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