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 #include <map>
  14 #include <exception>
  15 #include "sabundvector.hpp"
  16 #include "rabundvector.hpp"
  17 #include "ordervector.hpp"
  18 #include "datavector.hpp"
  19 #include "utilities.hpp"
  20 #include "sharedlistvector.h"
  21 #include "sharedordervector.h"
  22
  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                 f >> label >> hold;
  39
  40                 data.assign(hold, "");
  41                 string inputData = "";
  42
  43                 for(int i=0;i<hold;i++){
  44                         f >> inputData;
  45                         set(i, inputData);
  46                 }
  47         }
  48         catch(exception& e) {
  49                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  50                 exit(1);
  51         }
  52         catch(...) {
  53                 cout << "An unknown error has occurred in the SharedListVector class function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  54                 exit(1);
  55         }
  56 }
  57
  58 /***********************************************************************/
  59 void SharedListVector::set(int binNumber, string seqNames){
  60         try {
  61                 int nNames_old = getNumNames(data[binNumber]);
  62                 data[binNumber] = seqNames;
  63                 int nNames_new = getNumNames(seqNames);
  64
  65                 if(nNames_old == 0)                     {       numBins++;                              }
  66                 if(nNames_new == 0)                     {       numBins--;                              }
  67                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  68
  69                 numSeqs += (nNames_new - nNames_old);
  70         }
  71         catch(exception& e) {
  72                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  73                 exit(1);
  74         }
  75         catch(...) {
  76                 cout << "An unknown error has occurred in the SharedListVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  77                 exit(1);
  78         }
  79 }
  80
  81 /***********************************************************************/
  82
  83 string SharedListVector::get(int index){
  84         return data[index];
  85 }
  86
  87 /***********************************************************************/
  88
  89 void SharedListVector::push_back(string seqNames){
  90         try {
  91                 data.push_back(seqNames);
  92                 int nNames = getNumNames(seqNames);
  93
  94                 numBins++;
  95
  96                 if(nNames > maxRank)    {       maxRank = nNames;       }
  97
  98                 numSeqs += nNames;
  99         }
 100         catch(exception& e) {
 101                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 102                 exit(1);
 103         }
 104         catch(...) {
 105                 cout << "An unknown error has occurred in the SharedListVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 106                 exit(1);
 107         }
 108 }
 109
 110 /***********************************************************************/
 111
 112 void SharedListVector::resize(int size){
 113         data.resize(size);
 114 }
 115
 116 /***********************************************************************/
 117
 118 int SharedListVector::size(){
 119         return data.size();
 120 }
 121 /***********************************************************************/
 122
 123 void SharedListVector::clear(){
 124         numBins = 0;
 125         maxRank = 0;
 126         numSeqs = 0;
 127         return data.clear();
 128
 129 }
 130
 131 /***********************************************************************/
 132
 133 void SharedListVector::print(ostream& output){
 134         try {
 135                 output << label << '\t' << numBins << '\t';
 136
 137                 for(int i=0;i<data.size();i++){
 138                         if(data[i] != ""){
 139                                 output << data[i] << '\t';
 140                         }
 141                 }
 142                 output << endl;
 143         }
 144         catch(exception& e) {
 145                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 146                 exit(1);
 147         }
 148         catch(...) {
 149                 cout << "An unknown error has occurred in the SharedListVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 150                 exit(1);
 151         }
 152 }
 153
 154
 155 /***********************************************************************/
 156
 157 RAbundVector SharedListVector::getRAbundVector(){
 158         try {
 159                 RAbundVector rav;
 160
 161                 for(int i=0;i<data.size();i++){
 162                         int binSize = getNumNames(data[i]);
 163                         rav.push_back(binSize);
 164                 }
 165
 166         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 167         //      sort(rav.rbegin(), rav.rend());
 168         //
 169         //      for(int i=data.size()-1;i>=0;i--){
 170         //              if(rav.get(i) == 0){    rav.pop_back(); }
 171         //              else{
 172         //                      break;
 173         //              }
 174         //      }
 175                 rav.setLabel(label);
 176
 177                 return rav;
 178         }
 179         catch(exception& e) {
 180                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 181                 exit(1);
 182         }
 183         catch(...) {
 184                 cout << "An unknown error has occurred in the SharedListVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187 }
 188
 189 /***********************************************************************/
 190
 191 SAbundVector SharedListVector::getSAbundVector(){
 192         try {
 193                 SAbundVector sav(maxRank+1);
 194
 195                 for(int i=0;i<data.size();i++){
 196                         int binSize = getNumNames(data[i]);
 197                         sav.set(binSize, sav.get(binSize) + 1);
 198                 }
 199                 sav.set(0, 0);
 200                 sav.setLabel(label);
 201
 202                 return sav;
 203         }
 204         catch(exception& e) {
 205                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 206                 exit(1);
 207         }
 208         catch(...) {
 209                 cout << "An unknown error has occurred in the SharedListVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 210                 exit(1);
 211         }
 212 }
 213
 214 /***********************************************************************/
 215 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 216         try {
 217                 string groupName, names, name;
 218
 219                 SharedOrderVector* order = new SharedOrderVector();
 220                 order->setLabel(label);
 221
 222                 for(int i=0;i<numBins;i++){
 223                         int binSize = getNumNames(get(i));      //find number of individual in given bin
 224                         names = get(i);
 225                         while (names.find_first_of(',') != -1) {
 226                                 name = names.substr(0,names.find_first_of(','));
 227                                 names = names.substr(names.find_first_of(',')+1, names.length());
 228                                 groupName = groupmap->getGroup(name);
 229                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 230                         }
 231                         //get last name
 232                         groupName = groupmap->getGroup(names);
 233                         order->push_back(i, binSize, groupName);
 234                 }
 235                 random_shuffle(order->begin(), order->end());
 236                 return order;
 237         }
 238         catch(exception& e) {
 239                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 240                 exit(1);
 241         }
 242         catch(...) {
 243                 cout << "An unknown error has occurred in the SharedListVector class function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 244                 exit(1);
 245         }
 246
 247 }
 248 /***********************************************************************/
 249 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 250         try {
 251                 SharedRAbundVector rav(data.size());
 252                 string group, names, name;
 253
 254                 for(int i=0;i<numBins;i++){
 255                         names = get(i);
 256                         while (names.find_first_of(',') != -1) {
 257                                 name = names.substr(0,names.find_first_of(','));
 258                                 names = names.substr(names.find_first_of(',')+1, names.length());
 259                                 group = groupmap->getGroup(name);
 260                                 if (group == groupName) { //this name is in the group you want the vector for.
 261                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 262                                 }
 263                         }
 264
 265                         //get last name
 266                         groupName = groupmap->getGroup(names);
 267                         if (group == groupName) { //this name is in the group you want the vector for.
 268                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 269                         }
 270                 }
 271
 272                 rav.setLabel(label);
 273                 rav.setGroup(groupName);
 274                 return rav;
 275
 276         }
 277         catch(exception& e) {
 278                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 279                 exit(1);
 280         }
 281         catch(...) {
 282                 cout << "An unknown error has occurred in the SharedListVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 283                 exit(1);
 284         }
 285 }
 286
 287 /***********************************************************************/
 288 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 289         try {
 290                 SharedSAbundVector sav;
 291                 SharedRAbundVector rav;
 292
 293                 rav = this->getSharedRAbundVector(groupName);
 294                 sav = rav.getSharedSAbundVector();
 295
 296                 return sav;
 297         }
 298         catch(exception& e) {
 299                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 300                 exit(1);
 301         }
 302         catch(...) {
 303                 cout << "An unknown error has occurred in the SharedListVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 304                 exit(1);
 305         }
 306 }
 307 /***********************************************************************/
 308
 309 OrderVector SharedListVector::getOrderVector(map<string,int>* orderMap = NULL){
 310
 311         try {
 312                 if(orderMap == NULL){
 313                         OrderVector ov;
 314
 315                         for(int i=0;i<data.size();i++){
 316                                 int binSize = getNumNames(data[i]);
 317                                 for(int j=0;j<binSize;j++){
 318                                         ov.push_back(i);
 319                                 }
 320                         }
 321                         random_shuffle(ov.begin(), ov.end());
 322                         ov.setLabel(label);
 323                         ov.getNumBins();
 324
 325                         return ov;
 326
 327                 }
 328                 else{
 329                         OrderVector ov(numSeqs);
 330
 331                         for(int i=0;i<data.size();i++){
 332                                 string listOTU = data[i];
 333                                 int length = listOTU.size();
 334
 335                                 string seqName="";
 336
 337                                 for(int j=0;j<length;j++){
 338
 339                                         if(listOTU[j] != ','){
 340                                                 seqName += listOTU[j];
 341                                         }
 342                                         else{
 343                                                 if(orderMap->count(seqName) == 0){
 344                                                         cerr << seqName << " not found, check *.names file\n";
 345                                                         exit(1);
 346                                                 }
 347
 348                                                 ov.set((*orderMap)[seqName], i);
 349                                                 seqName = "";
 350                                         }
 351                                 }
 352
 353                                 if(orderMap->count(seqName) == 0){
 354                                         cerr << seqName << " not found, check *.names file\n";
 355                                         exit(1);
 356                                 }
 357                                 ov.set((*orderMap)[seqName], i);
 358                         }
 359
 360                         ov.setLabel(label);
 361                         ov.getNumBins();
 362
 363                         return ov;
 364                 }
 365         }
 366         catch(exception& e) {
 367                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 368                 exit(1);
 369         }
 370         catch(...) {
 371                 cout << "An unknown error has occurred in the SharedListVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 372                 exit(1);
 373         }
 374 }
 375
 376 /***********************************************************************/
 377