readtree.cpp

   1 /*
   2  *  readtree.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 "readtree.h"
  11
  12 /***********************************************************************/
  13 ReadTree::ReadTree() {
  14         try {
  15                 m = MothurOut::getInstance();
  16         }
  17         catch(exception& e) {
  18                 m->errorOut(e, "ReadTree", "ReadTree");
  19                 exit(1);
  20         }
  21 }
  22 /***********************************************************************/
  23 int ReadTree::AssembleTrees() {
  24          try {
  25                  //assemble users trees
  26                  for (int i = 0; i < Trees.size(); i++) {
  27                          if (m->control_pressed) { return 0;  }
  28                          Trees[i]->assembleTree();
  29                  }
  30                  return 0;
  31          }
  32         catch(exception& e) {
  33                 m->errorOut(e, "ReadTree", "AssembleTrees");
  34                 exit(1);
  35         }
  36 }
  37 /***********************************************************************/
  38 int ReadTree::readSpecialChar(istream& f, char c, string name) {
  39     try {
  40
  41                 m->gobble(f);
  42                 char d = f.get();
  43
  44                 if(d == EOF){
  45                         m->mothurOut("Error: Input file ends prematurely, expecting a " + name + "\n");
  46                         exit(1);
  47                 }
  48                 if(d != c){
  49                         m->mothurOut("Error: Expected " + name + " in input file.  Found " + toString(d) + ".\n");
  50                         exit(1);
  51                 }
  52                 if(d == ')' && f.peek() == '\n'){
  53                         m->gobble(f);
  54                 }
  55                 return d;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "ReadTree", "readSpecialChar");
  59                 exit(1);
  60         }
  61 }
  62 /**************************************************************************************************/
  63
  64 int ReadTree::readNodeChar(istream& f) {
  65         try {
  66 //              while(isspace(d=f.get()))               {;}
  67                 m->gobble(f);
  68                 char d = f.get();
  69
  70                 if(d == EOF){
  71                         m->mothurOut("Error: Input file ends prematurely, expecting a left parenthesis\n");
  72                         exit(1);
  73                 }
  74                 return d;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "ReadTree", "readNodeChar");
  78                 exit(1);
  79         }
  80 }
  81
  82 /**************************************************************************************************/
  83
  84 float ReadTree::readBranchLength(istream& f) {
  85     try {
  86                 float b;
  87
  88                 if(!(f >> b)){
  89                         m->mothurOut("Error: Missing branch length in input tree.\n");
  90                         exit(1);
  91                 }
  92                 m->gobble(f);
  93                 return b;
  94         }
  95         catch(exception& e) {
  96                 m->errorOut(e, "ReadTree", "readBranchLength");
  97                 exit(1);
  98         }
  99 }
 100
 101 /***********************************************************************/
 102 /***********************************************************************/
 103
 104 //Child Classes Below
 105
 106 /***********************************************************************/
 107 /***********************************************************************/
 108 //This class reads a file in Newick form and stores it in a tree.
 109
 110 int ReadNewickTree::read(CountTable* ct) {
 111         try {
 112                 holder = "";
 113                 int c, error;
 114                 int comment = 0;
 115
 116                 //if you are not a nexus file
 117                 if ((c = filehandle.peek()) != '#') {
 118                         while((c = filehandle.peek()) != EOF) {
 119                                 while ((c = filehandle.peek()) != EOF) {
 120                                         // get past comments
 121                                         if(c == '[') {
 122                                                 comment = 1;
 123                                         }
 124                                         if(c == ']'){
 125                                                 comment = 0;
 126                                         }
 127                                         if((c == '(') && (comment != 1)){ break; }
 128                                         filehandle.get();
 129                                 }
 130
 131                                 //make new tree
 132                                 T = new Tree(ct);
 133
 134                                 numNodes = T->getNumNodes();
 135                                 numLeaves = T->getNumLeaves();
 136
 137                                 error = readTreeString(ct);
 138
 139                                 //save trees for later commands
 140                                 Trees.push_back(T);
 141                                 m->gobble(filehandle);
 142                         }
 143                 //if you are a nexus file
 144                 }else if ((c = filehandle.peek()) == '#') {
 145                         //get right number of seqs from nexus file.
 146                         Tree* temp = new Tree(ct);  delete temp;
 147
 148                         nexusTranslation(ct);  //reads file through the translation and updates treemap
 149                         while((c = filehandle.peek()) != EOF) {
 150                                 // get past comments
 151                                 while ((c = filehandle.peek()) != EOF) {
 152                                         if(holder == "[" || holder == "[!"){
 153                                                 comment = 1;
 154                                         }
 155                                         if(holder == "]"){
 156                                                 comment = 0;
 157                                         }
 158                                         if((holder == "tree" || holder == "end;") && comment != 1){ holder = ""; comment = 0; break;}
 159                                         filehandle >> holder;
 160                                 }
 161
 162                                 //pass over the "tree rep.6878900 = "
 163                                 while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF) ) {;}
 164
 165                                 if (c == EOF ) { break; }
 166                                 filehandle.putback(c);  //put back first ( of tree.
 167
 168                                 //make new tree
 169                                 T = new Tree(ct);
 170                                 numNodes = T->getNumNodes();
 171                                 numLeaves = T->getNumLeaves();
 172
 173                                 //read tree info
 174                                 error = readTreeString(ct);
 175
 176                                 //save trees for later commands
 177                                 Trees.push_back(T);
 178                         }
 179                 }
 180
 181                 if (error != 0) { readOk = error; }
 182
 183                 filehandle.close();
 184
 185                 return readOk;
 186         }
 187         catch(exception& e) {
 188                 m->errorOut(e, "ReadNewickTree", "read");
 189                 exit(1);
 190         }
 191 }
 192 /**************************************************************************************************/
 193 //This function read the file through the translation of the sequences names and updates treemap.
 194 string ReadNewickTree::nexusTranslation(CountTable* ct) {
 195         try {
 196
 197                 holder = "";
 198                 int numSeqs = m->Treenames.size(); //must save this some when we clear old names we can still know how many sequences there were
 199                 int comment = 0;
 200
 201                 // get past comments
 202                 while(holder != "translate" && holder != "Translate"){
 203                         if(holder == "[" || holder == "[!"){
 204                                 comment = 1;
 205                         }
 206                         if(holder == "]"){
 207                                 comment = 0;
 208                         }
 209                         filehandle >> holder;
 210                         if(holder == "tree" && comment != 1){return holder;}
 211                 }
 212
 213                 string number, name;
 214                 for(int i=0;i<numSeqs;i++){
 215
 216                         filehandle >> number;
 217                         filehandle >> name;
 218                         name.erase(name.end()-1);  //erase the comma
 219                         ct->renameSeq(name, toString(number));
 220                 }
 221
 222                 return name;
 223         }
 224         catch(exception& e) {
 225                 m->errorOut(e, "ReadNewickTree", "nexusTranslation");
 226                 exit(1);
 227         }
 228 }
 229
 230 /**************************************************************************************************/
 231 int ReadNewickTree::readTreeString(CountTable* ct) {
 232         try {
 233
 234                 int n = 0;
 235                 int lc, rc;
 236
 237                 int rooted = 0;
 238
 239                 int ch = filehandle.peek();
 240
 241                 if(ch == '('){
 242                         n = numLeaves;  //number of leaves / sequences, we want node 1 to start where the leaves left off
 243
 244                         lc = readNewickInt(filehandle, n, T, ct);
 245                         if (lc == -1) { m->mothurOut("error with lc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
 246
 247                         if(filehandle.peek()==','){
 248                                 readSpecialChar(filehandle,',',"comma");
 249                         }
 250                         // ';' means end of tree.
 251                         else if((ch=filehandle.peek())==';' || ch=='['){
 252                                 rooted = 1;
 253                         }
 254
 255                         if(rooted != 1){
 256                                 rc = readNewickInt(filehandle, n, T, ct);
 257                                 if (rc == -1) { m->mothurOut("error with rc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
 258                                 if(filehandle.peek() == ')'){
 259                                         readSpecialChar(filehandle,')',"right parenthesis");
 260                                 }
 261                         }
 262                 }
 263                 //note: treeclimber had the code below added - not sure why?
 264                 else{
 265                         filehandle.putback(ch);
 266                         char name[MAX_LINE];
 267                         filehandle.get(name, MAX_LINE,'\n');
 268                         SKIPLINE(filehandle, ch);
 269
 270                         n = T->getIndex(name);
 271
 272                         if(n!=0){
 273                                 m->mothurOut("Internal error: The only taxon is not taxon 0.\n");
 274                                 //exit(1);
 275                                 readOk = -1; return -1;
 276                         }
 277                         lc = rc = -1;
 278                 }
 279
 280                 while(((ch=filehandle.get())!=';') && (filehandle.eof() != true)){;}
 281
 282                 if(rooted != 1){
 283                         T->tree[n].setChildren(lc,rc);
 284                         T->tree[n].setBranchLength(0);
 285                         T->tree[n].setParent(-1);
 286                         if(lc!=-1){             T->tree[lc].setParent(n);               }
 287                         if(rc!=-1){             T->tree[rc].setParent(n);               }
 288                 }
 289
 290                 //T->printTree(); cout << endl;
 291                 return 0;
 292
 293         }
 294         catch(exception& e) {
 295                 m->errorOut(e, "ReadNewickTree", "readTreeString");
 296                 exit(1);
 297         }
 298 }
 299 /**************************************************************************************************/
 300
 301 int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T, CountTable* ct) {
 302         try {
 303
 304                 if (m->control_pressed) { return -1; }
 305
 306                 int c = readNodeChar(f);
 307
 308                 if(c == '('){
 309
 310                         //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating
 311                         //read all children
 312                         vector<int> childrenNodes;
 313                         while(f.peek() != ')'){
 314                                 int child = readNewickInt(f, n, T, ct);
 315                                 if (child == -1) { return -1; } //reports an error in reading
 316                 //cout << "child = " << child << endl;
 317                                 childrenNodes.push_back(child);
 318
 319                                 //after a child you either have , or ), check for both
 320                                 if(f.peek()==')'){  break;  }
 321                                 else if (f.peek()==',') {   readSpecialChar(f,',',"comma");  }
 322                                 else {;}
 323                         }
 324         //cout << childrenNodes.size() << endl;
 325                         if (childrenNodes.size() < 2) {  m->mothurOut("Error in tree, please correct."); m->mothurOutEndLine(); return -1; }
 326
 327                         //then force into 2 node structure
 328                         for (int i = 1; i < childrenNodes.size(); i++) {
 329
 330                                 int lc, rc;
 331                                 if (i == 1) { lc = childrenNodes[i-1]; rc = childrenNodes[i]; }
 332                                 else { lc = n-1; rc = childrenNodes[i]; }
 333                         //cout << i << '\t' << lc << '\t' << rc << endl;
 334                                 T->tree[n].setChildren(lc,rc);
 335                                 T->tree[lc].setParent(n);
 336                                 T->tree[rc].setParent(n);
 337
 338                                 //T->printTree(); cout << endl;
 339                                 n++;
 340                         }
 341
 342                         //to account for extra ++ in looping
 343                         n--;
 344
 345                         if(f.peek()==')'){
 346                                 readSpecialChar(f,')',"right parenthesis");
 347                                 //to pass over labels in trees
 348                                 c=filehandle.get();
 349                                 while((c!=',') && (c != -1) && (c!= ':') && (c!=';')&& (c!=')')){ c=filehandle.get(); }
 350                                 filehandle.putback(c);
 351                         }
 352
 353                         if(f.peek() == ':'){
 354                                 readSpecialChar(f,':',"colon");
 355
 356                                 if(n >= numNodes){ m->mothurOut("Error: Too many nodes in input tree\n");  readOk = -1; return -1; }
 357
 358                                 T->tree[n].setBranchLength(readBranchLength(f));
 359                         }else{
 360                                 T->tree[n].setBranchLength(0.0);
 361                         }
 362
 363                         return n++;
 364
 365                 }else{
 366                         f.putback(c);
 367                         string name = "";
 368                         char d=f.get();
 369                         while(d != ':' && d != ',' && d!=')' && d!='\n'){
 370                                 name += d;
 371                                 d=f.get();
 372                         }
 373 //cout << name << endl;
 374                         int blen = 0;
 375                         if(d == ':')    {               blen = 1;       }
 376
 377                         f.putback(d);
 378
 379                         //set group info
 380                         vector<string> group = ct->getGroups(name);
 381             //cout << name << endl;
 382                         //find index in tree of name
 383                         int n1 = T->getIndex(name);
 384
 385                         //adds sequence names that are not in group file to the "xxx" group
 386                         if(group.size() == 0) {
 387                                 m->mothurOut("Name: " + name + " is not in your groupfile, and will be disregarded. \n");  //readOk = -1; return n1;
 388
 389                 vector<string> currentGroups = ct->getNamesOfGroups();
 390                 if (!m->inUsersGroups("xxx", currentGroups)) {  ct->addGroup("xxx");  }
 391                 currentGroups = ct->getNamesOfGroups();
 392                 vector<int> thisCounts; thisCounts.resize(currentGroups.size(), 0);
 393                 for (int h = 0; h < currentGroups.size(); h++) {
 394                     if (currentGroups[h] == "xxx") {  thisCounts[h] = 1;  break; }
 395                 }
 396                 ct->push_back(name, thisCounts);
 397
 398                                 group.push_back("xxx");
 399                         }
 400                         T->tree[n1].setGroup(group);
 401                         T->tree[n1].setChildren(-1,-1);
 402
 403                         if(blen == 1){
 404                                 f.get();
 405                                 T->tree[n1].setBranchLength(readBranchLength(f));
 406                         }else{
 407                                 T->tree[n1].setBranchLength(0.0);
 408                         }
 409
 410                         while((c=f.get())!=0 && (c != ':' && c != ',' && c!=')') )              {;}
 411
 412                         f.putback(c);
 413
 414                         return n1;
 415                 }
 416         }
 417         catch(exception& e) {
 418                 m->errorOut(e, "ReadNewickTree", "readNewickInt");
 419                 exit(1);
 420         }
 421 }
 422 /**************************************************************************************************/
 423 /**************************************************************************************************/
 424