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 #include "sharedrabundvector.h"
  11 #include "sabundvector.hpp"
  12 #include "ordervector.hpp"
  13 #include "sharedutilities.h"
  14
  15
  16 /***********************************************************************/
  17 SharedRAbundVector::SharedRAbundVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0) {}
  18 /***********************************************************************/
  19
  20 SharedRAbundVector::~SharedRAbundVector() {
  21         //for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
  22
  23 }
  24
  25 /***********************************************************************/
  26
  27 SharedRAbundVector::SharedRAbundVector(int n) : DataVector(), maxRank(0), numBins(n), numSeqs(0) {
  28                 individual newGuy;
  29                 //initialize data
  30                 for (int i=0; i< n; i++) {
  31                         newGuy.bin = i;
  32                         newGuy.abundance = 0;
  33                         data.push_back(newGuy);
  34                 }
  35 }
  36
  37 /***********************************************************************
  38
  39 SharedRAbundVector::SharedRAbundVector(string id, vector<individual> rav) : DataVector(id), data(rav) {
  40         try {
  41                 numBins = 0;
  42                 maxRank = 0;
  43                 numSeqs = 0;
  44
  45                 for(int i=0;i<data.size();i++){
  46                         if(data[i].abundance != 0)              {       numBins = i+1;          }
  47                         if(data[i].abundance > maxRank) {       maxRank = data[i].abundance;    }
  48                         numSeqs += data[i].abundance;
  49                 }
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "SharedRAbundVector", "SharedRAbundVector");
  53                 exit(1);
  54         }
  55 }
  56
  57
  58 /***********************************************************************/
  59 //reads a shared file
  60 SharedRAbundVector::SharedRAbundVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
  61         try {
  62                 m->namesOfGroups.clear();
  63
  64                 int num, inputData, count;
  65                 count = 0;
  66                 string holdLabel, nextLabel, groupN;
  67                 individual newguy;
  68
  69                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  lookup.clear();
  70
  71                 //are we at the beginning of the file??
  72                 if (m->saveNextLabel == "") {
  73                         f >> label;
  74
  75                         //is this a shared file that has headers
  76                         if (label == "label") {
  77                                 //gets "group"
  78                                 f >> label; m->gobble(f);
  79
  80                                 //gets "numOtus"
  81                                 f >> label; m->gobble(f);
  82
  83                                 //eat rest of line
  84                                 label = m->getline(f); m->gobble(f);
  85
  86                                 //parse labels to save
  87                                 istringstream iStringStream(label);
  88                                 m->binLabelsInFile.clear();
  89                                 while(!iStringStream.eof()){
  90                                         if (m->control_pressed) { break; }
  91                                         string temp;
  92                                         iStringStream >> temp;  m->gobble(iStringStream);
  93
  94                                         m->binLabelsInFile.push_back(temp);
  95                                 }
  96
  97                                 f >> label;
  98                         }
  99                 }else { label = m->saveNextLabel; }
 100
 101                 //reset labels, currentLabels may have gotten changed as otus were eliminated because of group choices or sampling
 102                 m->currentBinLabels = m->binLabelsInFile;
 103
 104                 //read in first row since you know there is at least 1 group.
 105                 f >> groupN >> num;
 106
 107                 holdLabel = label;
 108
 109                 //add new vector to lookup
 110                 SharedRAbundVector* temp = new SharedRAbundVector();
 111                 lookup.push_back(temp);
 112                 lookup[0]->setLabel(label);
 113                 lookup[0]->setGroup(groupN);
 114
 115                 m->namesOfGroups.push_back(groupN);
 116
 117                 //fill vector.  data = first sharedrabund in file
 118                 for(int i=0;i<num;i++){
 119                         f >> inputData;
 120
 121                         lookup[0]->push_back(inputData, groupN); //abundance, bin, group
 122                         push_back(inputData, groupN);
 123
 124                         if (inputData > maxRank) { maxRank = inputData; }
 125                 }
 126
 127                 m->gobble(f);
 128
 129                 if (!(f.eof())) { f >> nextLabel; }
 130
 131                 //read the rest of the groups info in
 132                 while ((nextLabel == holdLabel) && (f.eof() != true)) {
 133                         f >> groupN >> num;
 134                         count++;
 135
 136                         m->namesOfGroups.push_back(groupN);
 137
 138                         //add new vector to lookup
 139                         temp = new SharedRAbundVector();
 140                         lookup.push_back(temp);
 141                         lookup[count]->setLabel(label);
 142                         lookup[count]->setGroup(groupN);
 143
 144                         //fill vector.
 145                         for(int i=0;i<num;i++){
 146                                 f >> inputData;
 147
 148                                 lookup[count]->push_back(inputData, groupN); //abundance, bin, group
 149                         }
 150
 151                         m->gobble(f);
 152
 153                         if (f.eof() != true) { f >> nextLabel; }
 154                 }
 155                         m->saveNextLabel = nextLabel;
 156
 157         }
 158         catch(exception& e) {
 159                 m->errorOut(e, "SharedRAbundVector", "SharedRAbundVector");
 160                 exit(1);
 161         }
 162 }
 163
 164 /***********************************************************************/
 165
 166 void SharedRAbundVector::set(int binNumber, int newBinSize, string groupname){
 167         try {
 168                 int oldBinSize = data[binNumber].abundance;
 169                 data[binNumber].abundance = newBinSize;
 170                 data[binNumber].group = groupname;
 171
 172                 if(newBinSize > maxRank)        {       maxRank = newBinSize;   }
 173
 174                 numSeqs += (newBinSize - oldBinSize);
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "SharedRAbundVector", "set");
 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::clear(){
 222         numBins = 0;
 223         maxRank = 0;
 224         numSeqs = 0;
 225         data.clear();
 226         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 227         lookup.clear();
 228 }
 229 /***********************************************************************/
 230
 231 void SharedRAbundVector::push_back(int binSize, string groupName){
 232         try {
 233                 individual newGuy;
 234                 newGuy.abundance = binSize;
 235                 newGuy.group = groupName;
 236                 newGuy.bin = data.size();
 237
 238                 data.push_back(newGuy);
 239                 numBins++;
 240
 241                 if(binSize > maxRank){
 242                         maxRank = binSize;
 243                 }
 244
 245                 numSeqs += binSize;
 246         }
 247         catch(exception& e) {
 248                 m->errorOut(e, "SharedRAbundVector", "push_back");
 249                 exit(1);
 250         }
 251 }
 252
 253 /***********************************************************************/
 254
 255 void SharedRAbundVector::insert(int binSize, int otu, string groupName){
 256         try {
 257                 individual newGuy;
 258                 newGuy.abundance = binSize;
 259                 newGuy.group = groupName;
 260                 newGuy.bin = otu;
 261
 262                 data.insert(data.begin()+otu, newGuy);
 263                 numBins++;
 264
 265                 if(binSize > maxRank){
 266                         maxRank = binSize;
 267                 }
 268
 269                 numSeqs += binSize;
 270         }
 271         catch(exception& e) {
 272                 m->errorOut(e, "SharedRAbundVector", "insert");
 273                 exit(1);
 274         }
 275 }
 276
 277 /***********************************************************************/
 278
 279 void SharedRAbundVector::push_front(int binSize, int otu, string groupName){
 280         try {
 281                 individual newGuy;
 282                 newGuy.abundance = binSize;
 283                 newGuy.group = groupName;
 284                 newGuy.bin = otu;
 285
 286                 data.insert(data.begin(), newGuy);
 287                 numBins++;
 288
 289                 if(binSize > maxRank){
 290                         maxRank = binSize;
 291                 }
 292
 293                 numSeqs += binSize;
 294         }
 295         catch(exception& e) {
 296                 m->errorOut(e, "SharedRAbundVector", "push_front");
 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 /***********************************************************************/
 335 void SharedRAbundVector::printHeaders(ostream& output){
 336         try {
 337
 338                 output << "label\tGroup\tnumOtus\t";
 339                 if (m->sharedHeaderMode == "tax") {
 340                         for (int i = 0; i < numBins; i++) {
 341
 342                                 //if there is a bin label use it otherwise make one
 343                                 string binLabel = "PhyloType" + toString(i+1);
 344                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
 345
 346                                 output << binLabel << '\t';
 347                         }
 348                         output << endl;
 349                 }else {
 350                         for (int i = 0; i < numBins; i++) {
 351                                 //if there is a bin label use it otherwise make one
 352                                 string mybinLabel = "Otu" + toString(i+1);
 353                                 if (i < m->currentBinLabels.size()) {  mybinLabel = m->currentBinLabels[i]; }
 354
 355                                 output << mybinLabel << '\t';
 356                         }
 357
 358                         output << endl;
 359                 }
 360                 m->printedHeaders = true;
 361         }
 362         catch(exception& e) {
 363                 m->errorOut(e, "SharedRAbundVector", "printHeaders");
 364                 exit(1);
 365         }
 366 }
 367 /***********************************************************************/
 368 void SharedRAbundVector::print(ostream& output) {
 369         try {
 370                 output << numBins << '\t';
 371
 372                 for(int i=0;i<data.size();i++){         output << data[i].abundance << '\t';            }
 373                 output << endl;
 374         }
 375         catch(exception& e) {
 376                 m->errorOut(e, "SharedRAbundVector", "print");
 377                 exit(1);
 378         }
 379 }
 380 /***********************************************************************/
 381 string SharedRAbundVector::getGroup(){
 382         return group;
 383 }
 384
 385 /***********************************************************************/
 386
 387 void SharedRAbundVector::setGroup(string groupName){
 388         group = groupName;
 389 }
 390 /***********************************************************************/
 391 int SharedRAbundVector::getGroupIndex()  { return index; }
 392 /***********************************************************************/
 393 void SharedRAbundVector::setGroupIndex(int vIndex)      { index = vIndex; }
 394 /***********************************************************************/
 395 int SharedRAbundVector::getNumBins(){
 396         return numBins;
 397 }
 398
 399 /***********************************************************************/
 400
 401 int SharedRAbundVector::getNumSeqs(){
 402         return numSeqs;
 403 }
 404
 405 /***********************************************************************/
 406
 407 int SharedRAbundVector::getMaxRank(){
 408         return maxRank;
 409 }
 410 /***********************************************************************/
 411
 412 SharedRAbundVector SharedRAbundVector::getSharedRAbundVector(){
 413         return *this;
 414 }
 415 /***********************************************************************/
 416 vector<SharedRAbundVector*> SharedRAbundVector::getSharedRAbundVectors(){
 417         try {
 418                 SharedUtil* util;
 419                 util = new SharedUtil();
 420
 421                 util->setGroups(m->Groups, m->namesOfGroups);
 422
 423                 bool remove = false;
 424                 for (int i = 0; i < lookup.size(); i++) {
 425                         //if this sharedrabund is not from a group the user wants then delete it.
 426                         if (util->isValidGroup(lookup[i]->getGroup(), m->Groups) == false) {
 427                                 remove = true;
 428                                 delete lookup[i]; lookup[i] = NULL;
 429                                 lookup.erase(lookup.begin()+i);
 430                                 i--;
 431                         }
 432                 }
 433
 434                 delete util;
 435
 436                 if (remove) { eliminateZeroOTUS(lookup); }
 437
 438                 return lookup;
 439         }
 440         catch(exception& e) {
 441                 m->errorOut(e, "SharedRAbundVector", "getSharedRAbundVectors");
 442                 exit(1);
 443         }
 444 }
 445 //**********************************************************************************************************************
 446 int SharedRAbundVector::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
 447                 try {
 448
 449                         vector<SharedRAbundVector*> newLookup;
 450                         for (int i = 0; i < thislookup.size(); i++) {
 451                                 SharedRAbundVector* temp = new SharedRAbundVector();
 452                                 temp->setLabel(thislookup[i]->getLabel());
 453                                 temp->setGroup(thislookup[i]->getGroup());
 454                                 newLookup.push_back(temp);
 455                         }
 456
 457                         //for each bin
 458                         vector<string> newBinLabels;
 459                         for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
 460                                 if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 461
 462                                 //look at each sharedRabund and make sure they are not all zero
 463                                 bool allZero = true;
 464                                 for (int j = 0; j < thislookup.size(); j++) {
 465                                         if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
 466                                 }
 467
 468                                 //if they are not all zero add this bin
 469                                 if (!allZero) {
 470                                         for (int j = 0; j < thislookup.size(); j++) {
 471                                                 newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
 472                                         }
 473
 474                                         //if there is a bin label use it otherwise make one
 475                                         string binLabel = "Otu" + (i+1);
 476                                         if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
 477
 478                                         newBinLabels.push_back(binLabel);
 479                                 }
 480                         }
 481
 482                         for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
 483
 484                         thislookup = newLookup;
 485                         m->currentBinLabels = newBinLabels;
 486
 487                         return 0;
 488
 489                 }
 490                 catch(exception& e) {
 491                         m->errorOut(e, "SharedRAbundVector", "eliminateZeroOTUS");
 492                         exit(1);
 493                 }
 494         }
 495
 496 /***********************************************************************/
 497 vector<SharedRAbundFloatVector*> SharedRAbundVector::getSharedRAbundFloatVectors(vector<SharedRAbundVector*> thislookup){
 498         try {
 499                 vector<SharedRAbundFloatVector*> newLookupFloat;
 500                 for (int i = 0; i < lookup.size(); i++) {
 501                         SharedRAbundFloatVector* temp = new SharedRAbundFloatVector();
 502                         temp->setLabel(thislookup[i]->getLabel());
 503                         temp->setGroup(thislookup[i]->getGroup());
 504                         newLookupFloat.push_back(temp);
 505                 }
 506
 507                 for (int i = 0; i < thislookup.size(); i++) {
 508
 509                         for (int j = 0; j < thislookup[i]->getNumBins(); j++) {
 510
 511                                 if (m->control_pressed) { return newLookupFloat; }
 512
 513                                 int abund = thislookup[i]->getAbundance(j);
 514
 515                                 float relabund = abund / (float) thislookup[i]->getNumSeqs();
 516
 517                                 newLookupFloat[i]->push_back(relabund, thislookup[i]->getGroup());
 518                         }
 519                 }
 520
 521                 return newLookupFloat;
 522         }
 523         catch(exception& e) {
 524                 m->errorOut(e, "SharedRAbundVector", "getSharedRAbundVectors");
 525                 exit(1);
 526         }
 527 }
 528 /***********************************************************************/
 529
 530 RAbundVector SharedRAbundVector::getRAbundVector() {
 531         try {
 532                 RAbundVector rav;
 533
 534                 for (int i = 0; i < data.size(); i++) {
 535                         if(data[i].abundance != 0) {
 536                                 rav.push_back(data[i].abundance);
 537                         }
 538                 }
 539
 540                 rav.setLabel(label);
 541                 return rav;
 542         }
 543         catch(exception& e) {
 544                 m->errorOut(e, "SharedRAbundVector", "getRAbundVector");
 545                 exit(1);
 546         }
 547 }
 548 /***********************************************************************/
 549
 550 RAbundVector SharedRAbundVector::getRAbundVector2() {
 551         try {
 552                 RAbundVector rav;
 553                 for(int i = 0; i < numBins; i++)
 554                         if(data[i].abundance != 0)
 555                                 rav.push_back(data[i].abundance-1);
 556                 return rav;
 557         }
 558         catch(exception& e) {
 559                 m->errorOut(e, "SharedRAbundVector", "getRAbundVector2");
 560                 exit(1);
 561         }
 562 }
 563 /***********************************************************************/
 564
 565 SharedSAbundVector SharedRAbundVector::getSharedSAbundVector(){
 566         try {
 567                 SharedSAbundVector sav(maxRank+1);
 568
 569                 for(int i=0;i<data.size();i++){
 570                         int abund = data[i].abundance;
 571                         sav.set(abund, sav.getAbundance(abund) + 1, group);
 572                 }
 573
 574                 sav.set(0, 0, group);
 575                 sav.setLabel(label);
 576                 sav.setGroup(group);
 577
 578                 return sav;
 579         }
 580         catch(exception& e) {
 581                 m->errorOut(e, "SharedRAbundVector", "getSharedSAbundVector");
 582                 exit(1);
 583         }
 584 }
 585 /***********************************************************************/
 586
 587 SAbundVector SharedRAbundVector::getSAbundVector() {
 588         try {
 589                 SAbundVector sav(maxRank+1);
 590
 591                 for(int i=0;i<data.size();i++){
 592                         int abund = data[i].abundance;
 593                         sav.set(abund, sav.get(abund) + 1);
 594                 }
 595                 sav.set(0, 0);
 596                 sav.setLabel(label);
 597                 return sav;
 598         }
 599         catch(exception& e) {
 600                 m->errorOut(e, "SharedRAbundVector", "getSAbundVector");
 601                 exit(1);
 602         }
 603 }
 604
 605 /***********************************************************************/
 606
 607 SharedOrderVector SharedRAbundVector::getSharedOrderVector() {
 608         try {
 609                 SharedOrderVector ov;
 610
 611                 for(int i=0;i<data.size();i++){
 612                         for(int j=0;j<data[i].abundance;j++){
 613                                 ov.push_back(data[i].bin, data[i].abundance, data[i].group);
 614                         }
 615                 }
 616                 random_shuffle(ov.begin(), ov.end());
 617
 618                 ov.setLabel(label);
 619                 ov.updateStats();
 620
 621                 return ov;
 622         }
 623         catch(exception& e) {
 624                 m->errorOut(e, "SharedRAbundVector", "getSharedOrderVector");
 625                 exit(1);
 626         }
 627 }
 628 /***********************************************************************/
 629
 630 OrderVector SharedRAbundVector::getOrderVector(map<string,int>* nameMap = NULL) {
 631         try {
 632                 OrderVector ov;
 633                 for(int i=0;i<numBins;i++){
 634                         for(int j=0;j<data[i].abundance;j++){
 635                                 ov.push_back(i);
 636                         }
 637                 }
 638                 random_shuffle(ov.begin(), ov.end());
 639
 640                 ov.setLabel(label);
 641
 642                 return ov;
 643         }
 644         catch(exception& e) {
 645                 m->errorOut(e, "SharedRAbundVector", "getOrderVector");
 646                 exit(1);
 647         }
 648 }
 649
 650 /***********************************************************************/
 651