sharedrabundvector.cpp

   1 /*
   2  *  sharedvector.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 12/5/08.
   6  *  Copyright 2008 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10
  11 using namespace std;
  12
  13 #include "sharedrabundvector.h"
  14 #include "sabundvector.hpp"
  15 #include "ordervector.hpp"
  16 #include "sharedutilities.h"
  17
  18
  19 /***********************************************************************/
  20
  21 SharedRAbundVector::SharedRAbundVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0) {};
  22
  23 /***********************************************************************/
  24
  25 SharedRAbundVector::SharedRAbundVector(int n) : DataVector(), maxRank(0), numBins(n), numSeqs(0) {
  26                 individual newGuy;
  27                 //initialize data
  28                 for (int i=0; i< n; i++) {
  29                         newGuy.bin = i;
  30                         newGuy.abundance = 0;
  31                         data.push_back(newGuy);
  32                 }
  33 };
  34
  35 /***********************************************************************
  36
  37 SharedRAbundVector::SharedRAbundVector(string id, vector<individual> rav) : DataVector(id), data(rav) {
  38         try {
  39                 numBins = 0;
  40                 maxRank = 0;
  41                 numSeqs = 0;
  42
  43                 for(int i=0;i<data.size();i++){
  44                         if(data[i].abundance != 0)              {       numBins = i+1;          }
  45                         if(data[i].abundance > maxRank) {       maxRank = data[i].abundance;    }
  46                         numSeqs += data[i].abundance;
  47                 }
  48         }
  49         catch(exception& e) {
  50                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function SharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  51                 exit(1);
  52         }
  53         catch(...) {
  54                 cout << "An unknown error has occurred in the SharedRAbundVector class function SharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  55                 exit(1);
  56         }
  57 }
  58
  59
  60 /***********************************************************************/
  61 //reads a shared file
  62 SharedRAbundVector::SharedRAbundVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
  63         try {
  64                 globaldata = GlobalData::getInstance();
  65
  66                 if (globaldata->gGroupmap == NULL) {  groupmap = new GroupMap(); }
  67
  68                 int num, inputData, pos, count;
  69                 count = 0;
  70                 string holdLabel, nextLabel, groupN;
  71                 individual newguy;
  72
  73                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
  74                 lookup.clear();
  75
  76                 //read in first row since you know there is at least 1 group.
  77                 f >> label >> groupN >> num;
  78                 holdLabel = label;
  79
  80                 //add new vector to lookup
  81                 SharedRAbundVector* temp = new SharedRAbundVector();
  82                 lookup.push_back(temp);
  83                 lookup[0]->setLabel(label);
  84                 lookup[0]->setGroup(groupN);
  85
  86                 if (globaldata->gGroupmap == NULL) {
  87                         //save group in groupmap
  88                         groupmap->namesOfGroups.push_back(groupN);
  89                         groupmap->groupIndex[groupN] = 0;
  90                 }
  91
  92                 //fill vector.  data = first sharedrabund in file
  93                 for(int i=0;i<num;i++){
  94                         f >> inputData;
  95
  96                         lookup[0]->push_back(inputData, i, groupN); //abundance, bin, group
  97                         push_back(inputData, i, groupN);
  98                         numSeqs += inputData;
  99                         numBins++;
 100                         if (inputData > maxRank) { maxRank = inputData; }
 101
 102                 }
 103
 104                 //save position in file in case next line is a new label.
 105                 pos = f.tellg();
 106
 107                 if (f.eof() != true) { f >> nextLabel; }
 108
 109                 //read the rest of the groups info in
 110                 while ((nextLabel == holdLabel) && (f.eof() != true)) {
 111                         f >> groupN >> num;
 112                         count++;
 113
 114                         if (globaldata->gGroupmap == NULL) {
 115                                 //save group in groupmap
 116                                 groupmap->namesOfGroups.push_back(groupN);
 117                                 groupmap->groupIndex[groupN] = count;
 118                         }
 119
 120                         //add new vector to lookup
 121                         temp = new SharedRAbundVector();
 122                         lookup.push_back(temp);
 123                         lookup[count]->setLabel(label);
 124                         lookup[count]->setGroup(groupN);
 125
 126                         //fill vector.
 127                         for(int i=0;i<num;i++){
 128                                 f >> inputData;
 129                                 lookup[count]->push_back(inputData, i, groupN); //abundance, bin, group
 130                         }
 131
 132                         //save position in file in case next line is a new label.
 133                         pos = f.tellg();
 134
 135                         if (f.eof() != true) { f >> nextLabel; }
 136                 }
 137
 138                 //put file pointer back since you are now at a new distance label
 139                 f.seekg(pos, ios::beg);
 140
 141                 if (globaldata->gGroupmap == NULL) { globaldata->gGroupmap = groupmap; }
 142
 143         }
 144         catch(exception& e) {
 145                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function SharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 146                 exit(1);
 147         }
 148         catch(...) {
 149                 cout << "An unknown error has occurred in the SharedRAbundVector class function SharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 150                 exit(1);
 151         }
 152 }
 153
 154 /***********************************************************************/
 155
 156 void SharedRAbundVector::set(int binNumber, int newBinSize, string groupname){
 157         try {
 158                 int oldBinSize = data[binNumber].abundance;
 159                 data[binNumber].abundance = newBinSize;
 160                 data[binNumber].group = groupname;
 161
 162                 if(newBinSize > maxRank)        {       maxRank = newBinSize;   }
 163
 164                 numSeqs += (newBinSize - oldBinSize);
 165         }
 166         catch(exception& e) {
 167                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 168                 exit(1);
 169         }
 170         catch(...) {
 171                 cout << "An unknown error has occurred in the SharedRAbundVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 172                 exit(1);
 173         }
 174 }
 175 /***********************************************************************/
 176
 177 void SharedRAbundVector::setData(vector <individual> newData){
 178         data = newData;
 179 }
 180
 181 /***********************************************************************/
 182
 183 int SharedRAbundVector::getAbundance(int index){
 184         return data[index].abundance;
 185
 186 }
 187 /***********************************************************************/
 188
 189 int SharedRAbundVector::numNZ(){
 190         int sum = 0;
 191         for(int i = 1; i < numBins; i++)
 192                 if(data[i].abundance > 0)
 193                         sum++;
 194         return sum;
 195 }
 196 /***********************************************************************/
 197
 198 void SharedRAbundVector::sortD(){
 199         struct individual indObj;
 200         sort(data.begin()+1, data.end(), indObj);
 201 }
 202 /***********************************************************************/
 203
 204 individual SharedRAbundVector::get(int index){
 205         return data[index];
 206
 207 }
 208 /***********************************************************************/
 209
 210 vector <individual> SharedRAbundVector::getData(){
 211         return data;
 212 }
 213 /***********************************************************************/
 214
 215 void SharedRAbundVector::push_back(int binSize, int otu, string groupName){
 216         try {
 217                 individual newGuy;
 218                 newGuy.abundance = binSize;
 219                 newGuy.group = groupName;
 220                 newGuy.bin = otu;
 221
 222                 data.push_back(newGuy);
 223                 numBins++;
 224
 225                 if(binSize > maxRank){
 226                         maxRank = binSize;
 227                 }
 228
 229                 numSeqs += binSize;
 230         }
 231         catch(exception& e) {
 232                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 233                 exit(1);
 234         }
 235         catch(...) {
 236                 cout << "An unknown error has occurred in the SharedRAbundVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 237                 exit(1);
 238         }
 239 }
 240
 241 /***********************************************************************/
 242
 243 void SharedRAbundVector::insert(int binSize, int otu, string groupName){
 244         try {
 245                 individual newGuy;
 246                 newGuy.abundance = binSize;
 247                 newGuy.group = groupName;
 248                 newGuy.bin = otu;
 249
 250                 data.insert(data.begin()+otu, newGuy);
 251                 numBins++;
 252
 253                 if(binSize > maxRank){
 254                         maxRank = binSize;
 255                 }
 256
 257                 numSeqs += binSize;
 258         }
 259         catch(exception& e) {
 260                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function insert. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 261                 exit(1);
 262         }
 263         catch(...) {
 264                 cout << "An unknown error has occurred in the SharedRAbundVector class function insert. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 265                 exit(1);
 266         }
 267 }
 268
 269 /***********************************************************************/
 270
 271 void SharedRAbundVector::push_front(int binSize, int otu, string groupName){
 272         try {
 273                 individual newGuy;
 274                 newGuy.abundance = binSize;
 275                 newGuy.group = groupName;
 276                 newGuy.bin = otu;
 277
 278                 data.insert(data.begin(), newGuy);
 279                 numBins++;
 280
 281                 if(binSize > maxRank){
 282                         maxRank = binSize;
 283                 }
 284
 285                 numSeqs += binSize;
 286         }
 287         catch(exception& e) {
 288                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function push_front. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 289                 exit(1);
 290         }
 291         catch(...) {
 292                 cout << "An unknown error has occurred in the SharedRAbundVector class function push_front. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 293                 exit(1);
 294         }
 295 }
 296
 297 /***********************************************************************/
 298 void SharedRAbundVector::pop_back(){
 299         numSeqs -= data[data.size()-1].abundance;
 300         numBins--;
 301         return data.pop_back();
 302 }
 303
 304 /***********************************************************************/
 305
 306
 307 vector<individual>::reverse_iterator SharedRAbundVector::rbegin(){
 308         return data.rbegin();
 309 }
 310
 311 /***********************************************************************/
 312
 313 vector<individual>::reverse_iterator SharedRAbundVector::rend(){
 314         return data.rend();
 315 }
 316
 317 /***********************************************************************/
 318 void SharedRAbundVector::resize(int size){
 319
 320         data.resize(size);
 321 }
 322
 323 /***********************************************************************/
 324
 325 int SharedRAbundVector::size(){
 326         return data.size();
 327 }
 328
 329 /***********************************************************************/
 330 void SharedRAbundVector::print(ostream& output){
 331         try {
 332                 output << numBins << '\t';
 333
 334                 for(int i=0;i<data.size();i++){         output << data[i].abundance << '\t';            }
 335                 output << endl;
 336         }
 337         catch(exception& e) {
 338                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 339                 exit(1);
 340         }
 341         catch(...) {
 342                 cout << "An unknown error has occurred in the SharedRAbundVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 343                 exit(1);
 344         }
 345 }
 346 /***********************************************************************/
 347 string SharedRAbundVector::getGroup(){
 348         return group;
 349 }
 350
 351 /***********************************************************************/
 352
 353 void SharedRAbundVector::setGroup(string groupName){
 354         group = groupName;
 355 }
 356 /***********************************************************************/
 357 int SharedRAbundVector::getGroupIndex()  { return index; }
 358 /***********************************************************************/
 359 void SharedRAbundVector::setGroupIndex(int vIndex)      { index = vIndex; }
 360 /***********************************************************************/
 361 int SharedRAbundVector::getNumBins(){
 362         return numBins;
 363 }
 364
 365 /***********************************************************************/
 366
 367 int SharedRAbundVector::getNumSeqs(){
 368         return numSeqs;
 369 }
 370
 371 /***********************************************************************/
 372
 373 int SharedRAbundVector::getMaxRank(){
 374         return maxRank;
 375 }
 376 /***********************************************************************/
 377
 378 SharedRAbundVector SharedRAbundVector::getSharedRAbundVector(){
 379         return *this;
 380 }
 381 /***********************************************************************/
 382 vector<SharedRAbundVector*> SharedRAbundVector::getSharedRAbundVectors(){
 383         try {
 384                 SharedUtil* util;
 385                 util = new SharedUtil();
 386
 387                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups);
 388
 389                 for (int i = 0; i < lookup.size(); i++) {
 390                         //if this sharedrabund is not from a group the user wants then delete it.
 391                         if (util->isValidGroup(lookup[i]->getGroup(), globaldata->Groups) == false) {
 392                                 delete lookup[i];
 393                                 lookup.erase(lookup.begin()+i);
 394                                 i--;
 395                         }
 396                 }
 397
 398                 return lookup;
 399         }
 400         catch(exception& e) {
 401                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSharedRAbundVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 402                 exit(1);
 403         }
 404         catch(...) {
 405                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSharedRAbundVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 406                 exit(1);
 407         }
 408 }
 409 /***********************************************************************/
 410
 411 RAbundVector SharedRAbundVector::getRAbundVector() {
 412         try {
 413                 RAbundVector rav(data.size());
 414
 415                 for (int i = 0; i < data.size(); i++) {
 416                         rav.set(i, data[i].abundance);
 417                 }
 418
 419                 return rav;
 420         }
 421         catch(exception& e) {
 422                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 423                 exit(1);
 424         }
 425         catch(...) {
 426                 cout << "An unknown error has occurred in the SharedRAbundVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 427                 exit(1);
 428         }
 429 }
 430 /***********************************************************************/
 431
 432 RAbundVector SharedRAbundVector::getRAbundVector2() {
 433         try {
 434                 RAbundVector rav;
 435                 for(int i = 0; i < numBins; i++)
 436                         if(data[i].abundance != 0)
 437                                 rav.push_back(data[i].abundance-1);
 438                 return rav;
 439         }
 440         catch(exception& e) {
 441                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 442                 exit(1);
 443         }
 444         catch(...) {
 445                 cout << "An unknown error has occurred in the SharedRAbundVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 446                 exit(1);
 447         }
 448 }
 449 /***********************************************************************/
 450
 451 SharedSAbundVector SharedRAbundVector::getSharedSAbundVector(){
 452         try {
 453                 SharedSAbundVector sav(maxRank+1);
 454
 455                 for(int i=0;i<data.size();i++){
 456                         int abund = data[i].abundance;
 457                         sav.set(abund, sav.getAbundance(abund) + 1, group);
 458                 }
 459
 460                 sav.set(0, 0, group);
 461                 sav.setLabel(label);
 462                 sav.setGroup(group);
 463
 464                 return sav;
 465         }
 466         catch(exception& e) {
 467                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 468                 exit(1);
 469         }
 470         catch(...) {
 471                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 472                 exit(1);
 473         }
 474 }
 475 /***********************************************************************/
 476
 477 SAbundVector SharedRAbundVector::getSAbundVector() {
 478         try {
 479                 SAbundVector sav(maxRank+1);
 480
 481                 for(int i=0;i<data.size();i++){
 482                         int abund = data[i].abundance;
 483                         sav.set(abund, sav.get(abund) + 1);
 484                 }
 485                 sav.set(0, 0);
 486                 sav.setLabel(label);
 487                 return sav;
 488         }
 489         catch(exception& e) {
 490                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 491                 exit(1);
 492         }
 493         catch(...) {
 494                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 495                 exit(1);
 496         }
 497 }
 498
 499 /***********************************************************************/
 500
 501 SharedOrderVector SharedRAbundVector::getSharedOrderVector() {
 502         try {
 503                 SharedOrderVector ov;
 504
 505                 for(int i=0;i<data.size();i++){
 506                         for(int j=0;j<data[i].abundance;j++){
 507                                 ov.push_back(data[i].bin, data[i].abundance, data[i].group);
 508                         }
 509                 }
 510                 random_shuffle(ov.begin(), ov.end());
 511
 512                 ov.setLabel(label);
 513                 ov.updateStats();
 514
 515                 return ov;
 516         }
 517         catch(exception& e) {
 518                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 519                 exit(1);
 520         }
 521         catch(...) {
 522                 cout << "An unknown error has occurred in the SharedRAbundVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 523                 exit(1);
 524         }
 525 }
 526 /***********************************************************************/
 527
 528 OrderVector SharedRAbundVector::getOrderVector(map<string,int>* nameMap = NULL) {
 529         try {
 530                 OrderVector ov;
 531
 532                 for(int i=0;i<data.size();i++){
 533                         for(int j=0;j<data[i].abundance;j++){
 534                                 ov.push_back(i);
 535                         }
 536                 }
 537                 random_shuffle(ov.begin(), ov.end());
 538
 539                 ov.setLabel(label);
 540                 return ov;
 541         }
 542         catch(exception& e) {
 543                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 544                 exit(1);
 545         }
 546         catch(...) {
 547                 cout << "An unknown error has occurred in the SharedRAbundVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 548                 exit(1);
 549         }
 550 }
 551
 552 /***********************************************************************/
 553