listvector.cpp

   1 /*
   2  *  list.cpp
   3  *
   4  *
   5  *  Created by Pat Schloss on 8/8/08.
   6  *  Copyright 2008 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 using namespace std;
  11
  12 #include <map>
  13 #include <exception>
  14 #include "sabundvector.hpp"
  15 #include "rabundvector.hpp"
  16 #include "ordervector.hpp"
  17 #include "datavector.hpp"
  18 #include "utilities.hpp"
  19 #include "listvector.hpp"
  20
  21
  22 /***********************************************************************/
  23
  24 ListVector::ListVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0){};
  25
  26 /***********************************************************************/
  27
  28 ListVector::ListVector(int n):  DataVector(), data(n, "") , maxRank(0), numBins(0), numSeqs(0){};
  29
  30 /***********************************************************************/
  31
  32 ListVector::ListVector(string id, vector<string> lv) : DataVector(id), data(lv){
  33         try {
  34                 for(int i=0;i<data.size();i++){
  35                         if(data[i] != ""){
  36                                 int binSize = getNumNames(data[i]);
  37                                 numBins = i+1;
  38                                 if(binSize > maxRank)   {       maxRank = binSize;      }
  39                                 numSeqs += binSize;
  40                         }
  41                 }
  42         }
  43         catch(exception& e) {
  44                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function ListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  45                 exit(1);
  46         }
  47         catch(...) {
  48                 cout << "An unknown error has occurred in the ListVector class function ListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  49                 exit(1);
  50         }
  51 }
  52
  53 /**********************************************************************/
  54
  55 ListVector::ListVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
  56         try {
  57                 int hold;
  58                 f >> label >> hold;
  59
  60                 data.assign(hold, "");
  61                 string inputData = "";
  62
  63                 for(int i=0;i<hold;i++){
  64                         f >> inputData;
  65                         set(i, inputData);
  66                 }
  67         }
  68         catch(exception& e) {
  69                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function ListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  70                 exit(1);
  71         }
  72         catch(...) {
  73                 cout << "An unknown error has occurred in the ListVector class function ListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  74                 exit(1);
  75         }
  76 }
  77
  78 /***********************************************************************/
  79
  80 void ListVector::set(int binNumber, string seqNames){
  81         try {
  82                 int nNames_old = getNumNames(data[binNumber]);
  83                 data[binNumber] = seqNames;
  84                 int nNames_new = getNumNames(seqNames);
  85
  86                 if(nNames_old == 0)                     {       numBins++;                              }
  87                 if(nNames_new == 0)                     {       numBins--;                              }
  88                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  89
  90                 numSeqs += (nNames_new - nNames_old);
  91         }
  92         catch(exception& e) {
  93                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  94                 exit(1);
  95         }
  96         catch(...) {
  97                 cout << "An unknown error has occurred in the ListVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  98                 exit(1);
  99         }
 100 }
 101
 102 /***********************************************************************/
 103
 104 string ListVector::get(int index){
 105         return data[index];
 106 }
 107
 108 /***********************************************************************/
 109
 110 void ListVector::push_back(string seqNames){
 111         try {
 112                 data.push_back(seqNames);
 113                 int nNames = getNumNames(seqNames);
 114
 115                 numBins++;
 116
 117                 if(nNames > maxRank)    {       maxRank = nNames;       }
 118
 119                 numSeqs += nNames;
 120         }
 121         catch(exception& e) {
 122                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 123                 exit(1);
 124         }
 125         catch(...) {
 126                 cout << "An unknown error has occurred in the ListVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 127                 exit(1);
 128         }
 129 }
 130
 131 /***********************************************************************/
 132
 133 void ListVector::resize(int size){
 134         data.resize(size);
 135 }
 136
 137 /***********************************************************************/
 138
 139 int ListVector::size(){
 140         return data.size();
 141 }
 142 /***********************************************************************/
 143
 144 void ListVector::clear(){
 145         numBins = 0;
 146         maxRank = 0;
 147         numSeqs = 0;
 148         return data.clear();
 149
 150 }
 151
 152 /***********************************************************************/
 153
 154 void ListVector::print(ostream& output){
 155         try {
 156                 output << label << '\t' << numBins << '\t';
 157
 158                 for(int i=0;i<data.size();i++){
 159                         if(data[i] != ""){
 160                                 output << data[i] << '\t';
 161                         }
 162                 }
 163                 output << endl;
 164         }
 165         catch(exception& e) {
 166                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 167                 exit(1);
 168         }
 169         catch(...) {
 170                 cout << "An unknown error has occurred in the ListVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 171                 exit(1);
 172         }
 173 }
 174
 175
 176 /***********************************************************************/
 177
 178 RAbundVector ListVector::getRAbundVector(){
 179         try {
 180                 RAbundVector rav;
 181
 182                 for(int i=0;i<data.size();i++){
 183                         int binSize = getNumNames(data[i]);
 184                         rav.push_back(binSize);
 185                 }
 186
 187         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 188         //      sort(rav.rbegin(), rav.rend());
 189         //
 190         //      for(int i=data.size()-1;i>=0;i--){
 191         //              if(rav.get(i) == 0){    rav.pop_back(); }
 192         //              else{
 193         //                      break;
 194         //              }
 195         //      }
 196                 rav.setLabel(label);
 197
 198                 return rav;
 199         }
 200         catch(exception& e) {
 201                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 202                 exit(1);
 203         }
 204         catch(...) {
 205                 cout << "An unknown error has occurred in the ListVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 206                 exit(1);
 207         }
 208 }
 209
 210 /***********************************************************************/
 211
 212 SAbundVector ListVector::getSAbundVector(){
 213         try {
 214                 SAbundVector sav(maxRank+1);
 215
 216                 for(int i=0;i<data.size();i++){
 217                         int binSize = getNumNames(data[i]);
 218                         sav.set(binSize, sav.get(binSize) + 1);
 219                 }
 220                 sav.set(0, 0);
 221                 sav.setLabel(label);
 222
 223                 return sav;
 224         }
 225         catch(exception& e) {
 226                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 227                 exit(1);
 228         }
 229         catch(...) {
 230                 cout << "An unknown error has occurred in the ListVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 231                 exit(1);
 232         }
 233 }
 234
 235 /***********************************************************************/
 236 SharedOrderVector* ListVector::getSharedOrderVector(){
 237         globaldata = GlobalData::getInstance();
 238         string groupName, names, name;
 239         groupmap = globaldata->gGroupmap;
 240         SharedOrderVector* order;
 241         order = new SharedOrderVector();
 242         order->setLabel(label);
 243
 244         for(int i=0;i<numBins;i++){
 245                 int binSize = getNumNames(get(i));      //find number of individual in given bin
 246                 names = get(i);
 247                 while (names.find_first_of(',') != -1) {
 248                         name = names.substr(0,names.find_first_of(','));
 249                         names = names.substr(names.find_first_of(',')+1, names.length());
 250                         groupName = groupmap->getGroup(name);
 251                         order->push_back(i, binSize, groupName);  //i represents what bin you are in
 252                 }
 253                 //get last name
 254                 groupName = groupmap->getGroup(names);
 255                 order->push_back(i, binSize, groupName);
 256         }
 257         random_shuffle(order->begin(), order->end());
 258         return order;
 259 }
 260
 261 /***********************************************************************/
 262
 263 OrderVector ListVector::getOrderVector(map<string,int>* orderMap = NULL){
 264
 265         try {
 266                 if(orderMap == NULL){
 267                         OrderVector ov;
 268
 269                         for(int i=0;i<data.size();i++){
 270                                 int binSize = getNumNames(data[i]);
 271                                 for(int j=0;j<binSize;j++){
 272                                         ov.push_back(i);
 273                                 }
 274                         }
 275                         random_shuffle(ov.begin(), ov.end());
 276                         ov.setLabel(label);
 277                         ov.getNumBins();
 278
 279                         return ov;
 280
 281                 }
 282                 else{
 283                         OrderVector ov(numSeqs);
 284
 285                         for(int i=0;i<data.size();i++){
 286                                 string listOTU = data[i];
 287                                 int length = listOTU.size();
 288
 289                                 string seqName="";
 290
 291                                 for(int j=0;j<length;j++){
 292
 293                                         if(listOTU[j] != ','){
 294                                                 seqName += listOTU[j];
 295                                         }
 296                                         else{
 297                                                 if(orderMap->count(seqName) == 0){
 298                                                         cerr << seqName << " not found, check *.names file\n";
 299                                                         exit(1);
 300                                                 }
 301
 302                                                 ov.set((*orderMap)[seqName], i);
 303                                                 seqName = "";
 304                                         }
 305                                 }
 306
 307                                 if(orderMap->count(seqName) == 0){
 308                                         cerr << seqName << " not found, check *.names file\n";
 309                                         exit(1);
 310                                 }
 311                                 ov.set((*orderMap)[seqName], i);
 312                         }
 313
 314                         ov.setLabel(label);
 315                         ov.getNumBins();
 316
 317                         return ov;
 318                 }
 319         }
 320         catch(exception& e) {
 321                 cout << "Standard Error: " << e.what() << " has occurred in the ListVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 322                 exit(1);
 323         }
 324         catch(...) {
 325                 cout << "An unknown error has occurred in the ListVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 326                 exit(1);
 327         }
 328 }
 329
 330 /***********************************************************************/