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                 if (m->control_pressed) {  filehandle.close(); return 0; }
 120                                 while ((c = filehandle.peek()) != EOF) {
 121                     if (m->control_pressed) {  filehandle.close(); return 0; }
 122                                         // get past comments
 123                                         if(c == '[') {
 124                                                 comment = 1;
 125                                         }
 126                                         if(c == ']'){
 127                                                 comment = 0;
 128                                         }
 129                                         if((c == '(') && (comment != 1)){ break; }
 130                                         filehandle.get();
 131                                 }
 132
 133                                 //make new tree
 134                                 T = new Tree(ct);
 135
 136                                 numNodes = T->getNumNodes();
 137                                 numLeaves = T->getNumLeaves();
 138
 139                                 error = readTreeString(ct);
 140
 141                                 //save trees for later commands
 142                                 Trees.push_back(T);
 143                                 m->gobble(filehandle);
 144                         }
 145                 //if you are a nexus file
 146                 }else if ((c = filehandle.peek()) == '#') {
 147                         //get right number of seqs from nexus file.
 148                         Tree* temp = new Tree(ct);  delete temp;
 149
 150                         nexusTranslation(ct);  //reads file through the translation and updates treemap
 151                         while((c = filehandle.peek()) != EOF) {
 152                 if (m->control_pressed) {  filehandle.close(); return 0; }
 153                                 // get past comments
 154                                 while ((c = filehandle.peek()) != EOF) {
 155                     if (m->control_pressed) {  filehandle.close(); return 0; }
 156                                         if(holder == "[" || holder == "[!"){
 157                                                 comment = 1;
 158                                         }
 159                                         if(holder == "]"){
 160                                                 comment = 0;
 161                                         }
 162                                         if((holder == "tree" || holder == "end;") && comment != 1){ holder = ""; comment = 0; break;}
 163                                         filehandle >> holder;
 164                                 }
 165
 166                                 //pass over the "tree rep.6878900 = "
 167                                 while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF) ) {;}
 168
 169                                 if (c == EOF ) { break; }
 170                                 filehandle.putback(c);  //put back first ( of tree.
 171
 172                                 //make new tree
 173                                 T = new Tree(ct);
 174                                 numNodes = T->getNumNodes();
 175                                 numLeaves = T->getNumLeaves();
 176
 177                                 //read tree info
 178                                 error = readTreeString(ct);
 179
 180                                 //save trees for later commands
 181                                 Trees.push_back(T);
 182                         }
 183                 }
 184
 185                 if (error != 0) { readOk = error; }
 186
 187                 filehandle.close();
 188
 189                 return readOk;
 190         }
 191         catch(exception& e) {
 192                 m->errorOut(e, "ReadNewickTree", "read");
 193                 exit(1);
 194         }
 195 }
 196 /**************************************************************************************************/
 197 //This function read the file through the translation of the sequences names and updates treemap.
 198 string ReadNewickTree::nexusTranslation(CountTable* ct) {
 199         try {
 200
 201                 holder = "";
 202                 int numSeqs = m->Treenames.size(); //must save this some when we clear old names we can still know how many sequences there were
 203                 int comment = 0;
 204
 205                 // get past comments
 206                 while(holder != "translate" && holder != "Translate"){
 207                         if(holder == "[" || holder == "[!"){
 208                                 comment = 1;
 209                         }
 210                         if(holder == "]"){
 211                                 comment = 0;
 212                         }
 213                         filehandle >> holder;
 214                         if(holder == "tree" && comment != 1){return holder;}
 215                 }
 216
 217                 string number, name;
 218                 for(int i=0;i<numSeqs;i++){
 219
 220                         filehandle >> number;
 221                         filehandle >> name;
 222                         name.erase(name.end()-1);  //erase the comma
 223                         ct->renameSeq(name, toString(number));
 224                 }
 225
 226                 return name;
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "ReadNewickTree", "nexusTranslation");
 230                 exit(1);
 231         }
 232 }
 233
 234 /**************************************************************************************************/
 235 int ReadNewickTree::readTreeString(CountTable* ct) {
 236         try {
 237
 238                 int n = 0;
 239                 int lc, rc;
 240
 241                 int rooted = 0;
 242
 243                 int ch = filehandle.peek();
 244
 245                 if(ch == '('){
 246                         n = numLeaves;  //number of leaves / sequences, we want node 1 to start where the leaves left off
 247
 248                         lc = readNewickInt(filehandle, n, T, ct);
 249                         if (lc == -1) { m->mothurOut("error with lc"); m->mothurOutEndLine(); m->control_pressed = true; return -1; } //reports an error in reading
 250
 251                         if(filehandle.peek()==','){
 252                                 readSpecialChar(filehandle,',',"comma");
 253                         }
 254                         // ';' means end of tree.
 255                         else if((ch=filehandle.peek())==';' || ch=='['){
 256                                 rooted = 1;
 257                         }
 258
 259                         if(rooted != 1){
 260                                 rc = readNewickInt(filehandle, n, T, ct);
 261                                 if (rc == -1) { m->mothurOut("error with rc"); m->mothurOutEndLine(); m->control_pressed = true; return -1; } //reports an error in reading
 262                                 if(filehandle.peek() == ')'){
 263                                         readSpecialChar(filehandle,')',"right parenthesis");
 264                                 }
 265                         }
 266                 }
 267                 //note: treeclimber had the code below added - not sure why?
 268                 else{
 269                         filehandle.putback(ch);
 270                         char name[MAX_LINE];
 271                         filehandle.get(name, MAX_LINE,'\n');
 272                         SKIPLINE(filehandle, ch);
 273
 274                         n = T->getIndex(name);
 275
 276                         if(n!=0){
 277                                 m->mothurOut("Internal error: The only taxon is not taxon 0.\n");
 278                                 //exit(1);
 279                                 readOk = -1; return -1;
 280                         }
 281                         lc = rc = -1;
 282                 }
 283
 284                 while(((ch=filehandle.get())!=';') && (filehandle.eof() != true)){;}
 285
 286                 if(rooted != 1){
 287                         T->tree[n].setChildren(lc,rc);
 288                         T->tree[n].setBranchLength(0);
 289                         T->tree[n].setParent(-1);
 290                         if(lc!=-1){             T->tree[lc].setParent(n);               }
 291                         if(rc!=-1){             T->tree[rc].setParent(n);               }
 292                 }
 293
 294                 //T->printTree(); cout << endl;
 295                 return 0;
 296
 297         }
 298         catch(exception& e) {
 299                 m->errorOut(e, "ReadNewickTree", "readTreeString");
 300                 exit(1);
 301         }
 302 }
 303 /**************************************************************************************************/
 304
 305 int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T, CountTable* ct) {
 306         try {
 307
 308                 if (m->control_pressed) { return -1; }
 309
 310                 int c = readNodeChar(f);
 311
 312                 if(c == '('){
 313
 314                         //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating
 315                         //read all children
 316                         vector<int> childrenNodes;
 317                         while(f.peek() != ')'){
 318                                 int child = readNewickInt(f, n, T, ct);
 319                                 if (child == -1) { return -1; } //reports an error in reading
 320                 //cout << "child = " << child << endl;
 321                                 childrenNodes.push_back(child);
 322
 323                                 //after a child you either have , or ), check for both
 324                                 if(f.peek()==')'){  break;  }
 325                                 else if (f.peek()==',') {   readSpecialChar(f,',',"comma");  }
 326                                 else {;}
 327                         }
 328         //cout << childrenNodes.size() << endl;
 329                         if (childrenNodes.size() < 2) {  m->mothurOut("Error in tree, please correct."); m->mothurOutEndLine(); return -1; }
 330
 331                         //then force into 2 node structure
 332                         for (int i = 1; i < childrenNodes.size(); i++) {
 333
 334                                 int lc, rc;
 335                                 if (i == 1) { lc = childrenNodes[i-1]; rc = childrenNodes[i]; }
 336                                 else { lc = n-1; rc = childrenNodes[i]; }
 337                         //cout << i << '\t' << lc << '\t' << rc << endl;
 338                                 T->tree[n].setChildren(lc,rc);
 339                                 T->tree[lc].setParent(n);
 340                                 T->tree[rc].setParent(n);
 341
 342                                 //T->printTree(); cout << endl;
 343                                 n++;
 344                         }
 345
 346                         //to account for extra ++ in looping
 347                         n--;
 348
 349                         if(f.peek()==')'){
 350                                 readSpecialChar(f,')',"right parenthesis");
 351                                 //to pass over labels in trees
 352                                 c=filehandle.get();
 353                                 while((c!=',') && (c != -1) && (c!= ':') && (c!=';')&& (c!=')')){ c=filehandle.get(); }
 354                                 filehandle.putback(c);
 355                         }
 356
 357                         if(f.peek() == ':'){
 358                                 readSpecialChar(f,':',"colon");
 359
 360                                 if(n >= numNodes){ m->mothurOut("Error: Too many nodes in input tree\n");  readOk = -1; return -1; }
 361
 362                                 T->tree[n].setBranchLength(readBranchLength(f));
 363                         }else{
 364                                 T->tree[n].setBranchLength(0.0);
 365                         }
 366
 367                         return n++;
 368
 369                 }else{
 370                         f.putback(c);
 371                         string name = "";
 372                         char d=f.get();
 373                         while(d != ':' && d != ',' && d!=')' && d!='\n'){
 374                                 name += d;
 375                                 d=f.get();
 376                         }
 377 //cout << name << endl;
 378                         int blen = 0;
 379                         if(d == ':')    {               blen = 1;       }
 380
 381                         f.putback(d);
 382
 383                         //set group info
 384                         vector<string> group = ct->getGroups(name);
 385             //cout << name << endl;
 386                         //find index in tree of name
 387                         int n1 = T->getIndex(name);
 388
 389                         //adds sequence names that are not in group file to the "xxx" group
 390                         if(group.size() == 0) {
 391                                 m->mothurOut("Name: " + name + " is not in your groupfile, and will be disregarded. \n");  //readOk = -1; return n1;
 392
 393                 vector<string> currentGroups = ct->getNamesOfGroups();
 394                 if (!m->inUsersGroups("xxx", currentGroups)) {  ct->addGroup("xxx");  }
 395                 currentGroups = ct->getNamesOfGroups();
 396                 vector<int> thisCounts; thisCounts.resize(currentGroups.size(), 0);
 397                 for (int h = 0; h < currentGroups.size(); h++) {
 398                     if (currentGroups[h] == "xxx") {  thisCounts[h] = 1;  break; }
 399                 }
 400                 ct->push_back(name, thisCounts);
 401
 402                                 group.push_back("xxx");
 403                         }
 404                         T->tree[n1].setGroup(group);
 405                         T->tree[n1].setChildren(-1,-1);
 406
 407                         if(blen == 1){
 408                                 f.get();
 409                                 T->tree[n1].setBranchLength(readBranchLength(f));
 410                         }else{
 411                                 T->tree[n1].setBranchLength(0.0);
 412                         }
 413
 414                         while((c=f.get())!=0 && (c != ':' && c != ',' && c!=')') )              {;}
 415
 416                         f.putback(c);
 417
 418                         return n1;
 419                 }
 420         }
 421         catch(exception& e) {
 422                 m->errorOut(e, "ReadNewickTree", "readNewickInt");
 423                 exit(1);
 424         }
 425 }
 426 /**************************************************************************************************/
 427 /**************************************************************************************************/
 428