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                 string inputData;
  42                 f >> label >> hold;
  43
  44                 data.assign(hold, "");
  45
  46                 for(int i=0;i<hold;i++){
  47                         f >> inputData;
  48                         set(i, inputData);
  49                 }
  50
  51         }
  52         catch(exception& e) {
  53                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  54                 exit(1);
  55         }
  56         catch(...) {
  57                 cout << "An unknown error has occurred in the SharedListVector class function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  58                 exit(1);
  59         }
  60 }
  61
  62 /***********************************************************************/
  63 void SharedListVector::set(int binNumber, string seqNames){
  64         try {
  65                 int nNames_old = getNumNames(data[binNumber]);
  66                 data[binNumber] = seqNames;
  67                 int nNames_new = getNumNames(seqNames);
  68
  69                 if(nNames_old == 0)                     {       numBins++;                              }
  70                 if(nNames_new == 0)                     {       numBins--;                              }
  71                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  72
  73                 numSeqs += (nNames_new - nNames_old);
  74         }
  75         catch(exception& e) {
  76                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  77                 exit(1);
  78         }
  79         catch(...) {
  80                 cout << "An unknown error has occurred in the SharedListVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  81                 exit(1);
  82         }
  83 }
  84
  85 /***********************************************************************/
  86
  87 string SharedListVector::get(int index){
  88         return data[index];
  89 }
  90
  91 /***********************************************************************/
  92
  93 void SharedListVector::push_back(string seqNames){
  94         try {
  95                 data.push_back(seqNames);
  96                 int nNames = getNumNames(seqNames);
  97
  98                 numBins++;
  99
 100                 if(nNames > maxRank)    {       maxRank = nNames;       }
 101
 102                 numSeqs += nNames;
 103         }
 104         catch(exception& e) {
 105                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 106                 exit(1);
 107         }
 108         catch(...) {
 109                 cout << "An unknown error has occurred in the SharedListVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 110                 exit(1);
 111         }
 112 }
 113
 114 /***********************************************************************/
 115
 116 void SharedListVector::resize(int size){
 117         data.resize(size);
 118 }
 119
 120 /***********************************************************************/
 121
 122 int SharedListVector::size(){
 123         return data.size();
 124 }
 125 /***********************************************************************/
 126
 127 void SharedListVector::clear(){
 128         numBins = 0;
 129         maxRank = 0;
 130         numSeqs = 0;
 131         return data.clear();
 132
 133 }
 134
 135 /***********************************************************************/
 136
 137 void SharedListVector::print(ostream& output){
 138         try {
 139                 output << label << '\t' << numBins << '\t';
 140
 141                 for(int i=0;i<data.size();i++){
 142                         if(data[i] != ""){
 143                                 output << data[i] << '\t';
 144                         }
 145                 }
 146                 output << endl;
 147         }
 148         catch(exception& e) {
 149                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 150                 exit(1);
 151         }
 152         catch(...) {
 153                 cout << "An unknown error has occurred in the SharedListVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 154                 exit(1);
 155         }
 156 }
 157
 158
 159 /***********************************************************************/
 160
 161 RAbundVector SharedListVector::getRAbundVector(){
 162         try {
 163                 RAbundVector rav;
 164
 165                 for(int i=0;i<data.size();i++){
 166                         int binSize = getNumNames(data[i]);
 167                         rav.push_back(binSize);
 168                 }
 169
 170         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 171         //      sort(rav.rbegin(), rav.rend());
 172         //
 173         //      for(int i=data.size()-1;i>=0;i--){
 174         //              if(rav.get(i) == 0){    rav.pop_back(); }
 175         //              else{
 176         //                      break;
 177         //              }
 178         //      }
 179                 rav.setLabel(label);
 180
 181                 return rav;
 182         }
 183         catch(exception& e) {
 184                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187         catch(...) {
 188                 cout << "An unknown error has occurred in the SharedListVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 189                 exit(1);
 190         }
 191 }
 192
 193 /***********************************************************************/
 194
 195 SAbundVector SharedListVector::getSAbundVector(){
 196         try {
 197                 SAbundVector sav(maxRank+1);
 198
 199                 for(int i=0;i<data.size();i++){
 200                         int binSize = getNumNames(data[i]);
 201                         sav.set(binSize, sav.get(binSize) + 1);
 202                 }
 203                 sav.set(0, 0);
 204                 sav.setLabel(label);
 205
 206                 return sav;
 207         }
 208         catch(exception& e) {
 209                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 210                 exit(1);
 211         }
 212         catch(...) {
 213                 cout << "An unknown error has occurred in the SharedListVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 214                 exit(1);
 215         }
 216 }
 217
 218 /***********************************************************************/
 219 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 220         try {
 221                 string groupName, names, name;
 222
 223                 SharedOrderVector* order = new SharedOrderVector();
 224                 order->setLabel(label);
 225
 226                 for(int i=0;i<numBins;i++){
 227                         int binSize = getNumNames(get(i));      //find number of individual in given bin
 228                         names = get(i);
 229                         while (names.find_first_of(',') != -1) {
 230                                 name = names.substr(0,names.find_first_of(','));
 231                                 names = names.substr(names.find_first_of(',')+1, names.length());
 232                                 groupName = groupmap->getGroup(name);
 233                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 234                         }
 235                         //get last name
 236                         groupName = groupmap->getGroup(names);
 237                         order->push_back(i, binSize, groupName);
 238                 }
 239                 random_shuffle(order->begin(), order->end());
 240                 return order;
 241         }
 242         catch(exception& e) {
 243                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 244                 exit(1);
 245         }
 246         catch(...) {
 247                 cout << "An unknown error has occurred in the SharedListVector class function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 248                 exit(1);
 249         }
 250
 251 }
 252 /***********************************************************************/
 253 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 254         try {
 255                 SharedRAbundVector rav(data.size());
 256                 string group, names, name;
 257
 258                 for(int i=0;i<numBins;i++){
 259                         names = get(i);
 260                         while (names.find_first_of(',') != -1) {
 261                                 name = names.substr(0,names.find_first_of(','));
 262                                 names = names.substr(names.find_first_of(',')+1, names.length());
 263                                 group = groupmap->getGroup(name);
 264                                 if (group == groupName) { //this name is in the group you want the vector for.
 265                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 266                                 }
 267                         }
 268
 269                         //get last name
 270                         groupName = groupmap->getGroup(names);
 271                         if (group == groupName) { //this name is in the group you want the vector for.
 272                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 273                         }
 274                 }
 275
 276                 rav.setLabel(label);
 277                 rav.setGroup(groupName);
 278                 return rav;
 279
 280         }
 281         catch(exception& e) {
 282                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 283                 exit(1);
 284         }
 285         catch(...) {
 286                 cout << "An unknown error has occurred in the SharedListVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 287                 exit(1);
 288         }
 289 }
 290
 291 /***********************************************************************/
 292 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 293         try {
 294                 SharedSAbundVector sav;
 295                 SharedRAbundVector rav;
 296
 297                 rav = this->getSharedRAbundVector(groupName);
 298                 sav = rav.getSharedSAbundVector();
 299
 300                 return sav;
 301         }
 302         catch(exception& e) {
 303                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 304                 exit(1);
 305         }
 306         catch(...) {
 307                 cout << "An unknown error has occurred in the SharedListVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 308                 exit(1);
 309         }
 310 }
 311 /***********************************************************************/
 312
 313 OrderVector SharedListVector::getOrderVector(map<string,int>* orderMap = NULL){
 314
 315         try {
 316                 if(orderMap == NULL){
 317                         OrderVector ov;
 318
 319                         for(int i=0;i<data.size();i++){
 320                                 int binSize = getNumNames(data[i]);
 321                                 for(int j=0;j<binSize;j++){
 322                                         ov.push_back(i);
 323                                 }
 324                         }
 325                         random_shuffle(ov.begin(), ov.end());
 326                         ov.setLabel(label);
 327                         ov.getNumBins();
 328
 329                         return ov;
 330
 331                 }
 332                 else{
 333                         OrderVector ov(numSeqs);
 334
 335                         for(int i=0;i<data.size();i++){
 336                                 string listOTU = data[i];
 337                                 int length = listOTU.size();
 338
 339                                 string seqName="";
 340
 341                                 for(int j=0;j<length;j++){
 342
 343                                         if(listOTU[j] != ','){
 344                                                 seqName += listOTU[j];
 345                                         }
 346                                         else{
 347                                                 if(orderMap->count(seqName) == 0){
 348                                                         cerr << seqName << " not found, check *.names file\n";
 349                                                         exit(1);
 350                                                 }
 351
 352                                                 ov.set((*orderMap)[seqName], i);
 353                                                 seqName = "";
 354                                         }
 355                                 }
 356
 357                                 if(orderMap->count(seqName) == 0){
 358                                         cerr << seqName << " not found, check *.names file\n";
 359                                         exit(1);
 360                                 }
 361                                 ov.set((*orderMap)[seqName], i);
 362                         }
 363
 364                         ov.setLabel(label);
 365                         ov.getNumBins();
 366
 367                         return ov;
 368                 }
 369         }
 370         catch(exception& e) {
 371                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 372                 exit(1);
 373         }
 374         catch(...) {
 375                 cout << "An unknown error has occurred in the SharedListVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 376                 exit(1);
 377         }
 378 }
 379
 380 /***********************************************************************/
 381