globaldata.cpp

   1 #include "globaldata.hpp"
   2 #include "tree.h"
   3 #include "sparsematrix.hpp"
   4
   5 /*******************************************************/
   6
   7 /******************************************************/
   8 GlobalData* GlobalData::getInstance() {
   9         if( _uniqueInstance == 0 ) {
  10                 _uniqueInstance = new GlobalData();
  11         }
  12         return _uniqueInstance;
  13 }
  14 /*******************************************************/
  15
  16 /******************************************************/
  17 //This function parses through the option string of the command to remove its parameters
  18 void GlobalData::parseGlobalData(string commandString, string optionText){
  19         try {
  20                 commandName = commandString; //save command name to be used by other classes
  21
  22                 //set all non filename paramters to default
  23                 reset();
  24
  25                 //clears out data from previous read
  26                 if ((commandName == "read.dist") || (commandName == "read.otu") || (commandName == "read.tree")) {
  27                         clear();
  28                         gGroupmap = NULL;
  29                         gTree.clear();
  30                         Treenames.clear();
  31                         labels.clear(); lines.clear(); groups.clear();
  32                         allLines = 1;
  33                 }
  34
  35                 //saves help request
  36                 if (commandName =="help") {
  37                         helpRequest = optionText;
  38                 }
  39
  40                 if (commandName == "libshuff") {
  41                         iters = "10000";
  42                         cutoff = "1.0";
  43                 }
  44
  45                 //set default value for cutoff
  46                 if (commandName == "dist.seq") {        cutoff = "1.0";         }
  47
  48                 string key, value;
  49                 //reads in parameters and values
  50                 if((optionText != "") && (commandName != "help")){
  51                         while((optionText.find_first_of(',') != -1)) {  //while there are parameters
  52                                 splitAtComma(value, optionText);
  53                                 splitAtEquals(key, value);
  54
  55                                 if (key == "phylip" )   { phylipfile = value; inputFileName = value; fileroot = value; format = "phylip";       }
  56                                 if (key == "column" )   { columnfile = value; inputFileName = value; fileroot = value; format = "column";       }
  57                                 if (key == "list" )             { listfile = value; inputFileName = value; fileroot = value; format = "list";           }
  58                                 if (key == "rabund" )   { rabundfile = value; inputFileName = value; fileroot = value; format = "rabund";       }
  59                                 if (key == "sabund" )   { sabundfile = value; inputFileName = value; fileroot = value; format = "sabund";       }
  60                                 if (key == "fasta" )    { fastafile = value; inputFileName = value; fileroot = value; format = "fasta";         }
  61                                 if (key == "nexus" )    { nexusfile = value; inputFileName = value; fileroot = value; format = "nexus";         }
  62                                 if (key == "clustal" )  { clustalfile = value; inputFileName = value; fileroot = value; format = "clustal"; }
  63                                 if (key == "tree" )             { treefile = value; inputFileName = value; fileroot = value; format = "tree";           }
  64                                 if (key == "shared" )   { sharedfile = value; inputFileName = value; fileroot = value; format = "sharedfile";   }
  65                                 if (key == "name" )             { namefile = value;             }
  66                                 if (key == "order" )    { orderfile = value;    }
  67                                 if (key == "group" )    { groupfile = value;    }
  68                                 if (key == "cutoff" )           { cutoff = value;               }
  69                                 if (key == "precision" )        { precision = value;    }
  70                                 if (key == "iters" )            { iters = value;                }
  71                                 if (key == "jumble" )           { jumble = value;               }
  72                                 if (key == "freq" )                     { freq = value;                 }
  73                                 if (key == "method" )           { method = value;               }
  74                                 if (key == "fileroot" )         { fileroot = value;             }
  75                                 if (key == "abund" )        { abund = value;        }
  76                                 if (key == "random" )           { randomtree = value;   }
  77                                 if (key == "calc")                      { calc = value;                 }
  78                                 if (key == "step")                      { step = value;                 }
  79                                 if (key == "form")                      { form = value;                 }
  80                                 if (key == "sorted")            { sorted = value;               }
  81                                 if (key == "vertical")          { vertical = value;             }
  82                                 if (key == "trump")                 { trump = value;            }
  83                                 if (key == "filter")            { filter = value;               }
  84                                 if (key == "soft")                  { soft = value;                 }
  85                                 if (key == "scale")                     { scale = value;                }
  86                                 if (key == "ends" )                     { ends = value;                 }
  87                                 if (key == "processors" )       { processors = value;   }
  88
  89
  90
  91
  92
  93                                 if (key == "line") {//stores lines to be used in a set
  94                                         lines.clear();
  95                                         labels.clear();
  96                                         line = value;
  97                                         label = "";
  98                                         splitAtDash(value, lines);
  99                                         allLines = 0;
 100                                 }
 101                                 if (key == "label") {//stores labels to be used in a set
 102                                         labels.clear();
 103                                         lines.clear();
 104                                         label = value;
 105                                         line = "";
 106                                         splitAtDash(value, labels);
 107                                         allLines = 0;
 108                                 }
 109
 110                                 if (key == "groups") {//stores groups to be used in a vector
 111                                         Groups.clear();
 112                                         groups = value;
 113                                         splitAtDash(value, Groups);
 114                                 }
 115
 116                         }
 117
 118                         //saves the last parameter
 119                         value = optionText;
 120                         splitAtEquals(key, value);
 121                         if (key == "phylip" )   { phylipfile = value; inputFileName = value; fileroot = value; format = "phylip";       }
 122                         if (key == "column" )   { columnfile = value; inputFileName = value; fileroot = value; format = "column";       }
 123                         if (key == "list" )             { listfile = value; inputFileName = value; fileroot = value; format = "list";           }
 124                         if (key == "rabund" )   { rabundfile = value; inputFileName = value; fileroot = value; format = "rabund";       }
 125                         if (key == "sabund" )   { sabundfile = value; inputFileName = value; fileroot = value; format = "sabund";       }
 126                         if (key == "fasta" )    { fastafile = value; inputFileName = value; fileroot = value; format = "fasta";         }
 127                         if (key == "nexus" )    { nexusfile = value; inputFileName = value; fileroot = value; format = "nexus";         }
 128                         if (key == "clustal" )  { clustalfile = value; inputFileName = value; fileroot = value; format = "clustal"; }
 129                         if (key == "tree" )             { treefile = value; inputFileName = value; fileroot = value; format = "tree";           }
 130                         if (key == "shared" )   { sharedfile = value; inputFileName = value; fileroot = value; format = "sharedfile";   }
 131                         if (key == "name" )             { namefile = value;             }
 132                         if (key == "order" )    { orderfile = value;    }
 133                         if (key == "group" )    { groupfile = value;    }
 134                         if (key == "cutoff" )           { cutoff = value;               }
 135                         if (key == "precision" )        { precision = value;    }
 136                         if (key == "iters" )            { iters = value;                }
 137                         if (key == "jumble" )           { jumble = value;               }
 138                         if (key == "freq" )                     { freq = value;                 }
 139                         if (key == "method" )           { method = value;               }
 140                         if (key == "fileroot" )         { fileroot = value;             }
 141                         if (key == "abund" )        { abund = value;        }
 142                         if (key == "random" )           { randomtree = value;   }
 143                         if (key == "calc")                      { calc = value;                 }
 144                         if (key == "step")                      { step = value;                 }
 145                         if (key == "form")                      { form = value;                 }
 146                         if (key == "sorted")            { sorted = value;               }
 147                         if (key == "vertical")          { vertical = value;             }
 148                         if (key == "trump")                 { trump = value;            }
 149                         if (key == "filter")            { filter = value;               }
 150                         if (key == "soft")                  { soft = value;                 }
 151                         if (key == "scale")                     { scale = value;                }
 152                         if (key == "ends" )                     { ends = value;                 }
 153                         if (key == "processors" )       { processors = value;   }
 154
 155
 156                         if (key == "line") {//stores lines to be used in a vector
 157                                 lines.clear();
 158                                 labels.clear();
 159                                 line = value;
 160                                 label = "";
 161                                 if (line != "all") {  splitAtDash(value, lines);  allLines = 0;  }
 162                                 else { allLines = 1;  }
 163                         }
 164
 165                         if (key == "label") {//stores lines to be used in a vector
 166                                 labels.clear();
 167                                 lines.clear();
 168                                 label = value;
 169                                 line = "";
 170                                 if (label != "all") {  splitAtDash(value, labels);  allLines = 0;  }
 171                                 else { allLines = 1;  }
 172                         }
 173
 174                         if (key == "groups") {//stores groups to be used in a vector
 175                                         Groups.clear();
 176                                         groups = value;
 177                                         splitAtDash(value, Groups);
 178                         }
 179                 }
 180
 181                 //set format for shared
 182                 if ((listfile != "") && (groupfile != "")) { format = "shared"; }
 183                 if ((phylipfile != "") && (groupfile != "")) { format = "matrix"; }
 184
 185                 //input defaults for calculators
 186                 if (commandName == "collect.single") {
 187
 188                         if ((calc == "default") || (calc == "")) { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson"; }
 189                         Estimators.clear();
 190                         splitAtDash(calc, Estimators);
 191                 }
 192                 if (commandName == "rarefaction.single") {
 193                         if ((calc == "default") || (calc == "")) { calc = "sobs"; }
 194                         Estimators.clear();
 195                         splitAtDash(calc, Estimators);
 196                 }
 197                 if (commandName == "collect.shared") {
 198
 199                         if ((calc == "default") || (calc == "")) { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan"; }
 200                         Estimators.clear();
 201                         splitAtDash(calc, Estimators);
 202                 }
 203                 if (commandName == "summary.single") {
 204                         if ((calc == "default") || (calc == "")) { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson"; }
 205                         Estimators.clear();
 206                         splitAtDash(calc, Estimators);
 207                 }
 208                 if (commandName == "summary.shared") {
 209                         if ((calc == "default") || (calc == "")) { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan"; }
 210                         Estimators.clear();
 211                         splitAtDash(calc, Estimators);
 212                 }
 213                 if (commandName == "rarefaction.shared") {
 214                         if ((calc == "default") || (calc == "")) { calc = "sharedobserved"; }
 215                         Estimators.clear();
 216                         splitAtDash(calc, Estimators);
 217                 }
 218                 if (commandName == "dist.seq") {
 219                         if ((calc == "default") || (calc == "")) {  calc = "onegap";  }
 220                         Estimators.clear();
 221                         splitAtDash(calc, Estimators);
 222                 }
 223                 if (commandName == "venn") {
 224                         if ((calc == "default") || (calc == "")) {
 225                                 if (format == "list") { calc = "sobs"; }
 226                                 else { calc = "sharedsobs"; }
 227                         }
 228                         Estimators.clear();
 229                         splitAtDash(calc, Estimators);
 230                 }
 231                 if ((commandName == "tree.shared") || (commandName == "bootstrap.shared")) {
 232                         if (calc != "") {
 233                                 Estimators.clear();
 234                                 splitAtDash(calc, Estimators);
 235                         }else { cout << "You have not specified any calculators." << endl; }
 236                 }
 237
 238
 239                 //if you have done a read.otu with a groupfile but don't want to use it anymore because you want to do single commands
 240                 if ((commandName == "collect.single") || (commandName == "rarefaction.single") || (commandName == "summary.single")) {
 241                         if (listfile != "") { format = "list"; }
 242                         else if (sabundfile != "") { format = "sabund"; }
 243                         else if (rabundfile != "") { format = "rabund"; }
 244                 }
 245         }
 246         catch(exception& e) {
 247                 cout << "Standard Error: " << e.what() << " has occurred in the GlobalData class Function parseGlobalData. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 248                 exit(1);
 249         }
 250         catch(...) {
 251                 cout << "An unknown error has occurred in the GlobalData class function parseGlobalData. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 252                 exit(1);
 253         }
 254 }
 255 /*******************************************************/
 256
 257 /******************************************************/
 258 // These functions give you the option parameters of the commands
 259 string GlobalData::getPhylipFile()              {       return phylipfile;      }
 260 string GlobalData::getColumnFile()              {       return columnfile;      }
 261 string GlobalData::getListFile()                {       return listfile;        }
 262 string GlobalData::getRabundFile()              {       return rabundfile;      }
 263 string GlobalData::getSabundFile()              {       return sabundfile;      }
 264 string GlobalData::getNameFile()                {       return namefile;        }
 265 string GlobalData::getGroupFile()               {       return groupfile;       }
 266 string GlobalData::getOrderFile()               {       return orderfile;       }
 267 string GlobalData::getTreeFile()                {       return treefile;        }
 268 string GlobalData::getSharedFile()              {       return sharedfile;      }
 269 string GlobalData::getFastaFile()               {       return fastafile;       }
 270 string GlobalData::getNexusFile()               {       return nexusfile;       }
 271 string GlobalData::getClustalFile()     {   return clustalfile; }
 272 string GlobalData::getCutOff()                  {       return cutoff;          }
 273 string GlobalData::getFormat()                  {       return format;          }
 274 string GlobalData::getPrecision()               {       return precision;       }
 275 string GlobalData::getMethod()                  {       return method;          }
 276 string GlobalData::getFileRoot()                {       return fileroot;        }
 277 string GlobalData::getIters()                   {       return iters;           }
 278 string GlobalData::getJumble()                  {       return jumble;          }
 279 string GlobalData::getFreq()                    {       return freq;            }
 280 string GlobalData::getAbund()           {   return abund;       }
 281 string GlobalData::getRandomTree()              {       return randomtree;      }
 282 string GlobalData::getGroups()                  {       return groups;          }
 283 string GlobalData::getStep()                    {       return step;            }
 284 string GlobalData::getForm()                    {       return form;            }
 285 string GlobalData::getSorted()                  {       return sorted;          }
 286 string GlobalData::getTrump()                   {   return trump;       }
 287 string GlobalData::getSoft()                    {   return soft;                }
 288 string GlobalData::getFilter()                  {   return filter;              }
 289 string GlobalData::getScale()                   {       return scale;           }
 290 string GlobalData::getEnds()                    {   return ends;                }
 291 string GlobalData::getProcessors()              {       return processors;      }
 292
 293 void GlobalData::setListFile(string file)               {       listfile = file;        inputFileName = file;}
 294 void GlobalData::setGroupFile(string file)              {       groupfile = file;       }
 295 void GlobalData::setRabundFile(string file)             {       rabundfile = file;      inputFileName = file;}
 296 void GlobalData::setSabundFile(string file)             {       sabundfile = file;      inputFileName = file;}
 297 void GlobalData::setPhylipFile(string file)             {       phylipfile = file;    inputFileName = file;}
 298 void GlobalData::setColumnFile(string file)             {       columnfile = file;    inputFileName = file;}
 299 void GlobalData::setSharedFile(string file)             {       sharedfile = file;      inputFileName = file; fileroot = file;}
 300 void GlobalData::setNameFile(string file)               {       namefile = file;                }
 301 void GlobalData::setFormat(string Format)               {       format = Format;                }
 302 void GlobalData::setRandomTree(string Random)   {       randomtree = Random;    }
 303 void GlobalData::setGroups(string g)                    {       groups = g;                             }
 304 void GlobalData::setCalc(string Calc)                   {       calc = Calc;                    }
 305 void GlobalData::setEnds(string e)                              {   ends = e;                           }
 306 void GlobalData::setProcessors(string p)                {       processors = p;                 }
 307
 308
 309 /*******************************************************/
 310
 311 /******************************************************/
 312 GlobalData::GlobalData() {
 313         //option definitions should go here...
 314         helpRequest = "";
 315         clear();
 316         gListVector == NULL;
 317         gSparseMatrix == NULL;
 318 }
 319 /*******************************************************/
 320
 321 /******************************************************/
 322 void GlobalData::clear() {
 323         //option definitions should go here...
 324         phylipfile              =       "";
 325         columnfile              =       "";
 326         listfile                =       "";
 327         rabundfile              =       "";
 328         sabundfile              =       "";
 329         namefile                =       "";
 330         groupfile               =       "";
 331         orderfile               =       "";
 332         fastafile               =   "";
 333         nexusfile               =   "";
 334         clustalfile             =   "";
 335         treefile                =       "";
 336         sharedfile              =       "";
 337         cutoff                  =       "10.00";
 338         format                  =       "";
 339         precision               =       "100";
 340         iters                   =       "1000";
 341         line                    =   "";
 342         label                   =       "";
 343         groups                  =       "";
 344         jumble                  =       "1";    //0 means don't jumble, 1 means jumble.
 345         randomtree              =       "";  //"" means user will enter some user trees, "outputfile" means they just want the random tree distribution to be outputted to outputfile.
 346         freq                    =       "100";
 347         method                  =       "furthest";
 348         fileroot                =       "";
 349         abund           =   "10";
 350         step                    =       "0.01";
 351         form                    =       "integral";
 352         sorted                  =       "T";  //F means don't sort, T means sort.
 353         vertical        =   "";
 354         trump           =   "";
 355         filter          =   "";
 356         soft            =   "";
 357         scale                   =       "log10";
 358         ends                    =   "T";  //yes
 359         processors              =       "1";
 360
 361 }
 362
 363 //*******************************************************/
 364
 365 /******************************************************/
 366 void GlobalData::reset() {
 367         cutoff                  =       "10.00";
 368         precision               =       "100";
 369         iters                   =       "1000";
 370         groups                  =       "";
 371         jumble                  =       "1";    //0 means don't jumble, 1 means jumble.
 372         sorted                  =       "T";  //F means don't sort, T means sort.
 373         randomtree              =       "";  //"" means user will enter some user trees, "outputfile" means they just want the random tree distribution to be outputted to outputfile.
 374         freq                    =       "100";
 375         method                  =       "furthest";
 376         calc                    =       "";
 377         abund                   =   "10";
 378         step                    =       "0.01";
 379         form                    =       "integral";
 380         ends                    =   "T";
 381         processors              =       "1";
 382 }
 383 /*******************************************************/
 384
 385 /******************************************************/
 386 GlobalData::~GlobalData() {
 387         _uniqueInstance = 0;
 388         if(gListVector != NULL)         {       delete gListVector;             }
 389         if(gSparseMatrix != NULL)       {       delete gSparseMatrix;   }
 390         if(gorder != NULL)                      {       delete gorder;          }
 391 }
 392 /*******************************************************/
 393
 394 /*******************************************************/
 395 void GlobalData::parseTreeFile() {
 396         //only takes names from the first tree and assumes that all trees use the same names.
 397         try {
 398                 string filename = treefile;
 399                 ifstream filehandle;
 400                 openInputFile(filename, filehandle);
 401                 int c, comment;
 402                 comment = 0;
 403
 404                 //if you are not a nexus file
 405                 if ((c = filehandle.peek()) != '#') {
 406                         while((c = filehandle.peek()) != ';') {
 407                                 while ((c = filehandle.peek()) != ';') {
 408                                         // get past comments
 409                                         if(c == '[') {
 410                                                 comment = 1;
 411                                         }
 412                                         if(c == ']'){
 413                                                 comment = 0;
 414                                         }
 415                                         if((c == '(') && (comment != 1)){ break; }
 416                                         filehandle.get();
 417                                 }
 418
 419                                 readTreeString(filehandle);
 420                         }
 421                 //if you are a nexus file
 422                 }else if ((c = filehandle.peek()) == '#') {
 423                         string holder = "";
 424
 425                         // get past comments
 426                         while(holder != "translate" && holder != "Translate"){
 427                                 if(holder == "[" || holder == "[!"){
 428                                         comment = 1;
 429                                 }
 430                                 if(holder == "]"){
 431                                         comment = 0;
 432                                 }
 433                                 filehandle >> holder;
 434
 435                                 //if there is no translate then you must read tree string otherwise use translate to get names
 436                                 if(holder == "tree" && comment != 1){
 437                                         //pass over the "tree rep.6878900 = "
 438                                         while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF) ) {;}
 439
 440                                         if (c == EOF ) { break; }
 441                                         filehandle.putback(c);  //put back first ( of tree.
 442                                         readTreeString(filehandle);
 443                                         break;
 444                                 }
 445                         }
 446
 447                         //use nexus translation rather than parsing tree to save time
 448                         if ((holder == "translate") || (holder == "Translate")) {
 449
 450                                 string number, name, h;
 451                                 h = ""; // so it enters the loop the first time
 452                                 while((h != ";") && (number != ";")) {
 453                                         filehandle >> number;
 454                                         filehandle >> name;
 455
 456                                         //c = , until done with translation then c = ;
 457                                         h = name.substr(name.length()-1, name.length());
 458                                         name.erase(name.end()-1);  //erase the comma
 459                                         Treenames.push_back(number);
 460                                 }
 461                                 if (number == ";") { Treenames.pop_back(); }  //in case ';' from translation is on next line instead of next to last name
 462                         }
 463                 }
 464
 465         }
 466         catch(exception& e) {
 467                 cout << "Standard Error: " << e.what() << " has occurred in the GlobalData class Function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 468                 exit(1);
 469         }
 470         catch(...) {
 471                 cout << "An unknown error has occurred in the GlobalData class function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 472                 exit(1);
 473         }
 474 }
 475 /*******************************************************/
 476
 477 /*******************************************************/
 478 void GlobalData::readTreeString(ifstream& filehandle)   {
 479         try {
 480                 int c;
 481                 string name; //k
 482
 483                 while((c = filehandle.peek()) != ';') {
 484                                 //if you are a name
 485                         if ((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != '\t') && (c != 32)) { //32 is space
 486                                 name = "";
 487                                 c = filehandle.get();
 488         //              k = c;
 489 //cout << k << endl;
 490                                 while ((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != 32) && (c != '\t')) {
 491                                         name += c;
 492                                         c = filehandle.get();
 493                 //      k = c;
 494 //cout << " in name while " << k << endl;
 495                                 }
 496
 497 //cout << "name = " << name << endl;
 498                                 Treenames.push_back(name);
 499                                 filehandle.putback(c);
 500 //k = c;
 501 //cout << " after putback" <<  k << endl;
 502                         }
 503
 504                         if (c  == ':') { //read until you reach the end of the branch length
 505                                 while ((c != '(') && (c != ')') && (c != ',') && (c != ';') && (c != '\n') && (c != '\t') && (c != 32)) {
 506                                         c = filehandle.get();
 507                                 //      k = c;
 508         //cout << " in branch while " << k << endl;
 509                                 }
 510                                 filehandle.putback(c);
 511                         }
 512                         c = filehandle.get();
 513                         if (c == ';') { break; }
 514                 //      k = c;
 515 //cout << k << endl;
 516
 517                 }
 518         }
 519         catch(exception& e) {
 520                 cout << "Standard Error: " << e.what() << " has occurred in the GlobalData class Function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 521                 exit(1);
 522         }
 523         catch(...) {
 524                 cout << "An unknown error has occurred in the GlobalData class function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 525                 exit(1);
 526         }
 527 }
 528
 529 /*******************************************************/
 530
 531 /*******************************************************/
 532
 533