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 SharedRAbundVector::~SharedRAbundVector() {
 157
 158 }
 159
 160 /***********************************************************************/
 161
 162 void SharedRAbundVector::set(int binNumber, int newBinSize, string groupname){
 163         try {
 164                 int oldBinSize = data[binNumber].abundance;
 165                 data[binNumber].abundance = newBinSize;
 166                 data[binNumber].group = groupname;
 167
 168                 if(newBinSize > maxRank)        {       maxRank = newBinSize;   }
 169
 170                 numSeqs += (newBinSize - oldBinSize);
 171         }
 172         catch(exception& e) {
 173                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 174                 exit(1);
 175         }
 176         catch(...) {
 177                 cout << "An unknown error has occurred in the SharedRAbundVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 178                 exit(1);
 179         }
 180 }
 181 /***********************************************************************/
 182
 183 void SharedRAbundVector::setData(vector <individual> newData){
 184         data = newData;
 185 }
 186
 187 /***********************************************************************/
 188
 189 int SharedRAbundVector::getAbundance(int index){
 190         return data[index].abundance;
 191
 192 }
 193 /***********************************************************************/
 194
 195 int SharedRAbundVector::numNZ(){
 196         int sum = 0;
 197         for(int i = 1; i < numBins; i++)
 198                 if(data[i].abundance > 0)
 199                         sum++;
 200         return sum;
 201 }
 202 /***********************************************************************/
 203
 204 void SharedRAbundVector::sortD(){
 205         struct individual indObj;
 206         sort(data.begin()+1, data.end(), indObj);
 207 }
 208 /***********************************************************************/
 209
 210 individual SharedRAbundVector::get(int index){
 211         return data[index];
 212
 213 }
 214 /***********************************************************************/
 215
 216 vector <individual> SharedRAbundVector::getData(){
 217         return data;
 218 }
 219 /***********************************************************************/
 220
 221 void SharedRAbundVector::push_back(int binSize, int otu, string groupName){
 222         try {
 223                 individual newGuy;
 224                 newGuy.abundance = binSize;
 225                 newGuy.group = groupName;
 226                 newGuy.bin = otu;
 227
 228                 data.push_back(newGuy);
 229                 numBins++;
 230
 231                 if(binSize > maxRank){
 232                         maxRank = binSize;
 233                 }
 234
 235                 numSeqs += binSize;
 236         }
 237         catch(exception& e) {
 238                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 239                 exit(1);
 240         }
 241         catch(...) {
 242                 cout << "An unknown error has occurred in the SharedRAbundVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 243                 exit(1);
 244         }
 245 }
 246
 247 /***********************************************************************/
 248
 249 void SharedRAbundVector::insert(int binSize, int otu, string groupName){
 250         try {
 251                 individual newGuy;
 252                 newGuy.abundance = binSize;
 253                 newGuy.group = groupName;
 254                 newGuy.bin = otu;
 255
 256                 data.insert(data.begin()+otu, newGuy);
 257                 numBins++;
 258
 259                 if(binSize > maxRank){
 260                         maxRank = binSize;
 261                 }
 262
 263                 numSeqs += binSize;
 264         }
 265         catch(exception& e) {
 266                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function insert. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 267                 exit(1);
 268         }
 269         catch(...) {
 270                 cout << "An unknown error has occurred in the SharedRAbundVector class function insert. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 271                 exit(1);
 272         }
 273 }
 274
 275 /***********************************************************************/
 276
 277 void SharedRAbundVector::push_front(int binSize, int otu, string groupName){
 278         try {
 279                 individual newGuy;
 280                 newGuy.abundance = binSize;
 281                 newGuy.group = groupName;
 282                 newGuy.bin = otu;
 283
 284                 data.insert(data.begin(), newGuy);
 285                 numBins++;
 286
 287                 if(binSize > maxRank){
 288                         maxRank = binSize;
 289                 }
 290
 291                 numSeqs += binSize;
 292         }
 293         catch(exception& e) {
 294                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function push_front. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 295                 exit(1);
 296         }
 297         catch(...) {
 298                 cout << "An unknown error has occurred in the SharedRAbundVector class function push_front. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 299                 exit(1);
 300         }
 301 }
 302
 303 /***********************************************************************/
 304 void SharedRAbundVector::pop_back(){
 305         numSeqs -= data[data.size()-1].abundance;
 306         numBins--;
 307         return data.pop_back();
 308 }
 309
 310 /***********************************************************************/
 311
 312
 313 vector<individual>::reverse_iterator SharedRAbundVector::rbegin(){
 314         return data.rbegin();
 315 }
 316
 317 /***********************************************************************/
 318
 319 vector<individual>::reverse_iterator SharedRAbundVector::rend(){
 320         return data.rend();
 321 }
 322
 323 /***********************************************************************/
 324 void SharedRAbundVector::resize(int size){
 325
 326         data.resize(size);
 327 }
 328
 329 /***********************************************************************/
 330
 331 int SharedRAbundVector::size(){
 332         return data.size();
 333 }
 334
 335 /***********************************************************************/
 336 void SharedRAbundVector::print(ostream& output){
 337         try {
 338                 output << numBins << '\t';
 339
 340                 for(int i=0;i<data.size();i++){         output << data[i].abundance << '\t';            }
 341                 output << endl;
 342         }
 343         catch(exception& e) {
 344                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 345                 exit(1);
 346         }
 347         catch(...) {
 348                 cout << "An unknown error has occurred in the SharedRAbundVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 349                 exit(1);
 350         }
 351 }
 352 /***********************************************************************/
 353 string SharedRAbundVector::getGroup(){
 354         return group;
 355 }
 356
 357 /***********************************************************************/
 358
 359 void SharedRAbundVector::setGroup(string groupName){
 360         group = groupName;
 361 }
 362 /***********************************************************************/
 363 int SharedRAbundVector::getGroupIndex()  { return index; }
 364 /***********************************************************************/
 365 void SharedRAbundVector::setGroupIndex(int vIndex)      { index = vIndex; }
 366 /***********************************************************************/
 367 int SharedRAbundVector::getNumBins(){
 368         return numBins;
 369 }
 370
 371 /***********************************************************************/
 372
 373 int SharedRAbundVector::getNumSeqs(){
 374         return numSeqs;
 375 }
 376
 377 /***********************************************************************/
 378
 379 int SharedRAbundVector::getMaxRank(){
 380         return maxRank;
 381 }
 382 /***********************************************************************/
 383
 384 SharedRAbundVector SharedRAbundVector::getSharedRAbundVector(){
 385         return *this;
 386 }
 387 /***********************************************************************/
 388 vector<SharedRAbundVector*> SharedRAbundVector::getSharedRAbundVectors(){
 389         try {
 390                 SharedUtil* util;
 391                 util = new SharedUtil();
 392
 393                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups);
 394
 395                 for (int i = 0; i < lookup.size(); i++) {
 396                         //if this sharedrabund is not from a group the user wants then delete it.
 397                         if (util->isValidGroup(lookup[i]->getGroup(), globaldata->Groups) == false) {
 398                                 delete lookup[i];
 399                                 lookup.erase(lookup.begin()+i);
 400                                 i--;
 401                         }
 402                 }
 403
 404                 return lookup;
 405         }
 406         catch(exception& e) {
 407                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSharedRAbundVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 408                 exit(1);
 409         }
 410         catch(...) {
 411                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSharedRAbundVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 412                 exit(1);
 413         }
 414 }
 415 /***********************************************************************/
 416
 417 RAbundVector SharedRAbundVector::getRAbundVector() {
 418         try {
 419                 RAbundVector rav(data.size());
 420
 421                 for (int i = 0; i < data.size(); i++) {
 422                         rav.set(i, data[i].abundance);
 423                 }
 424
 425                 return rav;
 426         }
 427         catch(exception& e) {
 428                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 429                 exit(1);
 430         }
 431         catch(...) {
 432                 cout << "An unknown error has occurred in the SharedRAbundVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 433                 exit(1);
 434         }
 435 }
 436 /***********************************************************************/
 437
 438 RAbundVector SharedRAbundVector::getRAbundVector2() {
 439         try {
 440                 RAbundVector rav;
 441                 for(int i = 0; i < numBins; i++)
 442                         if(data[i].abundance != 0)
 443                                 rav.push_back(data[i].abundance-1);
 444                 return rav;
 445         }
 446         catch(exception& e) {
 447                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 448                 exit(1);
 449         }
 450         catch(...) {
 451                 cout << "An unknown error has occurred in the SharedRAbundVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 452                 exit(1);
 453         }
 454 }
 455 /***********************************************************************/
 456
 457 SharedSAbundVector SharedRAbundVector::getSharedSAbundVector(){
 458         try {
 459                 SharedSAbundVector sav(maxRank+1);
 460
 461                 for(int i=0;i<data.size();i++){
 462                         int abund = data[i].abundance;
 463                         sav.set(abund, sav.getAbundance(abund) + 1, group);
 464                 }
 465
 466                 sav.set(0, 0, group);
 467                 sav.setLabel(label);
 468                 sav.setGroup(group);
 469
 470                 return sav;
 471         }
 472         catch(exception& e) {
 473                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 474                 exit(1);
 475         }
 476         catch(...) {
 477                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 478                 exit(1);
 479         }
 480 }
 481 /***********************************************************************/
 482
 483 SAbundVector SharedRAbundVector::getSAbundVector() {
 484         try {
 485                 SAbundVector sav(maxRank+1);
 486
 487                 for(int i=0;i<data.size();i++){
 488                         int abund = data[i].abundance;
 489                         sav.set(abund, sav.get(abund) + 1);
 490                 }
 491                 sav.set(0, 0);
 492                 sav.setLabel(label);
 493                 return sav;
 494         }
 495         catch(exception& e) {
 496                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 497                 exit(1);
 498         }
 499         catch(...) {
 500                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 501                 exit(1);
 502         }
 503 }
 504
 505 /***********************************************************************/
 506
 507 SharedOrderVector SharedRAbundVector::getSharedOrderVector() {
 508         try {
 509                 SharedOrderVector ov;
 510
 511                 for(int i=0;i<data.size();i++){
 512                         for(int j=0;j<data[i].abundance;j++){
 513                                 ov.push_back(data[i].bin, data[i].abundance, data[i].group);
 514                         }
 515                 }
 516                 random_shuffle(ov.begin(), ov.end());
 517
 518                 ov.setLabel(label);
 519                 ov.updateStats();
 520
 521                 return ov;
 522         }
 523         catch(exception& e) {
 524                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 525                 exit(1);
 526         }
 527         catch(...) {
 528                 cout << "An unknown error has occurred in the SharedRAbundVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 529                 exit(1);
 530         }
 531 }
 532 /***********************************************************************/
 533
 534 OrderVector SharedRAbundVector::getOrderVector(map<string,int>* nameMap = NULL) {
 535         try {
 536                 OrderVector ov;
 537
 538                 for(int i=0;i<data.size();i++){
 539                         for(int j=0;j<data[i].abundance;j++){
 540                                 ov.push_back(i);
 541                         }
 542                 }
 543                 random_shuffle(ov.begin(), ov.end());
 544
 545                 ov.setLabel(label);
 546                 return ov;
 547         }
 548         catch(exception& e) {
 549                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 550                 exit(1);
 551         }
 552         catch(...) {
 553                 cout << "An unknown error has occurred in the SharedRAbundVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 554                 exit(1);
 555         }
 556 }
 557
 558 /***********************************************************************/
 559