sharedrabundfloatvector.cpp

   1 /*
   2  *  sharedrabundfloatvector.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 8/18/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "sharedrabundfloatvector.h"
  11 #include "sharedutilities.h"
  12
  13 /***********************************************************************/
  14
  15 SharedRAbundFloatVector::SharedRAbundFloatVector() : DataVector(), maxRank(0.0), numBins(0), numSeqs(0.0) {}
  16 /***********************************************************************/
  17
  18 SharedRAbundFloatVector::~SharedRAbundFloatVector() {}
  19
  20 /***********************************************************************/
  21 SharedRAbundFloatVector::SharedRAbundFloatVector(int n) : DataVector(), maxRank(0.0), numBins(n), numSeqs(0.0) {
  22
  23                 individualFloat newGuy;
  24                 //initialize data
  25                 for (int i=0; i< n; i++) {
  26                         newGuy.bin = i;
  27                         newGuy.abundance = 0.0;
  28                         data.push_back(newGuy);
  29                 }
  30 }
  31 /***********************************************************************/
  32 //reads a shared file
  33 SharedRAbundFloatVector::SharedRAbundFloatVector(ifstream& f) : DataVector(), maxRank(0.0), numBins(0), numSeqs(0.0) {
  34         try {
  35
  36                 m->clearAllGroups();
  37                 vector<string> allGroups;
  38
  39                 int num, count;
  40                 float inputData;
  41                 count = 0;
  42                 string holdLabel, nextLabel, groupN;
  43                 individualFloat newguy;
  44
  45                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
  46                 lookup.clear();
  47
  48                 //are we at the beginning of the file??
  49                 if (m->saveNextLabel == "") {
  50                         f >> label;
  51
  52                         //is this a shared file that has headers
  53                         if (label == "label") {
  54                                 //gets "group"
  55                                 f >> label; m->gobble(f);
  56
  57                                 //gets "numOtus"
  58                                 f >> label; m->gobble(f);
  59
  60                                 //eat rest of line
  61                                 label = m->getline(f); m->gobble(f);
  62
  63                                 //parse labels to save
  64                                 istringstream iStringStream(label);
  65                                 m->binLabelsInFile.clear();
  66                                 while(!iStringStream.eof()){
  67                                         if (m->control_pressed) { break; }
  68                                         string temp;
  69                                         iStringStream >> temp;  m->gobble(iStringStream);
  70
  71                                         m->binLabelsInFile.push_back(temp);
  72                                 }
  73
  74                                 f >> label;
  75                         }
  76                 }else { label = m->saveNextLabel; }
  77
  78                 //reset labels, currentLabels may have gotten changed as otus were eliminated because of group choices or sampling
  79                 m->currentBinLabels = m-> binLabelsInFile;
  80
  81                 //read in first row since you know there is at least 1 group.
  82                 f >> groupN >> num;
  83
  84                 holdLabel = label;
  85
  86                 //add new vector to lookup
  87                 SharedRAbundFloatVector* temp = new SharedRAbundFloatVector();
  88                 lookup.push_back(temp);
  89                 lookup[0]->setLabel(label);
  90                 lookup[0]->setGroup(groupN);
  91
  92                 allGroups.push_back(groupN);
  93
  94                 //fill vector.  data = first sharedrabund in file
  95                 for(int i=0;i<num;i++){
  96                         f >> inputData;
  97
  98                         lookup[0]->push_back(inputData, groupN); //abundance, bin, group
  99                         push_back(inputData, groupN);
 100
 101                         if (inputData > maxRank) { maxRank = inputData; }
 102                 }
 103
 104                 m->gobble(f);
 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                         allGroups.push_back(groupN);
 114
 115                         //add new vector to lookup
 116                         temp = new SharedRAbundFloatVector();
 117                         lookup.push_back(temp);
 118                         lookup[count]->setLabel(label);
 119                         lookup[count]->setGroup(groupN);
 120
 121                         //fill vector.
 122                         for(int i=0;i<num;i++){
 123                                 f >> inputData;
 124                                 lookup[count]->push_back(inputData, groupN); //abundance, bin, group
 125                         }
 126
 127                         m->gobble(f);
 128
 129                         if (f.eof() != true) { f >> nextLabel; }
 130                 }
 131
 132                 m->saveNextLabel = nextLabel;
 133                 m->setAllGroups(allGroups);
 134
 135         }
 136         catch(exception& e) {
 137                 m->errorOut(e, "SharedRAbundFloatVector", "SharedRAbundFloatVector");
 138                 exit(1);
 139         }
 140 }
 141
 142 /***********************************************************************/
 143
 144 void SharedRAbundFloatVector::set(int binNumber, float newBinSize, string groupname){
 145         try {
 146                 float oldBinSize = data[binNumber].abundance;
 147                 data[binNumber].abundance = newBinSize;
 148                 data[binNumber].group = groupname;
 149
 150                 if(newBinSize > maxRank)        {       newBinSize = newBinSize;        }
 151
 152                 numSeqs += (newBinSize - oldBinSize);
 153         }
 154         catch(exception& e) {
 155                 m->errorOut(e, "SharedRAbundFloatVector", "set");
 156                 exit(1);
 157         }
 158 }
 159 /***********************************************************************/
 160
 161 void SharedRAbundFloatVector::clear(){
 162         numBins = 0;
 163         maxRank = 0;
 164         numSeqs = 0;
 165         data.clear();
 166         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 167         lookup.clear();
 168 }
 169 /***********************************************************************/
 170 float SharedRAbundFloatVector::getAbundance(int index){
 171         return data[index].abundance;
 172 }
 173 /***********************************************************************/
 174 individualFloat SharedRAbundFloatVector::get(int index){
 175         return data[index];
 176 }
 177 /***********************************************************************/
 178 void SharedRAbundFloatVector::push_back(float binSize, string groupName){
 179         try {
 180                 individualFloat newGuy;
 181                 newGuy.abundance = binSize;
 182                 newGuy.group = groupName;
 183                 newGuy.bin = data.size();
 184
 185                 data.push_back(newGuy);
 186                 numBins++;
 187
 188                 if(binSize > maxRank){  maxRank = binSize;      }
 189
 190                 numSeqs += binSize;
 191         }
 192         catch(exception& e) {
 193                 m->errorOut(e, "SharedRAbundFloatVector", "push_back");
 194                 exit(1);
 195         }
 196 }
 197 /***********************************************************************/
 198 void SharedRAbundFloatVector::insert(float binSize, int otu, string groupName){
 199         try {
 200                 individualFloat newGuy;
 201                 newGuy.abundance = binSize;
 202                 newGuy.group = groupName;
 203                 newGuy.bin = otu;
 204
 205                 data.insert(data.begin()+otu, newGuy);
 206                 numBins++;
 207
 208                 if(binSize > maxRank){  maxRank = binSize;      }
 209
 210                 numSeqs += binSize;
 211         }
 212         catch(exception& e) {
 213                 m->errorOut(e, "SharedRAbundFloatVector", "insert");
 214                 exit(1);
 215         }
 216 }
 217
 218 /***********************************************************************/
 219 void SharedRAbundFloatVector::push_front(float binSize, int otu, string groupName){
 220         try {
 221                 individualFloat newGuy;
 222                 newGuy.abundance = binSize;
 223                 newGuy.group = groupName;
 224                 newGuy.bin = otu;
 225
 226                 data.insert(data.begin(), newGuy);
 227                 numBins++;
 228
 229                 if(binSize > maxRank){  maxRank = binSize;      }
 230
 231                 numSeqs += binSize;
 232         }
 233         catch(exception& e) {
 234                 m->errorOut(e, "SharedRAbundFloatVector", "push_front");
 235                 exit(1);
 236         }
 237 }
 238 /**********************************************************************/
 239 void SharedRAbundFloatVector::pop_back(){
 240         numSeqs -= data[data.size()-1].abundance;
 241         numBins--;
 242         data.pop_back();
 243 }
 244 /***********************************************************************/
 245 void SharedRAbundFloatVector::resize(int size){
 246         data.resize(size);
 247 }
 248 /**********************************************************************/
 249 int SharedRAbundFloatVector::size(){
 250         return data.size();
 251 }
 252 /***********************************************************************/
 253 void SharedRAbundFloatVector::printHeaders(ostream& output){
 254         try {
 255                 string snumBins = toString(numBins);
 256                 output << "label\tGroup\tnumOtus\t";
 257                 if (m->sharedHeaderMode == "tax") {
 258                         for (int i = 0; i < numBins; i++) {
 259
 260                                 //if there is a bin label use it otherwise make one
 261                                 string binLabel = "PhyloType";
 262                                 string sbinNumber = toString(i+1);
 263                                 if (sbinNumber.length() < snumBins.length()) {
 264                                         int diff = snumBins.length() - sbinNumber.length();
 265                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
 266                                 }
 267                                 binLabel += sbinNumber;
 268                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
 269
 270                                 output << binLabel << '\t';
 271                         }
 272                         output << endl;
 273                 }else {
 274                         for (int i = 0; i < numBins; i++) {
 275                                 //if there is a bin label use it otherwise make one
 276                                 string binLabel = "Otu";
 277                                 string sbinNumber = toString(i+1);
 278                                 if (sbinNumber.length() < snumBins.length()) {
 279                                         int diff = snumBins.length() - sbinNumber.length();
 280                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
 281                                 }
 282                                 binLabel += sbinNumber;
 283                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
 284
 285                                 output << binLabel << '\t';
 286                         }
 287
 288                         output << endl;
 289                 }
 290
 291                 m->printedHeaders = true;
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "SharedRAbundVector", "printHeaders");
 295                 exit(1);
 296         }
 297 }
 298 /***********************************************************************/
 299 void SharedRAbundFloatVector::print(ostream& output){
 300         try {
 301                 output << numBins << '\t';
 302
 303                 for(int i=0;i<data.size();i++){         output << data[i].abundance << '\t';            }
 304                 output << endl;
 305         }
 306         catch(exception& e) {
 307                 m->errorOut(e, "SharedRAbundFloatVector", "print");
 308                 exit(1);
 309         }
 310 }
 311 /***********************************************************************/
 312 string SharedRAbundFloatVector::getGroup(){
 313         return group;
 314 }
 315 /***********************************************************************/
 316 void SharedRAbundFloatVector::setGroup(string groupName){
 317         group = groupName;
 318 }
 319 /***********************************************************************/
 320 int SharedRAbundFloatVector::getGroupIndex()  { return index; }
 321 /***********************************************************************/
 322 void SharedRAbundFloatVector::setGroupIndex(int vIndex) { index = vIndex; }
 323 /***********************************************************************/
 324 int SharedRAbundFloatVector::getNumBins(){      return numBins; }
 325 /***********************************************************************/
 326 float SharedRAbundFloatVector::getNumSeqs(){    return numSeqs; }
 327 /***********************************************************************/
 328 float SharedRAbundFloatVector::getMaxRank(){    return maxRank; }
 329 /***********************************************************************/
 330 SharedRAbundFloatVector SharedRAbundFloatVector::getSharedRAbundFloatVector(){
 331         return *this;
 332 }
 333 /***********************************************************************
 334 SharedRAbundVector SharedRAbundFloatVector::getSharedRAbundVector(){
 335         try {
 336                 SharedRAbundVector rav(numBins);
 337                 rav.setLabel(label);
 338                 rav.setGroup(group);
 339
 340                 for (int i = 0; i < data.size(); i++) {
 341
 342                         rav.push_back(data[i].abundance);
 343                 }
 344
 345         }
 346         catch(exception& e) {
 347                 m->errorOut(e, "SharedRAbundFloatVector", "getSharedRAbundVector");
 348                 exit(1);
 349         }
 350 }
 351 /***********************************************************************/
 352 vector<SharedRAbundFloatVector*> SharedRAbundFloatVector::getSharedRAbundFloatVectors(){
 353         try {
 354                 SharedUtil* util;
 355                 util = new SharedUtil();
 356
 357                 vector<string> Groups = m->getGroups();
 358                 vector<string> allGroups = m->getAllGroups();
 359                 util->setGroups(Groups, allGroups);
 360                 m->setGroups(Groups);
 361
 362                 bool remove = false;
 363                 for (int i = 0; i < lookup.size(); i++) {
 364                         //if this sharedrabund is not from a group the user wants then delete it.
 365                         if (util->isValidGroup(lookup[i]->getGroup(), m->getGroups()) == false) {
 366                                 delete lookup[i]; lookup[i] = NULL;
 367                                 lookup.erase(lookup.begin()+i);
 368                                 i--;
 369                                 remove = true;
 370                         }
 371                 }
 372
 373                 delete util;
 374
 375                 if (remove) { eliminateZeroOTUS(lookup); }
 376
 377                 return lookup;
 378         }
 379         catch(exception& e) {
 380                 m->errorOut(e, "SharedRAbundFloatVector", "getSharedRAbundFloatVectors");
 381                 exit(1);
 382         }
 383 }
 384 /***********************************************************************/
 385
 386 RAbundVector SharedRAbundFloatVector::getRAbundVector() {
 387         try {
 388                 RAbundVector rav(numBins);
 389
 390                 //this is not functional, not sure how to handle it yet, but I need the stub because it is a pure function
 391
 392                 rav.setLabel(label);
 393                 return rav;
 394         }
 395         catch(exception& e) {
 396                 m->errorOut(e, "SharedRAbundFloatVector", "getRAbundVector");
 397                 exit(1);
 398         }
 399 }
 400 /***********************************************************************
 401
 402 SharedSAbundVector SharedRAbundVector::getSharedSAbundVector(){
 403         try {
 404                 SharedSAbundVector sav(maxRank+1);
 405
 406                 for(int i=0;i<data.size();i++){
 407                         int abund = data[i].abundance;
 408                         sav.set(abund, sav.getAbundance(abund) + 1, group);
 409                 }
 410
 411                 sav.set(0, 0, group);
 412                 sav.setLabel(label);
 413                 sav.setGroup(group);
 414
 415                 return sav;
 416         }
 417         catch(exception& e) {
 418                 m->errorOut(e, "SharedRAbundVector", "getSharedSAbundVector");
 419                 exit(1);
 420         }
 421 }
 422 /***********************************************************************/
 423
 424 SAbundVector SharedRAbundFloatVector::getSAbundVector() {
 425         try {
 426                 SAbundVector sav(ceil(maxRank)+1);
 427
 428                 //this is not functional, not sure how to handle it yet, but I need the stub because it is a pure function
 429
 430                 sav.set(0, 0);
 431                 sav.setLabel(label);
 432                 return sav;
 433         }
 434         catch(exception& e) {
 435                 m->errorOut(e, "SharedRAbundFloatVector", "getSAbundVector");
 436                 exit(1);
 437         }
 438 }
 439
 440 /***********************************************************************
 441
 442 SharedOrderVector SharedRAbundFloatVector::getSharedOrderVector() {
 443         try {
 444                 SharedOrderVector ov;
 445
 446                 for(int i=0;i<data.size();i++){
 447                         int round = ceil(data[i].abundance);
 448                         for(int j=0;j<round;j++){
 449                                 ov.push_back(data[i].bin, round, data[i].group);
 450                         }
 451                 }
 452                 random_shuffle(ov.begin(), ov.end());
 453
 454                 ov.setLabel(label);
 455                 ov.updateStats();
 456
 457                 return ov;
 458         }
 459         catch(exception& e) {
 460                 m->errorOut(e, "SharedRAbundFloatVector", "getSharedOrderVector");
 461                 exit(1);
 462         }
 463 }
 464 /***********************************************************************/
 465 //this is not functional, not sure how to handle it yet, but I need the stub because it is a pure function
 466 OrderVector SharedRAbundFloatVector::getOrderVector(map<string,int>* nameMap = NULL) {
 467         try {
 468                 OrderVector ov;
 469
 470                 for(int i=0;i<data.size();i++){
 471                         int round = ceil(data[i].abundance);
 472                         for(int j=0;j<round;j++){
 473                                 ov.push_back(i);
 474                         }
 475                 }
 476                 random_shuffle(ov.begin(), ov.end());
 477
 478                 ov.setLabel(label);
 479                 return ov;
 480         }
 481         catch(exception& e) {
 482                 m->errorOut(e, "SharedRAbundFloatVector", "getOrderVector");
 483                 exit(1);
 484         }
 485 }
 486 //**********************************************************************************************************************
 487 int SharedRAbundFloatVector::eliminateZeroOTUS(vector<SharedRAbundFloatVector*>& thislookup) {
 488         try {
 489
 490                 vector<SharedRAbundFloatVector*> newLookup;
 491                 for (int i = 0; i < thislookup.size(); i++) {
 492                         SharedRAbundFloatVector* temp = new SharedRAbundFloatVector();
 493                         temp->setLabel(thislookup[i]->getLabel());
 494                         temp->setGroup(thislookup[i]->getGroup());
 495                         newLookup.push_back(temp);
 496                 }
 497
 498                 //for each bin
 499                 vector<string> newBinLabels;
 500                 string snumBins = toString(thislookup[0]->getNumBins());
 501                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
 502                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 503
 504                         //look at each sharedRabund and make sure they are not all zero
 505                         bool allZero = true;
 506                         for (int j = 0; j < thislookup.size(); j++) {
 507                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
 508                         }
 509
 510                         //if they are not all zero add this bin
 511                         if (!allZero) {
 512                                 for (int j = 0; j < thislookup.size(); j++) {
 513                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
 514                                 }
 515                                 //if there is a bin label use it otherwise make one
 516                                 string binLabel = "Otu";
 517                                 string sbinNumber = toString(i+1);
 518                                 if (sbinNumber.length() < snumBins.length()) {
 519                                         int diff = snumBins.length() - sbinNumber.length();
 520                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
 521                                 }
 522                                 binLabel += sbinNumber;
 523                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
 524
 525                                 newBinLabels.push_back(binLabel);
 526                         }
 527                 }
 528
 529                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
 530
 531                 thislookup = newLookup;
 532                 m->currentBinLabels = newBinLabels;
 533
 534                 return 0;
 535
 536         }
 537         catch(exception& e) {
 538                 m->errorOut(e, "SharedRAbundFloatVector", "eliminateZeroOTUS");
 539                 exit(1);
 540         }
 541 }
 542 /***********************************************************************/
 543