 165                                                 //sort groups to be alphanumeric
 166                                                 if (lookup[k]->getGroup() > lookup[l]->getGroup()) {
 167                                                         outputFileHandle << (lookup[l]->getGroup() +'\t' + lookup[k]->getGroup()) << '\t'; //print out groups
 168                                                 }else{
 169                                                         outputFileHandle << (lookup[k]->getGroup() +'\t' + lookup[l]->getGroup()) << '\t'; //print out groups
 170                                                 }
 171
 172                                                 for(int i=0;i<sumCalculators.size();i++){
 173                                                         sumCalculators[i]->getValues(lookup[k], lookup[l]); //saves the calculator outputs
 174                                                         outputFileHandle << '\t';
 175                                                         sumCalculators[i]->print(outputFileHandle);
 176                                                 }
 177                                                 outputFileHandle << endl;
 178                                         }
 179                                         n++;
 180                                 }
 181                         }
 182
 183                         //get next line to process
 184                         if (format == "sharedfile") {
 185                                 order = input->getSharedOrderVector();
 186                         }else {
 187                                 //you are using a list and a groupfile
 188                                 SharedList = input->getSharedListVector(); //get new list vector to process
 189                                 if (SharedList != NULL) {
 190                                         order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
 191                                 }else {
 192                                         break;
 193                                 }
 194                         }
 195                         count++;
 196                 }
 197
 198                 //reset groups parameter
 199                 globaldata->Groups.clear();  globaldata->setGroups("");
 200
 201                 return 0;
 202         }
 203         catch(exception& e) {
 204                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 205                 exit(1);
 206         }
 207         catch(...) {
 208                 cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 209                 exit(1);
 210         }
 211 }
 212
 213 //**********************************************************************************************************************
 214
 215 void SummarySharedCommand::getSharedVectors(){
 216 try {
 217                 lookup.clear();
 218                 //create and initialize vector of sharedvectors, one for each group
 219                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 220                         SharedRAbundVector* temp = new SharedRAbundVector(order->getNumBins());
 221                         temp->setLabel(order->getLabel());
 222                         temp->setGroup(globaldata->Groups[i]);
 223                         lookup.push_back(temp);
 224                 }
 225
 226                 int numSeqs = order->size();
 227                 //sample all the members
 228                 for(int i=0;i<numSeqs;i++){
 229                         //get first sample
 230                         individual chosen = order->get(i);
 231                         int abundance;
 232
 233                         //set info for sharedvector in chosens group
 234                         for (int j = 0; j < lookup.size(); j++) {
 235                                 if (chosen.group == lookup[j]->getGroup()) {
 236                                          abundance = lookup[j]->getAbundance(chosen.bin);
 237                                          lookup[j]->set(chosen.bin, (abundance + 1), chosen.group);
 238                                          break;
 239                                 }
 240                         }
 241                 }
 242         }
 243         catch(exception& e) {
 244                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 245                 exit(1);
 246         }
 247         catch(...) {
 248                 cout << "An unknown error has occurred in the SummarySharedCommand class function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 249                 exit(1);
 250         }
 251
 252 }
 253
 254 //**********************************************************************************************************************
 255 void SummarySharedCommand::setGroups() {
 256         try {
 257                 //if the user has not entered specific groups to analyze then do them all
 258                 if (globaldata->Groups.size() != 0) {
 259                         if (globaldata->Groups[0] != "all") {
 260                                 //check that groups are valid
 261                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 262                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 263                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 264                                                 // erase the invalid group from globaldata->Groups
 265                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 266                                         }
 267                                 }
 268
 269                                 //if the user only entered invalid groups
 270                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 271                                         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;
 272                                         for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 273                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 274                                         }
 275                                 }
 276                         }else{//user has enter "all" and wants the default groups
 277                                 globaldata->Groups.clear();
 278                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 279                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 280                                 }
 281                                 globaldata->setGroups("");
 282                         }
 283                 }else {
 284                         for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 285                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 286                         }
 287                 }
 288
 289         }
 290         catch(exception& e) {
 291                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function setGroups. 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 SummarySharedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 296                 exit(1);
 297         }
 298
 299 }
 300 /***********************************************************/