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(); groupmap = NULL; }
  20
  21 /***********************************************************************/
  22
  23 SharedListVector::SharedListVector(int n):      DataVector(), data(n, "") , maxRank(0), numBins(0), numSeqs(0){globaldata = GlobalData::getInstance(); groupmap = NULL; }
  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                 errorOut(e, "SharedListVector", "SharedListVector");
  48                 exit(1);
  49         }
  50 }
  51
  52 /***********************************************************************/
  53 void SharedListVector::set(int binNumber, string seqNames){
  54         try {
  55                 int nNames_old = getNumNames(data[binNumber]);
  56                 data[binNumber] = seqNames;
  57                 int nNames_new = getNumNames(seqNames);
  58
  59                 if(nNames_old == 0)                     {       numBins++;                              }
  60                 if(nNames_new == 0)                     {       numBins--;                              }
  61                 if(nNames_new > maxRank)        {       maxRank = nNames_new;   }
  62
  63                 numSeqs += (nNames_new - nNames_old);
  64         }
  65         catch(exception& e) {
  66                 errorOut(e, "SharedListVector", "set");
  67                 exit(1);
  68         }
  69 }
  70
  71 /***********************************************************************/
  72
  73 string SharedListVector::get(int index){
  74         return data[index];
  75 }
  76
  77 /***********************************************************************/
  78
  79 void SharedListVector::push_back(string seqNames){
  80         try {
  81                 data.push_back(seqNames);
  82                 int nNames = getNumNames(seqNames);
  83
  84                 numBins++;
  85
  86                 if(nNames > maxRank)    {       maxRank = nNames;       }
  87
  88                 numSeqs += nNames;
  89         }
  90         catch(exception& e) {
  91                 errorOut(e, "SharedListVector", "push_back");
  92                 exit(1);
  93         }
  94 }
  95
  96 /***********************************************************************/
  97
  98 void SharedListVector::resize(int size){
  99         data.resize(size);
 100 }
 101
 102 /***********************************************************************/
 103
 104 int SharedListVector::size(){
 105         return data.size();
 106 }
 107 /***********************************************************************/
 108
 109 void SharedListVector::clear(){
 110         numBins = 0;
 111         maxRank = 0;
 112         numSeqs = 0;
 113         return data.clear();
 114
 115 }
 116
 117 /***********************************************************************/
 118
 119 void SharedListVector::print(ostream& output){
 120         try {
 121                 output << label << '\t' << numBins << '\t';
 122
 123                 for(int i=0;i<data.size();i++){
 124                         if(data[i] != ""){
 125                                 output << data[i] << '\t';
 126                         }
 127                 }
 128                 output << endl;
 129         }
 130         catch(exception& e) {
 131                 errorOut(e, "SharedListVector", "print");
 132                 exit(1);
 133         }
 134 }
 135
 136
 137 /***********************************************************************/
 138
 139 RAbundVector SharedListVector::getRAbundVector(){
 140         try {
 141                 RAbundVector rav;
 142
 143                 for(int i=0;i<data.size();i++){
 144                         int binSize = getNumNames(data[i]);
 145                         rav.push_back(binSize);
 146                 }
 147
 148         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 149         //      sort(rav.rbegin(), rav.rend());
 150         //
 151         //      for(int i=data.size()-1;i>=0;i--){
 152         //              if(rav.get(i) == 0){    rav.pop_back(); }
 153         //              else{
 154         //                      break;
 155         //              }
 156         //      }
 157                 rav.setLabel(label);
 158
 159                 return rav;
 160         }
 161         catch(exception& e) {
 162                 errorOut(e, "SharedListVector", "getRAbundVector");
 163                 exit(1);
 164         }
 165 }
 166
 167 /***********************************************************************/
 168
 169 SAbundVector SharedListVector::getSAbundVector(){
 170         try {
 171                 SAbundVector sav(maxRank+1);
 172
 173                 for(int i=0;i<data.size();i++){
 174                         int binSize = getNumNames(data[i]);
 175                         sav.set(binSize, sav.get(binSize) + 1);
 176                 }
 177                 sav.set(0, 0);
 178                 sav.setLabel(label);
 179
 180                 return sav;
 181         }
 182         catch(exception& e) {
 183                 errorOut(e, "SharedListVector", "getSAbundVector");
 184                 exit(1);
 185         }
 186 }
 187
 188 /***********************************************************************/
 189 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 190         try {
 191                 string groupName, names, name;
 192
 193                 SharedOrderVector* order = new SharedOrderVector();
 194                 order->setLabel(label);
 195
 196                 for(int i=0;i<numBins;i++){
 197                         int binSize = getNumNames(get(i));      //find number of individual in given bin
 198                         names = get(i);
 199                         while (names.find_first_of(',') != -1) {
 200                                 name = names.substr(0,names.find_first_of(','));
 201                                 names = names.substr(names.find_first_of(',')+1, names.length());
 202                                 groupName = groupmap->getGroup(name);
 203
 204                                 if(groupName == "not found") {  mothurOut("Error: Sequence '" + name + "' was not found in the group file, please correct."); mothurOutEndLine();  exit(1); }
 205
 206                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 207                         }
 208                         //get last name
 209                         groupName = groupmap->getGroup(names);
 210                         if(groupName == "not found") {  mothurOut("Error: Sequence '" + names + "' was not found in the group file, please correct."); mothurOutEndLine();  exit(1); }
 211                         order->push_back(i, binSize, groupName);
 212                 }
 213
 214                 random_shuffle(order->begin(), order->end());
 215                 order->updateStats();
 216
 217                 return order;
 218         }
 219         catch(exception& e) {
 220                 errorOut(e, "SharedListVector", "getSharedOrderVector");
 221                 exit(1);
 222         }
 223 }
 224 /***********************************************************************/
 225 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 226         try {
 227                 SharedRAbundVector rav(data.size());
 228                 string group, names, name;
 229
 230                 for(int i=0;i<numBins;i++){
 231                         names = get(i);
 232                         while (names.find_first_of(',') != -1) {
 233                                 name = names.substr(0,names.find_first_of(','));
 234                                 names = names.substr(names.find_first_of(',')+1, names.length());
 235                                 group = groupmap->getGroup(name);
 236                                 if(group == "not found") {      mothurOut("Error: Sequence '" + name + "' was not found in the group file, please correct."); mothurOutEndLine();  exit(1); }
 237                                 if (group == groupName) { //this name is in the group you want the vector for.
 238                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 239                                 }
 240                         }
 241
 242                         //get last name
 243                         groupName = groupmap->getGroup(names);
 244                         if(groupName == "not found") {  mothurOut("Error: Sequence '" + names + "' was not found in the group file, please correct."); mothurOutEndLine();  exit(1); }
 245                         if (group == groupName) { //this name is in the group you want the vector for.
 246                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 247                         }
 248                 }
 249
 250                 rav.setLabel(label);
 251                 rav.setGroup(groupName);
 252
 253                 return rav;
 254
 255         }
 256         catch(exception& e) {
 257                 errorOut(e, "SharedListVector", "getSharedRAbundVector");
 258                 exit(1);
 259         }
 260 }
 261 /***********************************************************************/
 262 vector<SharedRAbundVector*> SharedListVector::getSharedRAbundVector() {
 263         try {
 264                 SharedUtil* util;
 265                 util = new SharedUtil();
 266                 vector<SharedRAbundVector*> lookup;
 267                 vector<SharedRAbundVector*> lookup2;
 268                 map<string, SharedRAbundVector*> finder;
 269                 map<string, SharedRAbundVector*>::iterator it;
 270                 string group, names, name;
 271
 272                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups);
 273                 delete util;
 274
 275                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {
 276                         SharedRAbundVector* temp = new SharedRAbundVector(data.size());
 277                         finder[globaldata->gGroupmap->namesOfGroups[i]] = temp;
 278                         finder[globaldata->gGroupmap->namesOfGroups[i]]->setLabel(label);
 279                         finder[globaldata->gGroupmap->namesOfGroups[i]]->setGroup(globaldata->gGroupmap->namesOfGroups[i]);
 280                         //*temp = getSharedRAbundVector(globaldata->Groups[i]);
 281                         lookup.push_back(finder[globaldata->gGroupmap->namesOfGroups[i]]);
 282                 }
 283 //cout << "after blanks" << endl;
 284                 //fill vectors
 285                 for(int i=0;i<numBins;i++){
 286                         names = get(i);
 287                         while (names.find_first_of(',') != -1) {
 288                                 name = names.substr(0,names.find_first_of(','));
 289                                 names = names.substr(names.find_first_of(',')+1, names.length());
 290                                 group = groupmap->getGroup(name);
 291 //cout << i << '\t' << name << '\t' << group << endl;
 292                                 if(group == "not found") {      mothurOut("Error: Sequence '" + name + "' was not found in the group file, please correct."); mothurOutEndLine();  exit(1); }
 293                                 finder[group]->set(i, finder[group]->getAbundance(i) + 1, group);  //i represents what bin you are in
 294                         }
 295
 296                         //get last name
 297                         group = groupmap->getGroup(names);
 298 //cout << i << '\t' << names << '\t' << group << endl;
 299                         if(group == "not found") {      mothurOut("Error: Sequence '" + names + "' was not found in the group file, please correct."); mothurOutEndLine();  exit(1); }
 300                         finder[group]->set(i, finder[group]->getAbundance(i) + 1, group);  //i represents what bin you are in
 301
 302                 }
 303
 304                 if (globaldata->Groups.size() == globaldata->gGroupmap->namesOfGroups.size()) { //no groups specified
 305                         lookup2 = lookup;
 306                 }else{ //delete unwanted groups
 307                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 308                                 SharedRAbundVector* temp = new SharedRAbundVector(*finder[globaldata->Groups[i]]);
 309                                 lookup2.push_back(temp);
 310                                 delete finder[globaldata->Groups[i]];  //so we don't get dup memory
 311                         }
 312                 }
 313
 314                 return lookup2;
 315         }
 316         catch(exception& e) {
 317                 errorOut(e, "SharedListVector", "getSharedRAbundVector");
 318                 exit(1);
 319         }
 320 }
 321
 322 /***********************************************************************/
 323 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 324         try {
 325                 SharedSAbundVector sav;
 326                 SharedRAbundVector rav;
 327
 328                 rav = this->getSharedRAbundVector(groupName);
 329                 sav = rav.getSharedSAbundVector();
 330
 331                 return sav;
 332         }
 333         catch(exception& e) {
 334                 errorOut(e, "SharedListVector", "getSharedSAbundVector");
 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                                                         mothurOut(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                                         mothurOut(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                 errorOut(e, "SharedListVector", "getOrderVector");
 399                 exit(1);
 400         }
 401 }
 402
 403 /***********************************************************************/
 404