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                 m->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 = m->getNumNames(data[binNumber]);
  56                 data[binNumber] = seqNames;
  57                 int nNames_new = m->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
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "SharedListVector", "set");
  69                 exit(1);
  70         }
  71 }
  72
  73 /***********************************************************************/
  74
  75 string SharedListVector::get(int index){
  76         return data[index];
  77 }
  78
  79 /***********************************************************************/
  80
  81 void SharedListVector::push_back(string seqNames){
  82         try {
  83                 data.push_back(seqNames);
  84                 int nNames = m->getNumNames(seqNames);
  85
  86                 numBins++;
  87
  88                 if(nNames > maxRank)    {       maxRank = nNames;       }
  89
  90                 numSeqs += nNames;
  91         }
  92         catch(exception& e) {
  93                 m->errorOut(e, "SharedListVector", "push_back");
  94                 exit(1);
  95         }
  96 }
  97
  98 /***********************************************************************/
  99
 100 void SharedListVector::resize(int size){
 101         data.resize(size);
 102 }
 103
 104 /***********************************************************************/
 105
 106 int SharedListVector::size(){
 107         return data.size();
 108 }
 109 /***********************************************************************/
 110
 111 void SharedListVector::clear(){
 112         numBins = 0;
 113         maxRank = 0;
 114         numSeqs = 0;
 115         return data.clear();
 116
 117 }
 118
 119 /***********************************************************************/
 120
 121 void SharedListVector::print(ostream& output){
 122         try {
 123                 output << label << '\t' << numBins << '\t';
 124
 125                 for(int i=0;i<data.size();i++){
 126                         if(data[i] != ""){
 127                                 output << data[i] << '\t';
 128                         }
 129                 }
 130                 output << endl;
 131         }
 132         catch(exception& e) {
 133                 m->errorOut(e, "SharedListVector", "print");
 134                 exit(1);
 135         }
 136 }
 137
 138
 139 /***********************************************************************/
 140
 141 RAbundVector SharedListVector::getRAbundVector(){
 142         try {
 143                 RAbundVector rav;
 144
 145                 for(int i=0;i<data.size();i++){
 146                         int binSize = m->getNumNames(data[i]);
 147                         rav.push_back(binSize);
 148                 }
 149
 150         //  This was here before to output data in a nice format, but it screws up the name mapping steps
 151         //      sort(rav.rbegin(), rav.rend());
 152         //
 153         //      for(int i=data.size()-1;i>=0;i--){
 154         //              if(rav.get(i) == 0){    rav.pop_back(); }
 155         //              else{
 156         //                      break;
 157         //              }
 158         //      }
 159                 rav.setLabel(label);
 160
 161                 return rav;
 162         }
 163         catch(exception& e) {
 164                 m->errorOut(e, "SharedListVector", "getRAbundVector");
 165                 exit(1);
 166         }
 167 }
 168
 169 /***********************************************************************/
 170
 171 SAbundVector SharedListVector::getSAbundVector(){
 172         try {
 173                 SAbundVector sav(maxRank+1);
 174
 175                 for(int i=0;i<data.size();i++){
 176                         int binSize = m->getNumNames(data[i]);
 177                         sav.set(binSize, sav.get(binSize) + 1);
 178                 }
 179                 sav.set(0, 0);
 180                 sav.setLabel(label);
 181
 182                 return sav;
 183         }
 184         catch(exception& e) {
 185                 m->errorOut(e, "SharedListVector", "getSAbundVector");
 186                 exit(1);
 187         }
 188 }
 189
 190 /***********************************************************************/
 191 SharedOrderVector* SharedListVector::getSharedOrderVector(){
 192         try {
 193                 string groupName, names, name;
 194
 195                 SharedOrderVector* order = new SharedOrderVector();
 196                 order->setLabel(label);
 197
 198                 for(int i=0;i<numBins;i++){
 199                         int binSize = m->getNumNames(get(i));   //find number of individual in given bin
 200                         names = get(i);
 201                         while (names.find_first_of(',') != -1) {
 202                                 name = names.substr(0,names.find_first_of(','));
 203                                 names = names.substr(names.find_first_of(',')+1, names.length());
 204                                 groupName = groupmap->getGroup(name);
 205
 206                                 if(groupName == "not found") {  m->mothurOut("Error: Sequence '" + name + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 207
 208                                 order->push_back(i, binSize, groupName);  //i represents what bin you are in
 209                         }
 210                         //get last name
 211                         groupName = groupmap->getGroup(names);
 212                         if(groupName == "not found") {  m->mothurOut("Error: Sequence '" + names + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 213                         order->push_back(i, binSize, groupName);
 214                 }
 215
 216                 random_shuffle(order->begin(), order->end());
 217                 order->updateStats();
 218
 219                 return order;
 220         }
 221         catch(exception& e) {
 222                 m->errorOut(e, "SharedListVector", "getSharedOrderVector");
 223                 exit(1);
 224         }
 225 }
 226 /***********************************************************************/
 227 SharedRAbundVector SharedListVector::getSharedRAbundVector(string groupName) {
 228         try {
 229                 SharedRAbundVector rav(data.size());
 230                 string group, names, name;
 231
 232                 for(int i=0;i<numBins;i++){
 233                         names = get(i);
 234                         while (names.find_first_of(',') != -1) {
 235                                 name = names.substr(0,names.find_first_of(','));
 236                                 names = names.substr(names.find_first_of(',')+1, names.length());
 237                                 group = groupmap->getGroup(name);
 238                                 if(group == "not found") {      m->mothurOut("Error: Sequence '" + name + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 239                                 if (group == groupName) { //this name is in the group you want the vector for.
 240                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 241                                 }
 242                         }
 243
 244                         //get last name
 245                         groupName = groupmap->getGroup(names);
 246                         if(groupName == "not found") {  m->mothurOut("Error: Sequence '" + names + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 247                         if (group == groupName) { //this name is in the group you want the vector for.
 248                                         rav.set(i, rav.getAbundance(i) + 1, group);  //i represents what bin you are in
 249                         }
 250                 }
 251
 252                 rav.setLabel(label);
 253                 rav.setGroup(groupName);
 254
 255                 return rav;
 256
 257         }
 258         catch(exception& e) {
 259                 m->errorOut(e, "SharedListVector", "getSharedRAbundVector");
 260                 exit(1);
 261         }
 262 }
 263 /***********************************************************************/
 264 vector<SharedRAbundVector*> SharedListVector::getSharedRAbundVector() {
 265         try {
 266                 SharedUtil* util;
 267                 util = new SharedUtil();
 268                 vector<SharedRAbundVector*> lookup;  //contains just the groups the user selected
 269                 map<string, SharedRAbundVector*> finder;  //contains all groups in groupmap
 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                         if (m->inUsersGroups(globaldata->gGroupmap->namesOfGroups[i], globaldata->Groups)) {  //if this group is in user groups
 281                                 lookup.push_back(finder[globaldata->gGroupmap->namesOfGroups[i]]);
 282                         }
 283                 }
 284
 285                 //fill vectors
 286                 for(int i=0;i<numBins;i++){
 287                         names = get(i);
 288                         int nameLength = names.size();
 289                         string seqName = "";
 290
 291                         for(int j=0;j<nameLength;j++){
 292                                 if(names[j] == ','){
 293                                         group = groupmap->getGroup(seqName);
 294                                         if(group == "not found") {      m->mothurOut("Error: Sequence '" + seqName + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 295                                         finder[group]->set(i, finder[group]->getAbundance(i) + 1, group);  //i represents what bin you are in
 296
 297                                         seqName = "";
 298                                 }
 299                                 else{
 300                                         seqName += names[j];
 301                                 }
 302                         }
 303                         group = groupmap->getGroup(seqName);
 304                         if(group == "not found") {      m->mothurOut("Error: Sequence '" + seqName + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 305                         finder[group]->set(i, finder[group]->getAbundance(i) + 1, group);  //i represents what bin you are in
 306
 307
 308
 309 //                      while (names.find_first_of(',') != -1) {
 310 //                              name = names.substr(0,names.find_first_of(','));
 311 //                              names = names.substr(names.find_first_of(',')+1, names.length());
 312 //                              group = groupmap->getGroup(name);
 313 //                              if(group == "not found") {      m->mothurOut("Error: Sequence '" + name + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 314 //                              finder[group]->set(i, finder[group]->getAbundance(i) + 1, group);  //i represents what bin you are in
 315 //                      }
 316
 317                         //get last name
 318 //                      group = groupmap->getGroup(names);
 319 //                      if(group == "not found") {      m->mothurOut("Error: Sequence '" + names + "' was not found in the group file, please correct."); m->mothurOutEndLine();  exit(1); }
 320 //                      finder[group]->set(i, finder[group]->getAbundance(i) + 1, group);  //i represents what bin you are in
 321
 322                 }
 323
 324                 return lookup;
 325         }
 326         catch(exception& e) {
 327                 m->errorOut(e, "SharedListVector", "getSharedRAbundVector");
 328                 exit(1);
 329         }
 330 }
 331
 332 /***********************************************************************/
 333 SharedSAbundVector SharedListVector::getSharedSAbundVector(string groupName) {
 334         try {
 335                 SharedSAbundVector sav;
 336                 SharedRAbundVector rav;
 337
 338                 rav = this->getSharedRAbundVector(groupName);
 339                 sav = rav.getSharedSAbundVector();
 340
 341                 return sav;
 342         }
 343         catch(exception& e) {
 344                 m->errorOut(e, "SharedListVector", "getSharedSAbundVector");
 345                 exit(1);
 346         }
 347 }
 348 /***********************************************************************/
 349
 350 OrderVector SharedListVector::getOrderVector(map<string,int>* orderMap = NULL){
 351
 352         try {
 353                 if(orderMap == NULL){
 354                         OrderVector ov;
 355
 356                         for(int i=0;i<data.size();i++){
 357                                 int binSize = m->getNumNames(data[i]);
 358                                 for(int j=0;j<binSize;j++){
 359                                         ov.push_back(i);
 360                                 }
 361                         }
 362                         random_shuffle(ov.begin(), ov.end());
 363                         ov.setLabel(label);
 364                         ov.getNumBins();
 365
 366                         return ov;
 367
 368                 }
 369                 else{
 370                         OrderVector ov(numSeqs);
 371
 372                         for(int i=0;i<data.size();i++){
 373                                 string listOTU = data[i];
 374                                 int length = listOTU.size();
 375
 376                                 string seqName="";
 377
 378                                 for(int j=0;j<length;j++){
 379
 380                                         if(listOTU[j] != ','){
 381                                                 seqName += listOTU[j];
 382                                         }
 383                                         else{
 384                                                 if(orderMap->count(seqName) == 0){
 385                                                         m->mothurOut(seqName + " not found, check *.names file\n");
 386                                                         exit(1);
 387                                                 }
 388
 389                                                 ov.set((*orderMap)[seqName], i);
 390                                                 seqName = "";
 391                                         }
 392                                 }
 393
 394                                 if(orderMap->count(seqName) == 0){
 395                                         m->mothurOut(seqName + " not found, check *.names file\n");
 396                                         exit(1);
 397                                 }
 398                                 ov.set((*orderMap)[seqName], i);
 399                         }
 400
 401                         ov.setLabel(label);
 402                         ov.getNumBins();
 403
 404                         return ov;
 405                 }
 406         }
 407         catch(exception& e) {
 408                 m->errorOut(e, "SharedListVector", "getOrderVector");
 409                 exit(1);
 410         }
 411 }
 412
 413 /***********************************************************************/
 414