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