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 SharedListVector::SharedListVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0){};
  26
  27 /***********************************************************************/
  28
  29 SharedListVector::SharedListVector(int n):      DataVector(), data(n, "") , maxRank(0), numBins(0), numSeqs(0){};
  30
  31 /***********************************************************************/
  32 SharedListVector::SharedListVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
  33         try {
  34                 globaldata = GlobalData::getInstance();
  35
  36                 //set up groupmap for later.
  37                 groupmap = new GroupMap(globaldata->getGroupFile());
  38                 groupmap->readMap();
  39
  40                 int hold;
  41                 f >> label >> hold;
  42
  43                 data.assign(hold, "");
  44                 string inputData = "";
  45
  46                 for(int i=0;i<hold;i++){
  47                         f >> inputData;
  48                         set(i, inputData);
  49                 }
  50         }
  51         catch(exception& e) {
  52                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  53                 exit(1);
  54         }
  55         catch(...) {
  56                 cout << "An unknown error has occurred in the SharedListVector class function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  57                 exit(1);
  58         }
  59 }
  60
  61 /***********************************************************************/
  62 void SharedListVector::set(int binNumber, string seqNames){
  63         try {
  64                 int nNames_old = getNumNames(data[binNumber]);
  65                 data[binNumber] = seqNames;
  66                 int nNames_new = getNumNames(seqNames);
  67
  68                 if(nNames_old == 0)                     {       numBins++;                              }
  69                 if(nNames_new == 0)                     {       numBins--;                              }
  70                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  71
  72                 numSeqs += (nNames_new - nNames_old);
  73         }
  74         catch(exception& e) {
  75                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  76                 exit(1);
  77         }
  78         catch(...) {
  79                 cout << "An unknown error has occurred in the SharedListVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  80                 exit(1);
  81         }
  82 }
  83
  84 /***********************************************************************/
  85
  86 string SharedListVector::get(int index){
  87         return data[index];
  88 }
  89
  90 /***********************************************************************/
  91
  92 void SharedListVector::push_back(string seqNames){
  93         try {
  94                 data.push_back(seqNames);
  95                 int nNames = getNumNames(seqNames);
  96
  97                 numBins++;
  98
  99                 if(nNames > maxRank)    {       maxRank = nNames;       }
 100
 101                 numSeqs += nNames;
 102         }
 103         catch(exception& e) {
 104                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 105                 exit(1);
 106         }
 107         catch(...) {
 108                 cout << "An unknown error has occurred in the SharedListVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 109                 exit(1);
 110         }
 111 }
 112
 113 /***********************************************************************/
 114
 115 void SharedListVector::resize(int size){
 116         data.resize(size);
 117 }
 118
 119 /***********************************************************************/
 120
 121 int SharedListVector::size(){
 122         return data.size();
 123 }
 124 /***********************************************************************/
 125
 126 void SharedListVector::clear(){
 127         numBins = 0;
 128         maxRank = 0;
 129         numSeqs = 0;
 130         return data.clear();
 131
 132 }
 133
 134 /***********************************************************************/
 135
 136 void SharedListVector::print(ostream& output){
 137         try {
 138                 output << label << '\t' << numBins << '\t';
 139
 140                 for(int i=0;i<data.size();i++){
 141                         if(data[i] != ""){
 142                                 output << data[i] << '\t';
 143                         }
 144                 }
 145                 output << endl;
 146         }
 147         catch(exception& e) {
 148                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 149                 exit(1);
 150         }
 151         catch(...) {
 152                 cout << "An unknown error has occurred in the SharedListVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 153                 exit(1);
 154         }
 155 }
 156
 157
 158 /***********************************************************************/
 159
 160 RAbundVector SharedListVector::getRAbundVector(){
 161         try {
 162                 RAbundVector rav;
 163
 164                 for(int i=0;i<data.size();i++){
 165                         int binSize = getNumNames(data[i]);
 166                         rav.push_back(binSize);
 167                 }
 168
 169         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 170         //      sort(rav.rbegin(), rav.rend());
 171         //
 172         //      for(int i=data.size()-1;i>=0;i--){
 173         //              if(rav.get(i) == 0){    rav.pop_back(); }
 174         //              else{
 175         //                      break;
 176         //              }
 177         //      }
 178                 rav.setLabel(label);
 179
 180                 return rav;
 181         }
 182         catch(exception& e) {
 183                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 184                 exit(1);
 185         }
 186         catch(...) {
 187                 cout << "An unknown error has occurred in the SharedListVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 188                 exit(1);
 189         }
 190 }
 191
 192 /***********************************************************************/
 193
 194 SAbundVector SharedListVector::getSAbundVector(){
 195         try {
 196                 SAbundVector sav(maxRank+1);
 197
 198                 for(int i=0;i<data.size();i++){
 199                         int binSize = getNumNames(data[i]);
 200                         sav.set(binSize, sav.get(binSize) + 1);
 201                 }
 202                 sav.set(0, 0);
 203                 sav.setLabel(label);
 204
 205                 return sav;
 206         }
 207         catch(exception& e) {
 208                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 209                 exit(1);
 210         }
 211         catch(...) {
 212                 cout << "An unknown error has occurred in the SharedListVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 213                 exit(1);
 214         }
 215 }
 216
 217 /***********************************************************************/
 218 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 219         try {
 220                 string groupName, names, name;
 221
 222                 SharedOrderVector* order = new SharedOrderVector();
 223                 order->setLabel(label);
 224
 225                 for(int i=0;i<numBins;i++){
 226                         int binSize = getNumNames(get(i));      //find number of individual in given bin
 227                         names = get(i);
 228                         while (names.find_first_of(',') != -1) {
 229                                 name = names.substr(0,names.find_first_of(','));
 230                                 names = names.substr(names.find_first_of(',')+1, names.length());
 231                                 groupName = groupmap->getGroup(name);
 232                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 233                         }
 234                         //get last name
 235                         groupName = groupmap->getGroup(names);
 236                         order->push_back(i, binSize, groupName);
 237                 }
 238                 random_shuffle(order->begin(), order->end());
 239                 return order;
 240         }
 241         catch(exception& e) {
 242                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 243                 exit(1);
 244         }
 245         catch(...) {
 246                 cout << "An unknown error has occurred in the SharedListVector class function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 247                 exit(1);
 248         }
 249
 250 }
 251 /***********************************************************************/
 252 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 253         try {
 254                 SharedRAbundVector rav(data.size());
 255                 string group, names, name;
 256
 257                 for(int i=0;i<numBins;i++){
 258                         names = get(i);
 259                         while (names.find_first_of(',') != -1) {
 260                                 name = names.substr(0,names.find_first_of(','));
 261                                 names = names.substr(names.find_first_of(',')+1, names.length());
 262                                 group = groupmap->getGroup(name);
 263                                 if (group == groupName) { //this name is in the group you want the vector for.
 264                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 265                                 }
 266                         }
 267
 268                         //get last name
 269                         groupName = groupmap->getGroup(names);
 270                         if (group == groupName) { //this name is in the group you want the vector for.
 271                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 272                         }
 273                 }
 274
 275                 rav.setLabel(label);
 276                 rav.setGroup(groupName);
 277                 return rav;
 278
 279         }
 280         catch(exception& e) {
 281                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 282                 exit(1);
 283         }
 284         catch(...) {
 285                 cout << "An unknown error has occurred in the SharedListVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 286                 exit(1);
 287         }
 288 }
 289
 290 /***********************************************************************/
 291 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 292         try {
 293                 SharedSAbundVector sav;
 294                 SharedRAbundVector rav;
 295
 296                 rav = this->getSharedRAbundVector(groupName);
 297                 sav = rav.getSharedSAbundVector();
 298
 299                 return sav;
 300         }
 301         catch(exception& e) {
 302                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 303                 exit(1);
 304         }
 305         catch(...) {
 306                 cout << "An unknown error has occurred in the SharedListVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 307                 exit(1);
 308         }
 309 }
 310 /***********************************************************************/
 311
 312 OrderVector SharedListVector::getOrderVector(map<string,int>* orderMap = NULL){
 313
 314         try {
 315                 if(orderMap == NULL){
 316                         OrderVector ov;
 317
 318                         for(int i=0;i<data.size();i++){
 319                                 int binSize = getNumNames(data[i]);
 320                                 for(int j=0;j<binSize;j++){
 321                                         ov.push_back(i);
 322                                 }
 323                         }
 324                         random_shuffle(ov.begin(), ov.end());
 325                         ov.setLabel(label);
 326                         ov.getNumBins();
 327
 328                         return ov;
 329
 330                 }
 331                 else{
 332                         OrderVector ov(numSeqs);
 333
 334                         for(int i=0;i<data.size();i++){
 335                                 string listOTU = data[i];
 336                                 int length = listOTU.size();
 337
 338                                 string seqName="";
 339
 340                                 for(int j=0;j<length;j++){
 341
 342                                         if(listOTU[j] != ','){
 343                                                 seqName += listOTU[j];
 344                                         }
 345                                         else{
 346                                                 if(orderMap->count(seqName) == 0){
 347                                                         cerr << seqName << " not found, check *.names file\n";
 348                                                         exit(1);
 349                                                 }
 350
 351                                                 ov.set((*orderMap)[seqName], i);
 352                                                 seqName = "";
 353                                         }
 354                                 }
 355
 356                                 if(orderMap->count(seqName) == 0){
 357                                         cerr << seqName << " not found, check *.names file\n";
 358                                         exit(1);
 359                                 }
 360                                 ov.set((*orderMap)[seqName], i);
 361                         }
 362
 363                         ov.setLabel(label);
 364                         ov.getNumBins();
 365
 366                         return ov;
 367                 }
 368         }
 369         catch(exception& e) {
 370                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 371                 exit(1);
 372         }
 373         catch(...) {
 374                 cout << "An unknown error has occurred in the SharedListVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 375                 exit(1);
 376         }
 377 }
 378
 379 /***********************************************************************/
 380