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
  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                 maxRank = 0; numBins = 0; numSeqs = 0;
  37
  38                 if (globaldata->gGroupmap == NULL) {  groupmap = new GroupMap(); }
  39
  40                 int num, inputData, pos, count;
  41                 count = 0;  numSeqs = 0;
  42                 string holdLabel, nextLabel, groupN;
  43                 individual newguy;
  44
  45                 //read in first row since you know there is at least 1 group.
  46                 f >> label >> groupN >> num;
  47                 holdLabel = label;
  48
  49                 if (globaldata->gGroupmap == NULL) {
  50                         //save group in groupmap
  51                         groupmap->namesOfGroups.push_back(groupN);
  52                         groupmap->groupIndex[groupN] = 0;
  53                 }
  54
  55                 for(int i=0;i<num;i++){
  56                         f >> inputData;
  57
  58                         for (int j = 0; j < inputData; j++) {
  59                                 push_back(i+1, i+1, groupN);
  60                                 numSeqs++;
  61                         }
  62                 }
  63
  64                 //save position in file in case next line is a new label.
  65                 pos = f.tellg();
  66
  67                 if (f.eof() != true) { f >> nextLabel; }
  68
  69                 //read the rest of the groups info in
  70                 while ((nextLabel == holdLabel) && (f.eof() != true)) {
  71                         f >> groupN >> num;
  72                         count++;
  73
  74                         if (globaldata->gGroupmap == NULL) {
  75                                 //save group in groupmap
  76                                 groupmap->namesOfGroups.push_back(groupN);
  77                                 groupmap->groupIndex[groupN] = count;
  78                         }
  79
  80                         for(int i=0;i<num;i++){
  81                                 f >> inputData;
  82
  83                                 for (int j = 0; j < inputData; j++) {
  84                                         push_back(i+1, i+1, groupN);
  85                                         numSeqs++;
  86                                 }
  87                         }
  88
  89                         //save position in file in case next line is a new label.
  90                         pos = f.tellg();
  91
  92                         if (f.eof() != true) { f >> nextLabel; }
  93
  94                 }
  95
  96                 //put file pointer back since you are now at a new distance label
  97                 f.seekg(pos, ios::beg);
  98
  99                 if (globaldata->gGroupmap == NULL) { globaldata->gGroupmap = groupmap; }
 100
 101                 updateStats();
 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         if(needToUpdate == 1){  updateStats();  }
 116         return numBins;
 117 }
 118
 119 /***********************************************************************/
 120
 121 int SharedOrderVector::getNumSeqs(){
 122         if(needToUpdate == 1){  updateStats();  }
 123         return numSeqs;
 124 }
 125
 126 /***********************************************************************/
 127
 128 int SharedOrderVector::getMaxRank(){
 129         if(needToUpdate == 1){  updateStats();  }
 130         return maxRank;
 131 }
 132
 133
 134 /***********************************************************************/
 135
 136
 137
 138 void SharedOrderVector::set(int index, int binNumber, int abund, string groupName){
 139
 140         data[index].group = groupName;
 141         data[index].bin = binNumber;
 142         data[index].abundance = abund;
 143         needToUpdate = 1;
 144
 145 }
 146
 147 /***********************************************************************/
 148
 149 individual SharedOrderVector::get(int index){
 150         return data[index];
 151 }
 152
 153
 154 /***********************************************************************/
 155
 156 void SharedOrderVector::push_back(int binNumber, int abund, string groupName){
 157         individual newGuy;
 158         newGuy.group = groupName;
 159         newGuy.abundance = abund;
 160         newGuy.bin = binNumber;
 161         data.push_back(newGuy);
 162         needToUpdate = 1;
 163
 164 }
 165
 166 /***********************************************************************/
 167
 168 void SharedOrderVector::print(ostream& output){
 169         try {
 170                 output << label << '\t' << numSeqs << '\t';
 171
 172                 for(int i=0;i<data.size();i++){
 173                         output << data[i].bin << '\t';
 174                 }
 175                 output << endl;
 176         }
 177         catch(exception& e) {
 178                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 179                 exit(1);
 180         }
 181         catch(...) {
 182                 cout << "An unknown error has occurred in the SharedOrderVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 183                 exit(1);
 184         }
 185 }
 186
 187
 188 /***********************************************************************/
 189
 190 void SharedOrderVector::resize(int){
 191         cout << "resize() did nothing in class SharedOrderVector";
 192 }
 193
 194 /***********************************************************************/
 195
 196
 197 vector<individual>::iterator SharedOrderVector::begin(){
 198         return data.begin();
 199 }
 200
 201 /***********************************************************************/
 202
 203 vector<individual>::iterator SharedOrderVector::end(){
 204         return data.end();
 205 }
 206
 207 /***********************************************************************/
 208
 209 int SharedOrderVector::size(){
 210         return data.size();
 211 }
 212
 213 /***********************************************************************/
 214
 215 RAbundVector SharedOrderVector::getRAbundVector(){
 216         try {
 217                 RAbundVector rav(data.size());
 218
 219                 for(int i=0;i<numSeqs;i++){
 220                         rav.set(data[i].bin, rav.get(data[i].bin) + 1);
 221                 }
 222                 sort(rav.rbegin(), rav.rend());
 223                 for(int i=numSeqs-1;i>=0;i--){
 224                         if(rav.get(i) == 0){    rav.pop_back(); }
 225                         else{
 226                                 break;
 227                         }
 228                 }
 229                 rav.setLabel(label);
 230
 231                 return rav;
 232         }
 233         catch(exception& e) {
 234                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 235                 exit(1);
 236         }
 237         catch(...) {
 238                 cout << "An unknown error has occurred in the SharedOrderVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 239                 exit(1);
 240         }
 241 }
 242 /***********************************************************************/
 243
 244 OrderVector SharedOrderVector::getOrderVector(map<string,int>* nameMap = NULL) {
 245         try {
 246                 OrderVector ov;
 247
 248                 for (int i = 0; i < data.size(); i++) {
 249                         ov.push_back(data[i].bin);
 250                 }
 251
 252                 random_shuffle(ov.begin(), ov.end());
 253
 254                 ov.setLabel(label);
 255                 return ov;
 256         }
 257         catch(exception& e) {
 258                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 259                 exit(1);
 260         }
 261         catch(...) {
 262                 cout << "An unknown error has occurred in the SharedOrderVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 263                 exit(1);
 264         }
 265 }
 266
 267
 268 /***********************************************************************/
 269
 270 SAbundVector SharedOrderVector::getSAbundVector(){
 271
 272         RAbundVector rav(this->getRAbundVector());
 273         return rav.getSAbundVector();
 274
 275 }
 276 /***********************************************************************/
 277 SharedRAbundVector SharedOrderVector::getSharedRAbundVector(string group) {
 278         try {
 279                 SharedRAbundVector sharedRav(data.size());
 280
 281                 sharedRav.setLabel(label);
 282                 sharedRav.setGroup(group);
 283
 284                 for (int i = 0; i < data.size(); i++) {
 285                         if (data[i].group == group) {
 286                                 sharedRav.set(data[i].abundance, sharedRav.getAbundance(data[i].abundance) + 1, data[i].group);
 287                         }
 288                 }
 289                 return sharedRav;
 290         }
 291         catch(exception& e) {
 292                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 293                 exit(1);
 294         }
 295         catch(...) {
 296                 cout << "An unknown error has occurred in the SharedOrderVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 297                 exit(1);
 298         }
 299
 300 }
 301 /***********************************************************************/
 302 SharedSAbundVector SharedOrderVector::getSharedSAbundVector(string group) {
 303         try {
 304
 305                 SharedRAbundVector sharedRav(this->getSharedRAbundVector(group));
 306                 return sharedRav.getSharedSAbundVector();
 307
 308         }
 309         catch(exception& e) {
 310                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 311                 exit(1);
 312         }
 313         catch(...) {
 314                 cout << "An unknown error has occurred in the SharedOrderVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 315                 exit(1);
 316         }
 317
 318 }
 319
 320 /***********************************************************************/
 321
 322 SharedOrderVector SharedOrderVector::getSharedOrderVector(){
 323         return *this;
 324 }
 325
 326 /***********************************************************************/
 327
 328 void SharedOrderVector::updateStats(){
 329         try {
 330                 needToUpdate = 0;
 331                 numSeqs = 0;
 332                 numBins = 0;
 333                 maxRank = 0;
 334
 335                 for(int i=0;i<data.size();i++){
 336                         if(data[i].bin != -1){
 337                                 numSeqs++;
 338                         }
 339                 }
 340
 341                 vector<individual> hold(numSeqs);
 342
 343                 for(int i=0;i<numSeqs;i++){
 344                         if(data[i].bin != -1){
 345                                 hold[data[i].bin].bin = hold[data[i].bin].bin+1;
 346                         }
 347                 }
 348
 349                 for(int i=0;i<numSeqs;i++){
 350                         if(data[i].bin > numBins) { numBins = data[i].bin;      }
 351                         if(data[i].abundance > maxRank) {       maxRank = data[i].abundance;    }
 352                 }
 353                 numBins++; //if you have 10 bins largest .bin is 9 since we start at 0.
 354         }
 355         catch(exception& e) {
 356                 cout << "Standard Error: " << e.what() << " has occurred in the SharedOrderVector class Function updateStats. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 357                 exit(1);
 358         }
 359         catch(...) {
 360                 cout << "An unknown error has occurred in the SharedOrderVector class function updateStats. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 361                 exit(1);
 362         }
 363 }
 364
 365 /***********************************************************************/
 366
 367