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 "sharedjabund.h"
  15 #include "sharedsorabund.h"
  16 #include "sharedjclass.h"
  17 #include "sharedsorclass.h"
  18 #include "sharedjest.h"
  19 #include "sharedsorest.h"
  20 #include "sharedthetayc.h"
  21 #include "sharedthetan.h"
  22
  23
  24 //**********************************************************************************************************************
  25
  26 SummarySharedCommand::SummarySharedCommand(){
  27         try {
  28                 globaldata = GlobalData::getInstance();
  29                 outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
  30                 openOutputFile(outputFileName, outputFileHandle);
  31                 format = globaldata->getFormat();
  32                 validCalculator = new ValidCalculators();
  33
  34                 int i;
  35                 for (i=0; i<globaldata->Estimators.size(); i++) {
  36                         if (validCalculator->isValidCalculator("sharedsummary", globaldata->Estimators[i]) == true) {
  37                                 if (globaldata->Estimators[i] == "sharedsobs") {
  38                                         sumCalculators.push_back(new SharedSobsCS());
  39                                 }else if (globaldata->Estimators[i] == "sharedchao") {
  40                                         sumCalculators.push_back(new SharedChao1());
  41                                 }else if (globaldata->Estimators[i] == "sharedace") {
  42                                         sumCalculators.push_back(new SharedAce());
  43                                 }else if (globaldata->Estimators[i] == "sharedjabund") {
  44                                         sumCalculators.push_back(new SharedJAbund());
  45                                 }else if (globaldata->Estimators[i] == "sharedsorensonabund") {
  46                                         sumCalculators.push_back(new SharedSorAbund());
  47                                 }else if (globaldata->Estimators[i] == "sharedjclass") {
  48                                         sumCalculators.push_back(new SharedJclass());
  49                                 }else if (globaldata->Estimators[i] == "sharedsorclass") {
  50                                         sumCalculators.push_back(new SharedSorClass());
  51                                 }else if (globaldata->Estimators[i] == "sharedjest") {
  52                                         sumCalculators.push_back(new SharedJest());
  53                                 }else if (globaldata->Estimators[i] == "sharedsorest") {
  54                                         sumCalculators.push_back(new SharedSorEst());
  55                                 }else if (globaldata->Estimators[i] == "sharedthetayc") {
  56                                         sumCalculators.push_back(new SharedThetaYC());
  57                                 }else if (globaldata->Estimators[i] == "sharedthetan") {
  58                                         sumCalculators.push_back(new SharedThetaN());
  59                                 }
  60                         }
  61                 }
  62
  63                 //reset calc for next command
  64                 globaldata->setCalc("");
  65
  66         }
  67         catch(exception& e) {
  68                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  69                 exit(1);
  70         }
  71         catch(...) {
  72                 cout << "An unknown error has occurred in the SummarySharedCommand class function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  73                 exit(1);
  74         }
  75 }
  76 //**********************************************************************************************************************
  77
  78 SummarySharedCommand::~SummarySharedCommand(){
  79         delete input;
  80         delete read;
  81 }
  82
  83 //**********************************************************************************************************************
  84
  85 int SummarySharedCommand::execute(){
  86         try {
  87                 int count = 1;
  88
  89                 //if the users entered no valid calculators don't execute command
  90                 if (sumCalculators.size() == 0) { return 0; }
  91
  92                 if (format == "sharedfile") {
  93                         read = new ReadPhilFile(globaldata->inputFileName);
  94                         read->read(&*globaldata);
  95
  96                         input = globaldata->ginput;
  97                         order = input->getSharedOrderVector();
  98                 }else {
  99                         //you are using a list and a groupfile
 100                         read = new ReadPhilFile(globaldata->inputFileName);
 101                         read->read(&*globaldata);
 102
 103                         input = globaldata->ginput;
 104                         SharedList = globaldata->gSharedList;
 105                         order = SharedList->getSharedOrderVector();
 106                 }
 107
 108                 //output estimator names as column headers
 109                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 110                 for(int i=0;i<sumCalculators.size();i++){
 111                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 112                 }
 113                 outputFileHandle << endl;
 114
 115                 //set users groups
 116                 setGroups();
 117
 118                 while(order != NULL){
 119
 120                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 121
 122                                 cout << order->getLabel() << '\t' << count << endl;
 123                                 getSharedVectors();  //fills group vectors from order vector.
 124
 125                                 //randomize group order
 126                                 if (globaldata->getJumble() == "1") { random_shuffle(lookup.begin(), lookup.end()); }
 127
 128                                 int n = 1;
 129                                 for (int k = 0; k < (lookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 130                                         for (int l = n; l < lookup.size(); l++) {
 131                                                 outputFileHandle << order->getLabel() << '\t' << (lookup[k]->getGroup() + lookup[l]->getGroup()) << '\t' << '\t'; //print out label and group
 132                                                 for(int i=0;i<sumCalculators.size();i++){
 133                                                         sumCalculators[i]->getValues(lookup[k], lookup[l]); //saves the calculator outputs
 134                                                         outputFileHandle << '\t';
 135                                                         sumCalculators[i]->print(outputFileHandle);
 136                                                 }
 137                                                 outputFileHandle << endl;
 138                                         }
 139                                         n++;
 140                                 }
 141                         }
 142
 143                         //get next line to process
 144                         if (format == "sharedfile") {
 145                                 order = input->getSharedOrderVector();
 146                         }else {
 147                                 //you are using a list and a groupfile
 148                                 SharedList = input->getSharedListVector(); //get new list vector to process
 149                                 if (SharedList != NULL) {
 150                                         order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
 151                                 }else {
 152                                         break;
 153                                 }
 154                         }
 155                         count++;
 156                 }
 157
 158                 //reset groups parameter
 159                 globaldata->Groups.clear();  globaldata->setGroups("");
 160
 161                 return 0;
 162         }
 163         catch(exception& e) {
 164                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 165                 exit(1);
 166         }
 167         catch(...) {
 168                 cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 169                 exit(1);
 170         }
 171 }
 172
 173 //**********************************************************************************************************************
 174
 175 void SummarySharedCommand::getSharedVectors(){
 176 try {
 177                 lookup.clear();
 178                 //create and initialize vector of sharedvectors, one for each group
 179                 for (int i = 0; i < globaldata->gGroupmap->getNumGroups(); i++) {
 180                         SharedRAbundVector* temp = new SharedRAbundVector(order->getNumBins());
 181                         temp->setLabel(order->getLabel());
 182                         temp->setGroup(globaldata->gGroupmap->namesOfGroups[i]);
 183                         lookup.push_back(temp);
 184                 }
 185
 186                 int numSeqs = order->size();
 187                 //sample all the members
 188                 for(int i=0;i<numSeqs;i++){
 189                         //get first sample
 190                         individual chosen = order->get(i);
 191                         int abundance;
 192
 193                         //set info for sharedvector in chosens group
 194                         for (int j = 0; j < lookup.size(); j++) {
 195                                 if (chosen.group == lookup[j]->getGroup()) {
 196                                          abundance = lookup[j]->getAbundance(chosen.bin);
 197                                          lookup[j]->set(chosen.bin, (abundance + 1), chosen.group);
 198                                          break;
 199                                 }
 200                         }
 201                 }
 202
 203                 //get rid of vectors from groups you don't want to analyze
 204                 for (int r = 0; r < lookup.size(); r++) {
 205                         if (inUsersGroups(lookup[r]->getGroup(), globaldata->Groups) != true) {
 206                                 lookup.erase(lookup.begin()+r);
 207                         }
 208                 }
 209
 210         }
 211         catch(exception& e) {
 212                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 213                 exit(1);
 214         }
 215         catch(...) {
 216                 cout << "An unknown error has occurred in the SummarySharedCommand class function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 217                 exit(1);
 218         }
 219
 220 }
 221
 222 //**********************************************************************************************************************
 223 void SummarySharedCommand::setGroups() {
 224         try {
 225                 //if the user has not entered specific groups to analyze then do them all
 226                 if (globaldata->Groups.size() != 0) {
 227                         if (globaldata->Groups[0] != "all") {
 228                                 //check that groups are valid
 229                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 230                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 231                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 232                                                 // erase the invalid group from globaldata->Groups
 233                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 234                                         }
 235                                 }
 236
 237                                 //if the user only entered invalid groups
 238                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 239                                         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;
 240                                         for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 241                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 242                                         }
 243                                 }
 244                         }else{//user has enter "all" and wants the default groups
 245                                 globaldata->Groups.clear();
 246                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 247                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 248                                 }
 249                                 globaldata->setGroups("");
 250                         }
 251                 }else {
 252                         for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 253                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 254                         }
 255                 }
 256
 257         }
 258         catch(exception& e) {
 259                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 260                 exit(1);
 261         }
 262         catch(...) {
 263                 cout << "An unknown error has occurred in the SummarySharedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 264                 exit(1);
 265         }
 266
 267 }
 268 /***********************************************************/