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 #include "utilities.hpp"
  15 #include <exception>
  16
  17 /***********************************************************************/
  18
  19 SharedOrderVector::SharedOrderVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0)  {}
  20
  21 /***********************************************************************/
  22
  23 SharedOrderVector::SharedOrderVector(string id, vector<individual>  ov) :
  24                                                                                         DataVector(id), data(ov)
  25 {
  26                 updateStats();
  27 }
  28
  29 /***********************************************************************/
  30 //This function is used to read a .shared file for the collect.shared, rarefaction.shared and summary.shared commands
  31 //if you don't use a list and groupfile.
  32
  33 SharedOrderVector::SharedOrderVector(ifstream& f) : DataVector() {
  34         try {
  35                 globaldata = GlobalData::getInstance();
  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                         for (int j = 0; j < inputData; j++) {
  57                                 push_back(i+1, i+1, groupN);
  58                                 numSeqs++;
  59                         }
  60                 }
  61
  62                 //save position in file in case next line is a new label.
  63                 pos = f.tellg();
  64
  65                 if (f.eof() != true) { f >> nextLabel; }
  66
  67                 //read the rest of the groups info in
  68                 while ((nextLabel == holdLabel) && (f.eof() != true)) {
  69                         f >> groupN >> num;
  70                         count++;
  71
  72                         if (globaldata->gGroupmap == NULL) {
  73                                 //save group in groupmap
  74                                 groupmap->namesOfGroups.push_back(groupN);
  75                                 groupmap->groupIndex[groupN] = count;
  76                         }
  77
  78                         for(int i=0;i<num;i++){
  79                                 f >> inputData;
  80                                 for (int j = 0; j < inputData; j++) {
  81                                         push_back(i+1, i+1, groupN);
  82                                         numSeqs++;
  83                                 }
  84                         }
  85
  86                         //save position in file in case next line is a new label.
  87                         pos = f.tellg();
  88
  89                         if (f.eof() != true) { f >> nextLabel; }
  90
  91                 }
  92
  93                 //put file pointer back since you are now at a new distance label
  94                 f.seekg(pos, ios::beg);
  95
  96                 if (globaldata->gGroupmap == NULL) { globaldata->gGroupmap = groupmap; }
  97
  98                 updateStats();
  99         }
 100         catch(exception& e) {
 101                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function SharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 102                 exit(1);
 103         }
 104         catch(...) {
 105                 cout << "An unknown error has occurred in the SharedOrderVector class function SharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 106                 exit(1);
 107         }
 108 }
 109 /***********************************************************************/
 110
 111 int SharedOrderVector::getNumBins(){
 112         if(needToUpdate == 1){  updateStats();  }
 113         return numBins;
 114 }
 115
 116 /***********************************************************************/
 117
 118 int SharedOrderVector::getNumSeqs(){
 119         if(needToUpdate == 1){  updateStats();  }
 120         return numSeqs;
 121 }
 122
 123 /***********************************************************************/
 124
 125 int SharedOrderVector::getMaxRank(){
 126         if(needToUpdate == 1){  updateStats();  }
 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         needToUpdate = 1;
 141
 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         needToUpdate = 1;
 160
 161 }
 162
 163 /***********************************************************************/
 164
 165 void SharedOrderVector::print(ostream& output){
 166         try {
 167                 output << label << '\t' << numSeqs << '\t';
 168
 169                 for(int i=0;i<data.size();i++){
 170                         output << data[i].bin << '\t';
 171                 }
 172                 output << endl;
 173         }
 174         catch(exception& e) {
 175                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function print. 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 SharedOrderVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 180                 exit(1);
 181         }
 182 }
 183
 184
 185 /***********************************************************************/
 186
 187 void SharedOrderVector::resize(int){
 188         cout << "resize() did nothing in class SharedOrderVector";
 189 }
 190
 191 /***********************************************************************/
 192
 193
 194 vector<individual>::iterator SharedOrderVector::begin(){
 195         return data.begin();
 196 }
 197
 198 /***********************************************************************/
 199
 200 vector<individual>::iterator SharedOrderVector::end(){
 201         return data.end();
 202 }
 203
 204 /***********************************************************************/
 205
 206 int SharedOrderVector::size(){
 207         return data.size();
 208 }
 209
 210 /***********************************************************************/
 211
 212 RAbundVector SharedOrderVector::getRAbundVector(){
 213         try {
 214                 RAbundVector rav(data.size());
 215
 216                 for(int i=0;i<numSeqs;i++){
 217                         rav.set(data[i].bin, rav.get(data[i].bin) + 1);
 218                 }
 219                 sort(rav.rbegin(), rav.rend());
 220                 for(int i=numSeqs-1;i>=0;i--){
 221                         if(rav.get(i) == 0){    rav.pop_back(); }
 222                         else{
 223                                 break;
 224                         }
 225                 }
 226                 rav.setLabel(label);
 227
 228                 return rav;
 229         }
 230         catch(exception& e) {
 231                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 232                 exit(1);
 233         }
 234         catch(...) {
 235                 cout << "An unknown error has occurred in the SharedOrderVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 236                 exit(1);
 237         }
 238 }
 239 /***********************************************************************/
 240
 241 OrderVector SharedOrderVector::getOrderVector(map<string,int>* nameMap = NULL) {
 242         try {
 243                 OrderVector ov;
 244
 245                 for (int i = 0; i < data.size(); i++) {
 246                         ov.push_back(data[i].bin);
 247                 }
 248
 249                 random_shuffle(ov.begin(), ov.end());
 250
 251                 ov.setLabel(label);
 252                 return ov;
 253         }
 254         catch(exception& e) {
 255                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 256                 exit(1);
 257         }
 258         catch(...) {
 259                 cout << "An unknown error has occurred in the SharedOrderVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 260                 exit(1);
 261         }
 262 }
 263
 264
 265 /***********************************************************************/
 266
 267 SAbundVector SharedOrderVector::getSAbundVector(){
 268
 269         RAbundVector rav(this->getRAbundVector());
 270         return rav.getSAbundVector();
 271
 272 }
 273 /***********************************************************************/
 274 SharedRAbundVector SharedOrderVector::getSharedRAbundVector(string group) {
 275         try {
 276                 SharedRAbundVector sharedRav(data.size());
 277
 278                 sharedRav.setLabel(label);
 279                 sharedRav.setGroup(group);
 280
 281                 for (int i = 0; i < data.size(); i++) {
 282                         if (data[i].group == group) {
 283                                 sharedRav.set(data[i].abundance, sharedRav.getAbundance(data[i].abundance) + 1, data[i].group);
 284                         }
 285                 }
 286                 return sharedRav;
 287         }
 288         catch(exception& e) {
 289                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 290                 exit(1);
 291         }
 292         catch(...) {
 293                 cout << "An unknown error has occurred in the SharedOrderVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 294                 exit(1);
 295         }
 296
 297 }
 298 /***********************************************************************/
 299 SharedSAbundVector SharedOrderVector::getSharedSAbundVector(string group) {
 300         try {
 301
 302                 SharedRAbundVector sharedRav(this->getSharedRAbundVector(group));
 303                 return sharedRav.getSharedSAbundVector();
 304
 305         }
 306         catch(exception& e) {
 307                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 308                 exit(1);
 309         }
 310         catch(...) {
 311                 cout << "An unknown error has occurred in the SharedOrderVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 312                 exit(1);
 313         }
 314
 315 }
 316
 317 /***********************************************************************/
 318
 319 SharedOrderVector SharedOrderVector::getSharedOrderVector(){
 320         return *this;
 321 }
 322
 323 /***********************************************************************/
 324
 325 void SharedOrderVector::updateStats(){
 326         try {
 327                 needToUpdate = 0;
 328                 numSeqs = 0;
 329                 numBins = 0;
 330                 maxRank = 0;
 331
 332                 for(int i=0;i<data.size();i++){
 333                         if(data[i].bin != -1){
 334                                 numSeqs++;
 335                         }
 336                 }
 337
 338                 vector<individual> hold(numSeqs);
 339
 340                 for(int i=0;i<numSeqs;i++){
 341                         if(data[i].bin != -1){
 342                                 hold[data[i].bin].bin = hold[data[i].bin].bin+1;
 343                         }
 344                 }
 345
 346                 for(int i=0;i<numSeqs;i++){
 347                         if(data[i].bin > numBins) { numBins = data[i].bin;      }
 348                         if(data[i].abundance > maxRank) {       maxRank = data[i].abundance;    }
 349                 }
 350                 numBins++; //if you have 10 bins largest .bin is 9 since we start at 0.
 351         }
 352         catch(exception& e) {
 353                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function updateStats. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 354                 exit(1);
 355         }
 356         catch(...) {
 357                 cout << "An unknown error has occurred in the SharedOrderVector class function updateStats. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 358                 exit(1);
 359         }
 360 }
 361
 362 /***********************************************************************/
 363
 364