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