sharedordervector.cpp

   1 /*
   2  *  sharedSharedOrderVector.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 12/9/08.
   6  *  Copyright 2008 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 using namespace std;
  11
  12
  13 #include "sharedordervector.h"
  14
  15
  16 /***********************************************************************/
  17
  18 SharedOrderVector::SharedOrderVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0)  {}
  19
  20 /***********************************************************************/
  21
  22 SharedOrderVector::SharedOrderVector(string id, vector<individual>  ov) :
  23                                                                                         DataVector(id), data(ov)
  24 {
  25                 updateStats();
  26 }
  27
  28 /***********************************************************************/
  29 //This function is used to read a .shared file for the collect.shared, rarefaction.shared and summary.shared commands
  30 //if you don't use a list and groupfile.
  31
  32 SharedOrderVector::SharedOrderVector(ifstream& f) : DataVector() {  //reads in a shared file
  33         try {
  34                 globaldata = GlobalData::getInstance();
  35                 maxRank = 0; numBins = 0; numSeqs = 0;
  36
  37                 if (globaldata->gGroupmap == NULL) {  groupmap = new GroupMap(); }
  38
  39                 int num, inputData, pos, count;
  40                 count = 0;  numSeqs = 0;
  41                 string holdLabel, nextLabel, groupN;
  42                 individual newguy;
  43
  44                 //read in first row since you know there is at least 1 group.
  45                 f >> label >> groupN >> num;
  46                 holdLabel = label;
  47
  48                 if (globaldata->gGroupmap == NULL) {
  49                         //save group in groupmap
  50                         groupmap->namesOfGroups.push_back(groupN);
  51                         groupmap->groupIndex[groupN] = 0;
  52                 }
  53
  54                 for(int i=0;i<num;i++){
  55                         f >> inputData;
  56
  57                         for (int j = 0; j < inputData; j++) {
  58                                 push_back(i, i, groupN);
  59                                 numSeqs++;
  60                         }
  61                 }
  62
  63                 //save position in file in case next line is a new label.
  64                 pos = f.tellg();
  65
  66                 if (f.eof() != true) { f >> nextLabel; }
  67
  68                 //read the rest of the groups info in
  69                 while ((nextLabel == holdLabel) && (f.eof() != true)) {
  70                         f >> groupN >> num;
  71                         count++;
  72
  73                         if (globaldata->gGroupmap == NULL) {
  74                                 //save group in groupmap
  75                                 groupmap->namesOfGroups.push_back(groupN);
  76                                 groupmap->groupIndex[groupN] = count;
  77                         }
  78
  79                         for(int i=0;i<num;i++){
  80                                 f >> inputData;
  81
  82                                 for (int j = 0; j < inputData; j++) {
  83                                         push_back(i, i, groupN);
  84                                         numSeqs++;
  85                                 }
  86                         }
  87
  88                         //save position in file in case next line is a new label.
  89                         pos = f.tellg();
  90
  91                         if (f.eof() != true) { f >> nextLabel; }
  92
  93                 }
  94
  95                 //put file pointer back since you are now at a new distance label
  96                 f.seekg(pos, ios::beg);
  97
  98                 if (globaldata->gGroupmap == NULL) { globaldata->gGroupmap = groupmap; }
  99
 100                 updateStats();
 101
 102         }
 103         catch(exception& e) {
 104                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function SharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 105                 exit(1);
 106         }
 107         catch(...) {
 108                 cout << "An unknown error has occurred in the SharedOrderVector class function SharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 109                 exit(1);
 110         }
 111 }
 112 /***********************************************************************/
 113
 114 int SharedOrderVector::getNumBins(){
 115         return numBins;
 116 }
 117
 118 /***********************************************************************/
 119
 120 int SharedOrderVector::getNumSeqs(){
 121         return numSeqs;
 122 }
 123
 124 /***********************************************************************/
 125
 126 int SharedOrderVector::getMaxRank(){
 127         return maxRank;
 128 }
 129
 130
 131 /***********************************************************************/
 132
 133
 134
 135 void SharedOrderVector::set(int index, int binNumber, int abund, string groupName){
 136
 137         data[index].group = groupName;
 138         data[index].bin = binNumber;
 139         data[index].abundance = abund;
 140         //if (abund > maxRank) { maxRank = abund; }
 141         updateStats();
 142 }
 143
 144 /***********************************************************************/
 145
 146 individual SharedOrderVector::get(int index){
 147         return data[index];
 148 }
 149
 150
 151 /***********************************************************************/
 152
 153 void SharedOrderVector::push_back(int binNumber, int abund, string groupName){
 154         individual newGuy;
 155         newGuy.group = groupName;
 156         newGuy.abundance = abund;
 157         newGuy.bin = binNumber;
 158         data.push_back(newGuy);
 159         //numSeqs++;
 160         //numBins++;
 161         //if (abund > maxRank) { maxRank = abund; }
 162         updateStats();
 163 }
 164
 165 /***********************************************************************/
 166
 167 void SharedOrderVector::print(ostream& output){
 168         try {
 169                 output << label << '\t' << numSeqs << '\t';
 170
 171                 for(int i=0;i<data.size();i++){
 172                         output << data[i].bin << '\t';
 173                 }
 174                 output << endl;
 175         }
 176         catch(exception& e) {
 177                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 178                 exit(1);
 179         }
 180         catch(...) {
 181                 cout << "An unknown error has occurred in the SharedOrderVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 182                 exit(1);
 183         }
 184 }
 185
 186
 187 /***********************************************************************/
 188
 189 void SharedOrderVector::resize(int){
 190         cout << "resize() did nothing in class SharedOrderVector";
 191 }
 192
 193 /***********************************************************************/
 194
 195
 196 vector<individual>::iterator SharedOrderVector::begin(){
 197         return data.begin();
 198 }
 199
 200 /***********************************************************************/
 201
 202 vector<individual>::iterator SharedOrderVector::end(){
 203         return data.end();
 204 }
 205
 206 /***********************************************************************/
 207
 208 int SharedOrderVector::size(){
 209         return data.size();
 210 }
 211
 212 /***********************************************************************/
 213
 214 RAbundVector SharedOrderVector::getRAbundVector(){
 215         try {
 216                 RAbundVector rav(data.size());
 217
 218                 for(int i=0;i<numSeqs;i++){
 219                         rav.set(data[i].bin, rav.get(data[i].bin) + 1);
 220                 }
 221                 sort(rav.rbegin(), rav.rend());
 222                 for(int i=numSeqs-1;i>=0;i--){
 223                         if(rav.get(i) == 0){    rav.pop_back(); }
 224                         else{
 225                                 break;
 226                         }
 227                 }
 228                 rav.setLabel(label);
 229
 230                 return rav;
 231         }
 232         catch(exception& e) {
 233                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 234                 exit(1);
 235         }
 236         catch(...) {
 237                 cout << "An unknown error has occurred in the SharedOrderVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 238                 exit(1);
 239         }
 240 }
 241 /***********************************************************************/
 242
 243 OrderVector SharedOrderVector::getOrderVector(map<string,int>* nameMap = NULL) {
 244         try {
 245                 OrderVector ov;
 246
 247                 for (int i = 0; i < data.size(); i++) {
 248                         ov.push_back(data[i].bin);
 249                 }
 250
 251                 random_shuffle(ov.begin(), ov.end());
 252
 253                 ov.setLabel(label);
 254                 return ov;
 255         }
 256         catch(exception& e) {
 257                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 258                 exit(1);
 259         }
 260         catch(...) {
 261                 cout << "An unknown error has occurred in the SharedOrderVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 262                 exit(1);
 263         }
 264 }
 265
 266
 267 /***********************************************************************/
 268
 269 SAbundVector SharedOrderVector::getSAbundVector(){
 270
 271         RAbundVector rav(this->getRAbundVector());
 272         return rav.getSAbundVector();
 273
 274 }
 275 /***********************************************************************/
 276 SharedRAbundVector SharedOrderVector::getSharedRAbundVector(string group) {
 277         try {
 278                 SharedRAbundVector sharedRav(data.size());
 279
 280                 sharedRav.setLabel(label);
 281                 sharedRav.setGroup(group);
 282
 283                 for (int i = 0; i < data.size(); i++) {
 284                         if (data[i].group == group) {
 285                                 sharedRav.set(data[i].abundance, sharedRav.getAbundance(data[i].abundance) + 1, data[i].group);
 286                         }
 287                 }
 288                 return sharedRav;
 289         }
 290         catch(exception& e) {
 291                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getSharedRAbundVector. 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 SharedOrderVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 296                 exit(1);
 297         }
 298
 299 }
 300 /***********************************************************************/
 301 SharedSAbundVector SharedOrderVector::getSharedSAbundVector(string group) {
 302         try {
 303
 304                 SharedRAbundVector sharedRav(this->getSharedRAbundVector(group));
 305                 return sharedRav.getSharedSAbundVector();
 306
 307         }
 308         catch(exception& e) {
 309                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 310                 exit(1);
 311         }
 312         catch(...) {
 313                 cout << "An unknown error has occurred in the SharedOrderVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 314                 exit(1);
 315         }
 316
 317 }
 318
 319 /***********************************************************************/
 320
 321 SharedOrderVector SharedOrderVector::getSharedOrderVector(){
 322         return *this;
 323 }
 324
 325 /***********************************************************************/
 326
 327 void SharedOrderVector::updateStats(){
 328         try {
 329                 needToUpdate = 0;
 330                 numSeqs = 0;
 331                 numBins = 0;
 332                 maxRank = 0;
 333
 334                 for(int i=0;i<data.size();i++){
 335                         if(data[i].bin != -1){
 336                                 numSeqs++;
 337                         }
 338                 }
 339
 340                 vector<individual> hold(numSeqs);
 341
 342                 for(int i=0;i<numSeqs;i++){
 343                         if(data[i].bin != -1){
 344                                 hold[data[i].bin].bin = hold[data[i].bin].bin+1;
 345                         }
 346                 }
 347
 348                 for(int i=0;i<numSeqs;i++){
 349                         if(data[i].bin > numBins) { numBins = data[i].bin;      }
 350                         if(data[i].abundance > maxRank) {       maxRank = data[i].abundance;    }
 351                 }
 352                 numBins++; //if you have 10 bins largest .bin is 9 since we start at 0.
 353         }
 354         catch(exception& e) {
 355                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function updateStats. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 356                 exit(1);
 357         }
 358         catch(...) {
 359                 cout << "An unknown error has occurred in the SharedOrderVector class function updateStats. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 360                 exit(1);
 361         }
 362 }
 363
 364 /***********************************************************************/
 365
 366