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                 lookup.push_back(new SharedRAbundVector(num));
  82                 lookup[0]->setLabel(label);
  83                 lookup[0]->setGroup(groupN);
  84
  85                 if (globaldata->gGroupmap == NULL) {
  86                         //save group in groupmap
  87                         groupmap->namesOfGroups.push_back(groupN);
  88                         groupmap->groupIndex[groupN] = 0;
  89                 }
  90
  91                 //fill vector.  data = first sharedrabund in file
  92                 for(int i=0;i<num;i++){
  93                         f >> inputData;
  94
  95                         lookup[0]->push_back(inputData, i, groupN); //abundance, bin, group
  96                         push_back(inputData, i, groupN);
  97                         numSeqs += inputData;
  98                         numBins++;
  99                         if (inputData > maxRank) { maxRank = inputData; }
 100
 101                 }
 102
 103                 //save position in file in case next line is a new label.
 104                 pos = f.tellg();
 105
 106                 if (f.eof() != true) { f >> nextLabel; }
 107
 108                 //read the rest of the groups info in
 109                 while ((nextLabel == holdLabel) && (f.eof() != true)) {
 110                         f >> groupN >> num;
 111                         count++;
 112
 113                         if (globaldata->gGroupmap == NULL) {
 114                                 //save group in groupmap
 115                                 groupmap->namesOfGroups.push_back(groupN);
 116                                 groupmap->groupIndex[groupN] = count;
 117                         }
 118
 119                         //add new vector to lookup
 120                         lookup.push_back(new SharedRAbundVector(num));
 121                         lookup[count]->setLabel(label);
 122                         lookup[count]->setGroup(groupN);
 123
 124                         //fill vector.
 125                         for(int i=0;i<num;i++){
 126                                 f >> inputData;
 127                                 lookup[count]->push_back(inputData, i, groupN); //abundance, bin, group
 128                         }
 129
 130                         //save position in file in case next line is a new label.
 131                         pos = f.tellg();
 132
 133                         if (f.eof() != true) { f >> nextLabel; }
 134                 }
 135
 136                 //put file pointer back since you are now at a new distance label
 137                 f.seekg(pos, ios::beg);
 138
 139                 if (globaldata->gGroupmap == NULL) { globaldata->gGroupmap = groupmap; }
 140
 141         }
 142         catch(exception& e) {
 143                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function SharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 144                 exit(1);
 145         }
 146         catch(...) {
 147                 cout << "An unknown error has occurred in the SharedRAbundVector class function SharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 148                 exit(1);
 149         }
 150 }
 151
 152 /***********************************************************************/
 153
 154 SharedRAbundVector::~SharedRAbundVector() {
 155
 156 }
 157
 158 /***********************************************************************/
 159
 160 void SharedRAbundVector::set(int binNumber, int newBinSize, string groupname){
 161         try {
 162                 int oldBinSize = data[binNumber].abundance;
 163                 data[binNumber].abundance = newBinSize;
 164                 data[binNumber].group = groupname;
 165
 166                 if(newBinSize > maxRank)        {       maxRank = newBinSize;   }
 167
 168                 numSeqs += (newBinSize - oldBinSize);
 169         }
 170         catch(exception& e) {
 171                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 172                 exit(1);
 173         }
 174         catch(...) {
 175                 cout << "An unknown error has occurred in the SharedRAbundVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 176                 exit(1);
 177         }
 178 }
 179 /***********************************************************************/
 180
 181 void SharedRAbundVector::setData(vector <individual> newData){
 182         data = newData;
 183 }
 184
 185 /***********************************************************************/
 186
 187 int SharedRAbundVector::getAbundance(int index){
 188         return data[index].abundance;
 189
 190 }
 191 /***********************************************************************/
 192
 193 int SharedRAbundVector::numNZ(){
 194         int sum = 0;
 195         for(int i = 1; i < numBins; i++)
 196                 if(data[i].abundance > 0)
 197                         sum++;
 198         return sum;
 199 }
 200 /***********************************************************************/
 201
 202 void SharedRAbundVector::sortD(){
 203         struct individual indObj;
 204         sort(data.begin()+1, data.end(), indObj);
 205 }
 206 /***********************************************************************/
 207
 208 individual SharedRAbundVector::get(int index){
 209         return data[index];
 210
 211 }
 212 /***********************************************************************/
 213
 214 vector <individual> SharedRAbundVector::getData(){
 215         return data;
 216 }
 217 /***********************************************************************/
 218
 219 void SharedRAbundVector::push_back(int binSize, int otu, string groupName){
 220         try {
 221                 individual newGuy;
 222                 newGuy.abundance = binSize;
 223                 newGuy.group = groupName;
 224                 newGuy.bin = otu;
 225
 226                 data.push_back(newGuy);
 227                 numBins++;
 228
 229                 if(binSize > maxRank){
 230                         maxRank = binSize;
 231                 }
 232
 233                 numSeqs += binSize;
 234         }
 235         catch(exception& e) {
 236                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 237                 exit(1);
 238         }
 239         catch(...) {
 240                 cout << "An unknown error has occurred in the SharedRAbundVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 241                 exit(1);
 242         }
 243 }
 244
 245 /***********************************************************************/
 246
 247 void SharedRAbundVector::insert(int binSize, int otu, string groupName){
 248         try {
 249                 individual newGuy;
 250                 newGuy.abundance = binSize;
 251                 newGuy.group = groupName;
 252                 newGuy.bin = otu;
 253
 254                 data.insert(data.begin()+otu, newGuy);
 255                 numBins++;
 256
 257                 if(binSize > maxRank){
 258                         maxRank = binSize;
 259                 }
 260
 261                 numSeqs += binSize;
 262         }
 263         catch(exception& e) {
 264                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function insert. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 265                 exit(1);
 266         }
 267         catch(...) {
 268                 cout << "An unknown error has occurred in the SharedRAbundVector class function insert. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 269                 exit(1);
 270         }
 271 }
 272
 273 /***********************************************************************/
 274
 275 void SharedRAbundVector::push_front(int binSize, int otu, string groupName){
 276         try {
 277                 individual newGuy;
 278                 newGuy.abundance = binSize;
 279                 newGuy.group = groupName;
 280                 newGuy.bin = otu;
 281
 282                 data.insert(data.begin(), newGuy);
 283                 numBins++;
 284
 285                 if(binSize > maxRank){
 286                         maxRank = binSize;
 287                 }
 288
 289                 numSeqs += binSize;
 290         }
 291         catch(exception& e) {
 292                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function push_front. 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 SharedRAbundVector class function push_front. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 297                 exit(1);
 298         }
 299 }
 300
 301 /***********************************************************************/
 302 void SharedRAbundVector::pop_back(){
 303         numSeqs -= data[data.size()-1].abundance;
 304         numBins--;
 305         return data.pop_back();
 306 }
 307
 308 /***********************************************************************/
 309
 310
 311 vector<individual>::reverse_iterator SharedRAbundVector::rbegin(){
 312         return data.rbegin();
 313 }
 314
 315 /***********************************************************************/
 316
 317 vector<individual>::reverse_iterator SharedRAbundVector::rend(){
 318         return data.rend();
 319 }
 320
 321 /***********************************************************************/
 322 void SharedRAbundVector::resize(int size){
 323
 324         data.resize(size);
 325 }
 326
 327 /***********************************************************************/
 328
 329 int SharedRAbundVector::size(){
 330         return data.size();
 331 }
 332
 333 /***********************************************************************/
 334 void SharedRAbundVector::print(ostream& output){
 335         try {
 336                 output << numBins << '\t';
 337
 338                 for(int i=0;i<data.size();i++){         output << data[i].abundance << '\t';            }
 339                 output << endl;
 340         }
 341         catch(exception& e) {
 342                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 343                 exit(1);
 344         }
 345         catch(...) {
 346                 cout << "An unknown error has occurred in the SharedRAbundVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 347                 exit(1);
 348         }
 349 }
 350 /***********************************************************************/
 351 string SharedRAbundVector::getGroup(){
 352         return group;
 353 }
 354
 355 /***********************************************************************/
 356
 357 void SharedRAbundVector::setGroup(string groupName){
 358         group = groupName;
 359 }
 360 /***********************************************************************/
 361 int SharedRAbundVector::getGroupIndex()  { return index; }
 362 /***********************************************************************/
 363 void SharedRAbundVector::setGroupIndex(int vIndex)      { index = vIndex; }
 364 /***********************************************************************/
 365 int SharedRAbundVector::getNumBins(){
 366         return numBins;
 367 }
 368
 369 /***********************************************************************/
 370
 371 int SharedRAbundVector::getNumSeqs(){
 372         return numSeqs;
 373 }
 374
 375 /***********************************************************************/
 376
 377 int SharedRAbundVector::getMaxRank(){
 378         return maxRank;
 379 }
 380 /***********************************************************************/
 381
 382 SharedRAbundVector SharedRAbundVector::getSharedRAbundVector(){
 383         return *this;
 384 }
 385 /***********************************************************************/
 386 vector<SharedRAbundVector*> SharedRAbundVector::getSharedRAbundVectors(){
 387         try {
 388                 SharedUtil* util;
 389                 util = new SharedUtil();
 390
 391                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups);
 392
 393                 for (int i = 0; i < lookup.size(); i++) {
 394                         //if this sharedrabund is not from a group the user wants then delete it.
 395                         if (util->isValidGroup(lookup[i]->getGroup(), globaldata->Groups) == false) {
 396                                 delete lookup[i];
 397                                 lookup.erase(lookup.begin()+i);
 398                                 i--;
 399                         }
 400                 }
 401
 402                 return lookup;
 403         }
 404         catch(exception& e) {
 405                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSharedRAbundVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 406                 exit(1);
 407         }
 408         catch(...) {
 409                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSharedRAbundVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 410                 exit(1);
 411         }
 412 }
 413 /***********************************************************************/
 414
 415 RAbundVector SharedRAbundVector::getRAbundVector() {
 416         try {
 417                 RAbundVector rav(data.size());
 418
 419                 for (int i = 0; i < data.size(); i++) {
 420                         rav.set(i, data[i].abundance);
 421                 }
 422
 423                 return rav;
 424         }
 425         catch(exception& e) {
 426                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 427                 exit(1);
 428         }
 429         catch(...) {
 430                 cout << "An unknown error has occurred in the SharedRAbundVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 431                 exit(1);
 432         }
 433 }
 434 /***********************************************************************/
 435
 436 RAbundVector SharedRAbundVector::getRAbundVector2() {
 437         try {
 438                 RAbundVector rav;
 439                 for(int i = 0; i < numBins; i++)
 440                         if(data[i].abundance != 0)
 441                                 rav.push_back(data[i].abundance-1);
 442                 return rav;
 443         }
 444         catch(exception& e) {
 445                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 446                 exit(1);
 447         }
 448         catch(...) {
 449                 cout << "An unknown error has occurred in the SharedRAbundVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 450                 exit(1);
 451         }
 452 }
 453 /***********************************************************************/
 454
 455 SharedSAbundVector SharedRAbundVector::getSharedSAbundVector(){
 456         try {
 457                 SharedSAbundVector sav(maxRank+1);
 458
 459                 for(int i=0;i<data.size();i++){
 460                         int abund = data[i].abundance;
 461                         sav.set(abund, sav.getAbundance(abund) + 1, group);
 462                 }
 463
 464                 sav.set(0, 0, group);
 465                 sav.setLabel(label);
 466                 sav.setGroup(group);
 467
 468                 return sav;
 469         }
 470         catch(exception& e) {
 471                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 472                 exit(1);
 473         }
 474         catch(...) {
 475                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 476                 exit(1);
 477         }
 478 }
 479 /***********************************************************************/
 480
 481 SAbundVector SharedRAbundVector::getSAbundVector() {
 482         try {
 483                 SAbundVector sav(maxRank+1);
 484
 485                 for(int i=0;i<data.size();i++){
 486                         int abund = data[i].abundance;
 487                         sav.set(abund, sav.get(abund) + 1);
 488                 }
 489                 sav.set(0, 0);
 490                 sav.setLabel(label);
 491                 return sav;
 492         }
 493         catch(exception& e) {
 494                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 495                 exit(1);
 496         }
 497         catch(...) {
 498                 cout << "An unknown error has occurred in the SharedRAbundVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 499                 exit(1);
 500         }
 501 }
 502
 503 /***********************************************************************/
 504
 505 SharedOrderVector SharedRAbundVector::getSharedOrderVector() {
 506         try {
 507                 SharedOrderVector ov;
 508
 509                 for(int i=0;i<data.size();i++){
 510                         for(int j=0;j<data[i].abundance;j++){
 511                                 ov.push_back(data[i].bin, data[i].abundance, data[i].group);
 512                         }
 513                 }
 514                 random_shuffle(ov.begin(), ov.end());
 515
 516                 ov.setLabel(label);
 517                 return ov;
 518         }
 519         catch(exception& e) {
 520                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 521                 exit(1);
 522         }
 523         catch(...) {
 524                 cout << "An unknown error has occurred in the SharedRAbundVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 525                 exit(1);
 526         }
 527 }
 528 /***********************************************************************/
 529
 530 OrderVector SharedRAbundVector::getOrderVector(map<string,int>* nameMap = NULL) {
 531         try {
 532                 OrderVector ov;
 533
 534                 for(int i=0;i<data.size();i++){
 535                         for(int j=0;j<data[i].abundance;j++){
 536                                 ov.push_back(i);
 537                         }
 538                 }
 539                 random_shuffle(ov.begin(), ov.end());
 540
 541                 ov.setLabel(label);
 542                 return ov;
 543         }
 544         catch(exception& e) {
 545                 cout << "Standard Error: " << e.what() << " has occurred in the SharedRAbundVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 546                 exit(1);
 547         }
 548         catch(...) {
 549                 cout << "An unknown error has occurred in the SharedRAbundVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 550                 exit(1);
 551         }
 552 }
 553
 554 /***********************************************************************/
 555