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