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                 return readOk;
 169         }
 170         catch(exception& e) {
 171                 m->errorOut(e, "ReadNewickTree", "read");
 172                 exit(1);
 173         }
 174 }
 175 /**************************************************************************************************/
 176 //This function read the file through the translation of the sequences names and updates treemap.
 177 void ReadNewickTree::nexusTranslation() {
 178         try {
 179
 180                 holder = "";
 181                 int numSeqs = globaldata->gTreemap->getNumSeqs(); //must save this some when we clear old names we can still know how many sequences there were
 182                 int comment = 0;
 183
 184                 // get past comments
 185                 while(holder != "translate" && holder != "Translate"){
 186                         if(holder == "[" || holder == "[!"){
 187                                 comment = 1;
 188                         }
 189                         if(holder == "]"){
 190                                 comment = 0;
 191                         }
 192                         filehandle >> holder;
 193                         if(holder == "tree" && comment != 1){return;}
 194                 }
 195
 196                 //update treemap
 197                 globaldata->gTreemap->namesOfSeqs.clear();
 198                 for(int i=0;i<numSeqs;i++){
 199                         string number, name;
 200                         filehandle >> number;
 201                         filehandle >> name;
 202                         name.erase(name.end()-1);  //erase the comma
 203                         //insert new one with new name
 204                         globaldata->gTreemap->treemap[toString(number)].groupname = globaldata->gTreemap->treemap[name].groupname;
 205                         globaldata->gTreemap->treemap[toString(number)].vectorIndex = globaldata->gTreemap->treemap[name].vectorIndex;
 206                         //erase old one.  so treemap[sarah].groupnumber is now treemap[1].groupnumber. if number is 1 and name is sarah.
 207                         globaldata->gTreemap->treemap.erase(name);
 208                         globaldata->gTreemap->namesOfSeqs.push_back(number);
 209                 }
 210         }
 211         catch(exception& e) {
 212                 m->errorOut(e, "ReadNewickTree", "nexusTranslation");
 213                 exit(1);
 214         }
 215 }
 216
 217 /**************************************************************************************************/
 218 int ReadNewickTree::readTreeString() {
 219         try {
 220
 221                 int n = 0;
 222                 int lc, rc;
 223
 224                 int rooted = 0;
 225
 226                 int ch = filehandle.peek();
 227
 228                 if(ch == '('){
 229                         n = numLeaves;  //number of leaves / sequences, we want node 1 to start where the leaves left off
 230
 231                         lc = readNewickInt(filehandle, n, T);
 232                         if (lc == -1) { m->mothurOut("error with lc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
 233
 234                         if(filehandle.peek()==','){
 235                                 readSpecialChar(filehandle,',',"comma");
 236                         }
 237                         // ';' means end of tree.
 238                         else if((ch=filehandle.peek())==';' || ch=='['){
 239                                 rooted = 1;
 240                         }
 241                         if(rooted != 1){
 242                                 rc = readNewickInt(filehandle, n, T);
 243                                 if (rc == -1) { m->mothurOut("error with rc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
 244                                 if(filehandle.peek() == ')'){
 245                                         readSpecialChar(filehandle,')',"right parenthesis");
 246                                 }
 247                         }
 248                 }
 249                 //note: treeclimber had the code below added - not sure why?
 250                 else{
 251                         filehandle.putback(ch);
 252                         char name[MAX_LINE];
 253                         filehandle.get(name, MAX_LINE,'\n');
 254                         SKIPLINE(filehandle, ch);
 255
 256                         n = T->getIndex(name);
 257
 258                         if(n!=0){
 259                                 m->mothurOut("Internal error: The only taxon is not taxon 0.\n");
 260                                 //exit(1);
 261                                 readOk = -1; return -1;
 262                         }
 263                         lc = rc = -1;
 264                 }
 265
 266                 while(((ch=filehandle.get())!=';') && (filehandle.eof() != true)){;}
 267                 if(rooted != 1){
 268                         T->tree[n].setChildren(lc,rc);
 269                         T->tree[n].setBranchLength(0);
 270                         T->tree[n].setParent(-1);
 271                         if(lc!=-1){             T->tree[lc].setParent(n);               }
 272                         if(rc!=-1){             T->tree[rc].setParent(n);               }
 273                 }
 274                 return 0;
 275
 276         }
 277         catch(exception& e) {
 278                 m->errorOut(e, "ReadNewickTree", "readTreeString");
 279                 exit(1);
 280         }
 281 }
 282 /**************************************************************************************************/
 283
 284 int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
 285         try {
 286
 287                 if (m->control_pressed) { return -1; }
 288
 289                 int c = readNodeChar(f);
 290   string k;
 291 k = c;
 292         cout << "at beginning = " << k <<endl;
 293                 if(c == '('){
 294
 295                         //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating
 296                         //read all children
 297                         vector<int> childrenNodes;
 298                         while(f.peek() != ')'){
 299                                 int child = readNewickInt(f, n, T);
 300                                 if (child == -1) { return -1; } //reports an error in reading
 301
 302                                 childrenNodes.push_back(child);
 303
 304                                 //after a child you either have , or ), check for both
 305                                 if(f.peek()==')'){  break;  }
 306                                 else if (f.peek()==',') {   readSpecialChar(f,',',"comma");  }
 307                                 else { string k;
 308                         k = f.peek();
 309         cout << "in here k = " << k << '\t' << f.tellg() <<endl;
 310  }
 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; 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                         //int lc = readNewickInt(f, n, T);
 333                         //if (lc == -1) { return -1; } //reports an error in reading
 334
 335                         //readSpecialChar(f,',',"comma");
 336
 337                         //int rc = readNewickInt(f, n, T);
 338                         //if (rc == -1) { return -1; }  //reports an error in reading
 339
 340                         if(f.peek()==')'){
 341                                 readSpecialChar(f,')',"right parenthesis");
 342                                 //to pass over labels in trees
 343                                 c=filehandle.get();
 344                                 while((c!=',') && (c != -1) && (c!= ':') && (c!=';')){ c=filehandle.get(); }
 345                                 filehandle.putback(c);
 346                         }
 347
 348                         if(f.peek() == ':'){
 349                                 readSpecialChar(f,':',"colon");
 350
 351                                 if(n >= numNodes){      m->mothurOut("Error: Too many nodes in input tree\n");  readOk = -1; return -1; }
 352
 353                                 T->tree[n].setBranchLength(readBranchLength(f));
 354                         }else{
 355                                 T->tree[n].setBranchLength(0.0);
 356                         }
 357
 358                         //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating
 359                         /*while(f.peek() == ','){
 360                         string k;
 361                         k = f.peek();
 362         cout << "in here k = " << k << '\t' << f.tellg() <<endl;
 363                                 //force this node to be left child and read new rc
 364                                 T->tree[n].setChildren(lc,rc);
 365                                 T->tree[lc].setParent(n);
 366                                 T->tree[rc].setParent(n);
 367
 368                                 T->printTree(); cout << endl;
 369                                 lc = n;
 370                                 n++;
 371
 372                                 readSpecialChar(f,',',"comma");
 373
 374                                 rc = readNewickInt(f, n, T);
 375
 376                                 if (rc == -1) { return -1; }  //reports an error in reading
 377
 378                                 if(f.peek()==')'){
 379                                         readSpecialChar(f,')',"right parenthesis");
 380                                         //to pass over labels in trees
 381                                         c=filehandle.get();
 382                                         while((c!=',') && (c != -1) && (c!= ':') && (c!=';')){ c=filehandle.get(); }
 383                                         filehandle.putback(c);
 384
 385                                         if(f.peek() == ':'){
 386                                                 readSpecialChar(f,':',"colon");
 387
 388                                                 if(n >= numNodes){      m->mothurOut("Error: Too many nodes in input tree\n");  readOk = -1; return -1; }
 389
 390                                                 T->tree[n].setBranchLength(readBranchLength(f));
 391                                         }else{
 392                                                 T->tree[n].setBranchLength(0.0);
 393                                         }
 394
 395                                         break;
 396                                 }
 397                         }*/
 398
 399                         //T->tree[n].setChildren(lc,rc);
 400                         //T->tree[lc].setParent(n);
 401                         //T->tree[rc].setParent(n);
 402                         //T->printTree();  cout << endl;
 403
 404                         return n++;
 405
 406                 }else{
 407                         f.putback(c);
 408                         string name = "";
 409                         char d=f.get();
 410                         while(d != ':' && d != ',' && d!=')' && d!='\n'){
 411                                 name += d;
 412                                 d=f.get();
 413                         }
 414                 cout << name << endl;
 415                         int blen = 0;
 416                         if(d == ':')    {               blen = 1;       }
 417
 418                         f.putback(d);
 419
 420                         //set group info
 421                         string group = globaldata->gTreemap->getGroup(name);
 422
 423                         //find index in tree of name
 424                         int n1 = T->getIndex(name);
 425
 426                         //adds sequence names that are not in group file to the "xxx" group
 427                         if(group == "not found") {
 428                                 m->mothurOut("Name: " + name + " is not in your groupfile, and will be disregarded. \n");  //readOk = -1; return n1;
 429
 430                                 globaldata->gTreemap->namesOfSeqs.push_back(name);
 431                                 globaldata->gTreemap->treemap[name].groupname = "xxx";
 432
 433                                 map<string, int>::iterator it;
 434                                 it = globaldata->gTreemap->seqsPerGroup.find("xxx");
 435                                 if (it == globaldata->gTreemap->seqsPerGroup.end()) { //its a new group
 436                                         globaldata->gTreemap->namesOfGroups.push_back("xxx");
 437                                         globaldata->gTreemap->seqsPerGroup["xxx"] = 1;
 438                                 }else {
 439                                         globaldata->gTreemap->seqsPerGroup["xxx"]++;
 440                                 }
 441
 442                                 group = "xxx";
 443                         }
 444
 445                         vector<string> tempGroup; tempGroup.push_back(group);
 446
 447                         T->tree[n1].setGroup(tempGroup);
 448                         T->tree[n1].setChildren(-1,-1);
 449
 450                         if(blen == 1){
 451                                 f.get();
 452                                 T->tree[n1].setBranchLength(readBranchLength(f));
 453                         }else{
 454                                 T->tree[n1].setBranchLength(0.0);
 455                         }
 456
 457                         while((c=f.get())!=0 && (c != ':' && c != ',' && c!=')') )              {;}
 458                         f.putback(c);
 459
 460                         return n1;
 461                 }
 462         }
 463         catch(exception& e) {
 464                 m->errorOut(e, "ReadNewickTree", "readNewickInt");
 465                 exit(1);
 466         }
 467 }
 468 /**************************************************************************************************/
 469 /**************************************************************************************************/
 470