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