sharedlistvector.cpp

   1 /*
   2  *  sharedSharedListVector.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/22/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10
  11 using namespace std;
  12
  13
  14 #include "sabundvector.hpp"
  15 #include "rabundvector.hpp"
  16 #include "ordervector.hpp"
  17 #include "sharedlistvector.h"
  18 #include "sharedordervector.h"
  19
  20 /***********************************************************************/
  21
  22 SharedListVector::SharedListVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0){};
  23
  24 /***********************************************************************/
  25
  26 SharedListVector::SharedListVector(int n):      DataVector(), data(n, "") , maxRank(0), numBins(0), numSeqs(0){};
  27
  28 /***********************************************************************/
  29 SharedListVector::SharedListVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
  30         try {
  31                 globaldata = GlobalData::getInstance();
  32
  33                 //set up groupmap for later.
  34                 groupmap = new GroupMap(globaldata->getGroupFile());
  35                 groupmap->readMap();
  36
  37                 int hold;
  38                 string inputData;
  39                 f >> label >> hold;
  40
  41                 data.assign(hold, "");
  42
  43                 for(int i=0;i<hold;i++){
  44                         f >> inputData;
  45                         set(i, inputData);
  46                 }
  47
  48         }
  49         catch(exception& e) {
  50                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function SharedListVector. 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 SharedListVector class function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  55                 exit(1);
  56         }
  57 }
  58
  59 /***********************************************************************/
  60 void SharedListVector::set(int binNumber, string seqNames){
  61         try {
  62                 int nNames_old = getNumNames(data[binNumber]);
  63                 data[binNumber] = seqNames;
  64                 int nNames_new = getNumNames(seqNames);
  65
  66                 if(nNames_old == 0)                     {       numBins++;                              }
  67                 if(nNames_new == 0)                     {       numBins--;                              }
  68                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  69
  70                 numSeqs += (nNames_new - nNames_old);
  71         }
  72         catch(exception& e) {
  73                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  74                 exit(1);
  75         }
  76         catch(...) {
  77                 cout << "An unknown error has occurred in the SharedListVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  78                 exit(1);
  79         }
  80 }
  81
  82 /***********************************************************************/
  83
  84 string SharedListVector::get(int index){
  85         return data[index];
  86 }
  87
  88 /***********************************************************************/
  89
  90 void SharedListVector::push_back(string seqNames){
  91         try {
  92                 data.push_back(seqNames);
  93                 int nNames = getNumNames(seqNames);
  94
  95                 numBins++;
  96
  97                 if(nNames > maxRank)    {       maxRank = nNames;       }
  98
  99                 numSeqs += nNames;
 100         }
 101         catch(exception& e) {
 102                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 103                 exit(1);
 104         }
 105         catch(...) {
 106                 cout << "An unknown error has occurred in the SharedListVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 107                 exit(1);
 108         }
 109 }
 110
 111 /***********************************************************************/
 112
 113 void SharedListVector::resize(int size){
 114         data.resize(size);
 115 }
 116
 117 /***********************************************************************/
 118
 119 int SharedListVector::size(){
 120         return data.size();
 121 }
 122 /***********************************************************************/
 123
 124 void SharedListVector::clear(){
 125         numBins = 0;
 126         maxRank = 0;
 127         numSeqs = 0;
 128         return data.clear();
 129
 130 }
 131
 132 /***********************************************************************/
 133
 134 void SharedListVector::print(ostream& output){
 135         try {
 136                 output << label << '\t' << numBins << '\t';
 137
 138                 for(int i=0;i<data.size();i++){
 139                         if(data[i] != ""){
 140                                 output << data[i] << '\t';
 141                         }
 142                 }
 143                 output << endl;
 144         }
 145         catch(exception& e) {
 146                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 147                 exit(1);
 148         }
 149         catch(...) {
 150                 cout << "An unknown error has occurred in the SharedListVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 151                 exit(1);
 152         }
 153 }
 154
 155
 156 /***********************************************************************/
 157
 158 RAbundVector SharedListVector::getRAbundVector(){
 159         try {
 160                 RAbundVector rav;
 161
 162                 for(int i=0;i<data.size();i++){
 163                         int binSize = getNumNames(data[i]);
 164                         rav.push_back(binSize);
 165                 }
 166
 167         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 168         //      sort(rav.rbegin(), rav.rend());
 169         //
 170         //      for(int i=data.size()-1;i>=0;i--){
 171         //              if(rav.get(i) == 0){    rav.pop_back(); }
 172         //              else{
 173         //                      break;
 174         //              }
 175         //      }
 176                 rav.setLabel(label);
 177
 178                 return rav;
 179         }
 180         catch(exception& e) {
 181                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 182                 exit(1);
 183         }
 184         catch(...) {
 185                 cout << "An unknown error has occurred in the SharedListVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 186                 exit(1);
 187         }
 188 }
 189
 190 /***********************************************************************/
 191
 192 SAbundVector SharedListVector::getSAbundVector(){
 193         try {
 194                 SAbundVector sav(maxRank+1);
 195
 196                 for(int i=0;i<data.size();i++){
 197                         int binSize = getNumNames(data[i]);
 198                         sav.set(binSize, sav.get(binSize) + 1);
 199                 }
 200                 sav.set(0, 0);
 201                 sav.setLabel(label);
 202
 203                 return sav;
 204         }
 205         catch(exception& e) {
 206                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 207                 exit(1);
 208         }
 209         catch(...) {
 210                 cout << "An unknown error has occurred in the SharedListVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 211                 exit(1);
 212         }
 213 }
 214
 215 /***********************************************************************/
 216 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 217         try {
 218                 string groupName, names, name;
 219
 220                 SharedOrderVector* order = new SharedOrderVector();
 221                 order->setLabel(label);
 222
 223                 for(int i=0;i<numBins;i++){
 224                         int binSize = getNumNames(get(i));      //find number of individual in given bin
 225                         names = get(i);
 226                         while (names.find_first_of(',') != -1) {
 227                                 name = names.substr(0,names.find_first_of(','));
 228                                 names = names.substr(names.find_first_of(',')+1, names.length());
 229                                 groupName = groupmap->getGroup(name);
 230                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 231                         }
 232                         //get last name
 233                         groupName = groupmap->getGroup(names);
 234                         order->push_back(i, binSize, groupName);
 235                 }
 236                 random_shuffle(order->begin(), order->end());
 237                 return order;
 238         }
 239         catch(exception& e) {
 240                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 241                 exit(1);
 242         }
 243         catch(...) {
 244                 cout << "An unknown error has occurred in the SharedListVector class function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 245                 exit(1);
 246         }
 247
 248 }
 249 /***********************************************************************/
 250 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 251         try {
 252                 SharedRAbundVector rav(data.size());
 253                 string group, names, name;
 254
 255                 for(int i=0;i<numBins;i++){
 256                         names = get(i);
 257                         while (names.find_first_of(',') != -1) {
 258                                 name = names.substr(0,names.find_first_of(','));
 259                                 names = names.substr(names.find_first_of(',')+1, names.length());
 260                                 group = groupmap->getGroup(name);
 261                                 if (group == groupName) { //this name is in the group you want the vector for.
 262                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 263                                 }
 264                         }
 265
 266                         //get last name
 267                         groupName = groupmap->getGroup(names);
 268                         if (group == groupName) { //this name is in the group you want the vector for.
 269                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 270                         }
 271                 }
 272
 273                 rav.setLabel(label);
 274                 rav.setGroup(groupName);
 275                 return rav;
 276
 277         }
 278         catch(exception& e) {
 279                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 280                 exit(1);
 281         }
 282         catch(...) {
 283                 cout << "An unknown error has occurred in the SharedListVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 284                 exit(1);
 285         }
 286 }
 287
 288 /***********************************************************************/
 289 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 290         try {
 291                 SharedSAbundVector sav;
 292                 SharedRAbundVector rav;
 293
 294                 rav = this->getSharedRAbundVector(groupName);
 295                 sav = rav.getSharedSAbundVector();
 296
 297                 return sav;
 298         }
 299         catch(exception& e) {
 300                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 301                 exit(1);
 302         }
 303         catch(...) {
 304                 cout << "An unknown error has occurred in the SharedListVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 305                 exit(1);
 306         }
 307 }
 308 /***********************************************************************/
 309
 310 OrderVector SharedListVector::getOrderVector(map<string,int>* orderMap = NULL){
 311
 312         try {
 313                 if(orderMap == NULL){
 314                         OrderVector ov;
 315
 316                         for(int i=0;i<data.size();i++){
 317                                 int binSize = getNumNames(data[i]);
 318                                 for(int j=0;j<binSize;j++){
 319                                         ov.push_back(i);
 320                                 }
 321                         }
 322                         random_shuffle(ov.begin(), ov.end());
 323                         ov.setLabel(label);
 324                         ov.getNumBins();
 325
 326                         return ov;
 327
 328                 }
 329                 else{
 330                         OrderVector ov(numSeqs);
 331
 332                         for(int i=0;i<data.size();i++){
 333                                 string listOTU = data[i];
 334                                 int length = listOTU.size();
 335
 336                                 string seqName="";
 337
 338                                 for(int j=0;j<length;j++){
 339
 340                                         if(listOTU[j] != ','){
 341                                                 seqName += listOTU[j];
 342                                         }
 343                                         else{
 344                                                 if(orderMap->count(seqName) == 0){
 345                                                         cerr << seqName << " not found, check *.names file\n";
 346                                                         exit(1);
 347                                                 }
 348
 349                                                 ov.set((*orderMap)[seqName], i);
 350                                                 seqName = "";
 351                                         }
 352                                 }
 353
 354                                 if(orderMap->count(seqName) == 0){
 355                                         cerr << seqName << " not found, check *.names file\n";
 356                                         exit(1);
 357                                 }
 358                                 ov.set((*orderMap)[seqName], i);
 359                         }
 360
 361                         ov.setLabel(label);
 362                         ov.getNumBins();
 363
 364                         return ov;
 365                 }
 366         }
 367         catch(exception& e) {
 368                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 369                 exit(1);
 370         }
 371         catch(...) {
 372                 cout << "An unknown error has occurred in the SharedListVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 373                 exit(1);
 374         }
 375 }
 376
 377 /***********************************************************************/
 378