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 #include "sabundvector.hpp"
  11 #include "rabundvector.hpp"
  12 #include "ordervector.hpp"
  13 #include "sharedlistvector.h"
  14 #include "sharedordervector.h"
  15 #include "sharedutilities.h"
  16
  17 /***********************************************************************/
  18
  19 SharedListVector::SharedListVector() : DataVector(), maxRank(0), numBins(0), numSeqs(0){globaldata = GlobalData::getInstance();}
  20
  21 /***********************************************************************/
  22
  23 SharedListVector::SharedListVector(int n):      DataVector(), data(n, "") , maxRank(0), numBins(0), numSeqs(0){globaldata = GlobalData::getInstance();}
  24
  25 /***********************************************************************/
  26 SharedListVector::SharedListVector(ifstream& f) : DataVector(), maxRank(0), numBins(0), numSeqs(0) {
  27         try {
  28                 globaldata = GlobalData::getInstance();
  29
  30                 //set up groupmap for later.
  31                 groupmap = new GroupMap(globaldata->getGroupFile());
  32                 groupmap->readMap();
  33
  34                 int hold;
  35                 string inputData;
  36                 f >> label >> hold;
  37
  38                 data.assign(hold, "");
  39
  40                 for(int i=0;i<hold;i++){
  41                         f >> inputData;
  42                         set(i, inputData);
  43                 }
  44
  45         }
  46         catch(exception& e) {
  47                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  48                 exit(1);
  49         }
  50         catch(...) {
  51                 cout << "An unknown error has occurred in the SharedListVector class function SharedListVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  52                 exit(1);
  53         }
  54 }
  55
  56 /***********************************************************************/
  57 void SharedListVector::set(int binNumber, string seqNames){
  58         try {
  59                 int nNames_old = getNumNames(data[binNumber]);
  60                 data[binNumber] = seqNames;
  61                 int nNames_new = getNumNames(seqNames);
  62
  63                 if(nNames_old == 0)                     {       numBins++;                              }
  64                 if(nNames_new == 0)                     {       numBins--;                              }
  65                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  66
  67                 numSeqs += (nNames_new - nNames_old);
  68         }
  69         catch(exception& e) {
  70                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  71                 exit(1);
  72         }
  73         catch(...) {
  74                 cout << "An unknown error has occurred in the SharedListVector class function set. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  75                 exit(1);
  76         }
  77 }
  78
  79 /***********************************************************************/
  80
  81 string SharedListVector::get(int index){
  82         return data[index];
  83 }
  84
  85 /***********************************************************************/
  86
  87 void SharedListVector::push_back(string seqNames){
  88         try {
  89                 data.push_back(seqNames);
  90                 int nNames = getNumNames(seqNames);
  91
  92                 numBins++;
  93
  94                 if(nNames > maxRank)    {       maxRank = nNames;       }
  95
  96                 numSeqs += nNames;
  97         }
  98         catch(exception& e) {
  99                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 100                 exit(1);
 101         }
 102         catch(...) {
 103                 cout << "An unknown error has occurred in the SharedListVector class function push_back. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 104                 exit(1);
 105         }
 106 }
 107
 108 /***********************************************************************/
 109
 110 void SharedListVector::resize(int size){
 111         data.resize(size);
 112 }
 113
 114 /***********************************************************************/
 115
 116 int SharedListVector::size(){
 117         return data.size();
 118 }
 119 /***********************************************************************/
 120
 121 void SharedListVector::clear(){
 122         numBins = 0;
 123         maxRank = 0;
 124         numSeqs = 0;
 125         return data.clear();
 126
 127 }
 128
 129 /***********************************************************************/
 130
 131 void SharedListVector::print(ostream& output){
 132         try {
 133                 output << label << '\t' << numBins << '\t';
 134
 135                 for(int i=0;i<data.size();i++){
 136                         if(data[i] != ""){
 137                                 output << data[i] << '\t';
 138                         }
 139                 }
 140                 output << endl;
 141         }
 142         catch(exception& e) {
 143                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 144                 exit(1);
 145         }
 146         catch(...) {
 147                 cout << "An unknown error has occurred in the SharedListVector class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 148                 exit(1);
 149         }
 150 }
 151
 152
 153 /***********************************************************************/
 154
 155 RAbundVector SharedListVector::getRAbundVector(){
 156         try {
 157                 RAbundVector rav;
 158
 159                 for(int i=0;i<data.size();i++){
 160                         int binSize = getNumNames(data[i]);
 161                         rav.push_back(binSize);
 162                 }
 163
 164         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 165         //      sort(rav.rbegin(), rav.rend());
 166         //
 167         //      for(int i=data.size()-1;i>=0;i--){
 168         //              if(rav.get(i) == 0){    rav.pop_back(); }
 169         //              else{
 170         //                      break;
 171         //              }
 172         //      }
 173                 rav.setLabel(label);
 174
 175                 return rav;
 176         }
 177         catch(exception& e) {
 178                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 179                 exit(1);
 180         }
 181         catch(...) {
 182                 cout << "An unknown error has occurred in the SharedListVector class function getRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 183                 exit(1);
 184         }
 185 }
 186
 187 /***********************************************************************/
 188
 189 SAbundVector SharedListVector::getSAbundVector(){
 190         try {
 191                 SAbundVector sav(maxRank+1);
 192
 193                 for(int i=0;i<data.size();i++){
 194                         int binSize = getNumNames(data[i]);
 195                         sav.set(binSize, sav.get(binSize) + 1);
 196                 }
 197                 sav.set(0, 0);
 198                 sav.setLabel(label);
 199
 200                 return sav;
 201         }
 202         catch(exception& e) {
 203                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 204                 exit(1);
 205         }
 206         catch(...) {
 207                 cout << "An unknown error has occurred in the SharedListVector class function getSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 208                 exit(1);
 209         }
 210 }
 211
 212 /***********************************************************************/
 213 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 214         try {
 215                 string groupName, names, name;
 216
 217                 SharedOrderVector* order = new SharedOrderVector();
 218                 order->setLabel(label);
 219
 220                 for(int i=0;i<numBins;i++){
 221                         int binSize = getNumNames(get(i));      //find number of individual in given bin
 222                         names = get(i);
 223                         while (names.find_first_of(',') != -1) {
 224                                 name = names.substr(0,names.find_first_of(','));
 225                                 names = names.substr(names.find_first_of(',')+1, names.length());
 226                                 groupName = groupmap->getGroup(name);
 227                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 228                         }
 229                         //get last name
 230                         groupName = groupmap->getGroup(names);
 231                         order->push_back(i, binSize, groupName);
 232                 }
 233
 234                 random_shuffle(order->begin(), order->end());
 235                 order->updateStats();
 236
 237                 return order;
 238         }
 239         catch(exception& e) {
 240                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 241                 exit(1);
 242         }
 243         catch(...) {
 244                 cout << "An unknown error has occurred in the SharedListVector class function getSharedOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 245                 exit(1);
 246         }
 247
 248 }
 249 /***********************************************************************/
 250 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 251         try {
 252                 SharedRAbundVector rav(data.size());
 253                 string group, names, name;
 254
 255                 for(int i=0;i<numBins;i++){
 256                         names = get(i);
 257                         while (names.find_first_of(',') != -1) {
 258                                 name = names.substr(0,names.find_first_of(','));
 259                                 names = names.substr(names.find_first_of(',')+1, names.length());
 260                                 group = groupmap->getGroup(name);
 261                                 if (group == groupName) { //this name is in the group you want the vector for.
 262                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 263                                 }
 264                         }
 265
 266                         //get last name
 267                         groupName = groupmap->getGroup(names);
 268                         if (group == groupName) { //this name is in the group you want the vector for.
 269                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 270                         }
 271                 }
 272
 273                 rav.setLabel(label);
 274                 rav.setGroup(groupName);
 275
 276                 return rav;
 277
 278         }
 279         catch(exception& e) {
 280                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 281                 exit(1);
 282         }
 283         catch(...) {
 284                 cout << "An unknown error has occurred in the SharedListVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 285                 exit(1);
 286         }
 287 }
 288 /***********************************************************************/
 289 vector<SharedRAbundVector*> SharedListVector::getSharedRAbundVector() {
 290         try {
 291                 SharedUtil* util;
 292                 util = new SharedUtil();
 293                 vector<SharedRAbundVector*> lookup;
 294
 295                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups);
 296
 297                 delete util;
 298
 299                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 300                         SharedRAbundVector* temp = new SharedRAbundVector();
 301                         *temp = getSharedRAbundVector(globaldata->Groups[i]);
 302                         lookup.push_back(temp);
 303                 }
 304
 305                 return lookup;
 306         }
 307         catch(exception& e) {
 308                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 309                 exit(1);
 310         }
 311         catch(...) {
 312                 cout << "An unknown error has occurred in the SharedListVector class function getSharedRAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 313                 exit(1);
 314         }
 315
 316 }
 317
 318 /***********************************************************************/
 319 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 320         try {
 321                 SharedSAbundVector sav;
 322                 SharedRAbundVector rav;
 323
 324                 rav = this->getSharedRAbundVector(groupName);
 325                 sav = rav.getSharedSAbundVector();
 326
 327                 return sav;
 328         }
 329         catch(exception& e) {
 330                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 331                 exit(1);
 332         }
 333         catch(...) {
 334                 cout << "An unknown error has occurred in the SharedListVector class function getSharedSAbundVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 335                 exit(1);
 336         }
 337 }
 338 /***********************************************************************/
 339
 340 OrderVector SharedListVector::getOrderVector(map<string,int>* orderMap = NULL){
 341
 342         try {
 343                 if(orderMap == NULL){
 344                         OrderVector ov;
 345
 346                         for(int i=0;i<data.size();i++){
 347                                 int binSize = getNumNames(data[i]);
 348                                 for(int j=0;j<binSize;j++){
 349                                         ov.push_back(i);
 350                                 }
 351                         }
 352                         random_shuffle(ov.begin(), ov.end());
 353                         ov.setLabel(label);
 354                         ov.getNumBins();
 355
 356                         return ov;
 357
 358                 }
 359                 else{
 360                         OrderVector ov(numSeqs);
 361
 362                         for(int i=0;i<data.size();i++){
 363                                 string listOTU = data[i];
 364                                 int length = listOTU.size();
 365
 366                                 string seqName="";
 367
 368                                 for(int j=0;j<length;j++){
 369
 370                                         if(listOTU[j] != ','){
 371                                                 seqName += listOTU[j];
 372                                         }
 373                                         else{
 374                                                 if(orderMap->count(seqName) == 0){
 375                                                         cerr << seqName << " not found, check *.names file\n";
 376                                                         exit(1);
 377                                                 }
 378
 379                                                 ov.set((*orderMap)[seqName], i);
 380                                                 seqName = "";
 381                                         }
 382                                 }
 383
 384                                 if(orderMap->count(seqName) == 0){
 385                                         cerr << seqName << " not found, check *.names file\n";
 386                                         exit(1);
 387                                 }
 388                                 ov.set((*orderMap)[seqName], i);
 389                         }
 390
 391                         ov.setLabel(label);
 392                         ov.getNumBins();
 393
 394                         return ov;
 395                 }
 396         }
 397         catch(exception& e) {
 398                 cout << "Standard Error: " << e.what() << " has occurred in the SharedListVector class Function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 399                 exit(1);
 400         }
 401         catch(...) {
 402                 cout << "An unknown error has occurred in the SharedListVector class function getOrderVector. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 403                 exit(1);
 404         }
 405 }
 406
 407 /***********************************************************************/
 408