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