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                 m->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                         m->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                 m->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                 m->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                                 m->gobble(filehandle);
 129                         }
 130                 //if you are a nexus file
 131                 }else if ((c = filehandle.peek()) == '#') {
 132                         //get right number of seqs from nexus file.
 133                         Tree* temp = new Tree();  delete temp;
 134
 135                         nexusTranslation();  //reads file through the translation and updates treemap
 136                         while((c = filehandle.peek()) != EOF) {
 137                                 // get past comments
 138                                 while ((c = filehandle.peek()) != EOF) {
 139                                         if(holder == "[" || holder == "[!"){
 140                                                 comment = 1;
 141                                         }
 142                                         if(holder == "]"){
 143                                                 comment = 0;
 144                                         }
 145                                         if((holder == "tree" || holder == "end;") && comment != 1){ holder = ""; comment = 0; break;}
 146                                         filehandle >> holder;
 147                                 }
 148
 149                                 //pass over the "tree rep.6878900 = "
 150                                 while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF) ) {;}
 151
 152                                 if (c == EOF ) { break; }
 153                                 filehandle.putback(c);  //put back first ( of tree.
 154
 155                                 //make new tree
 156                                 T = new Tree();
 157                                 numNodes = T->getNumNodes();
 158                                 numLeaves = T->getNumLeaves();
 159
 160                                 //read tree info
 161                                 error = readTreeString();
 162
 163                                 //save trees for later commands
 164                                 globaldata->gTree.push_back(T);
 165                         }
 166                 }
 167
 168                 if (error != 0) { readOk = error; }
 169
 170                 filehandle.close();
 171
 172                 return readOk;
 173         }
 174         catch(exception& e) {
 175                 m->errorOut(e, "ReadNewickTree", "read");
 176                 exit(1);
 177         }
 178 }
 179 /**************************************************************************************************/
 180 //This function read the file through the translation of the sequences names and updates treemap.
 181 string ReadNewickTree::nexusTranslation() {
 182         try {
 183
 184                 holder = "";
 185                 int numSeqs = globaldata->Treenames.size(); //must save this some when we clear old names we can still know how many sequences there were
 186                 int comment = 0;
 187
 188                 // get past comments
 189                 while(holder != "translate" && holder != "Translate"){
 190                         if(holder == "[" || holder == "[!"){
 191                                 comment = 1;
 192                         }
 193                         if(holder == "]"){
 194                                 comment = 0;
 195                         }
 196                         filehandle >> holder;
 197                         if(holder == "tree" && comment != 1){return holder;}
 198                 }
 199
 200                 //update treemap
 201                 globaldata->gTreemap->namesOfSeqs.clear();
 202
 203                 /*char c;
 204                 string number, name;
 205                 while ((c = filehandle.peek()) != EOF) {
 206
 207                         filehandle >> number;
 208
 209                         if ((number == "tree") || (number == ";") ) {  name = number; break;  }
 210
 211                         filehandle >> name;
 212
 213                         char lastChar;
 214                         if (name.length() != 0) { lastChar = name[name.length()-1]; }
 215
 216                         if ((name == "tree") || (name == ";") ) {  break;  }
 217
 218                         if (lastChar == ',') {  name.erase(name.end()-1); } //erase the comma
 219                         */
 220                 cout << "numseqs = " << numSeqs << endl;
 221                 string number, name;
 222                 for(int i=0;i<numSeqs;i++){
 223
 224                         filehandle >> number;
 225                         filehandle >> name;
 226                         name.erase(name.end()-1);  //erase the comma
 227
 228                         //insert new one with new name
 229                         string group = globaldata->gTreemap->getGroup(name);
 230                         globaldata->gTreemap->treemap[toString(number)].groupname = group;
 231                         globaldata->gTreemap->treemap[toString(number)].vectorIndex = globaldata->gTreemap->getIndex(name);
 232                         //erase old one.  so treemap[sarah].groupnumber is now treemap[1].groupnumber. if number is 1 and name is sarah.
 233                         globaldata->gTreemap->treemap.erase(name);
 234                         globaldata->gTreemap->namesOfSeqs.push_back(number);
 235                 }
 236
 237                 return name;
 238         }
 239         catch(exception& e) {
 240                 m->errorOut(e, "ReadNewickTree", "nexusTranslation");
 241                 exit(1);
 242         }
 243 }
 244
 245 /**************************************************************************************************/
 246 int ReadNewickTree::readTreeString() {
 247         try {
 248
 249                 int n = 0;
 250                 int lc, rc;
 251
 252                 int rooted = 0;
 253
 254                 int ch = filehandle.peek();
 255
 256                 if(ch == '('){
 257                         n = numLeaves;  //number of leaves / sequences, we want node 1 to start where the leaves left off
 258
 259                         lc = readNewickInt(filehandle, n, T);
 260                         if (lc == -1) { m->mothurOut("error with lc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
 261
 262                         if(filehandle.peek()==','){
 263                                 readSpecialChar(filehandle,',',"comma");
 264                         }
 265                         // ';' means end of tree.
 266                         else if((ch=filehandle.peek())==';' || ch=='['){
 267                                 rooted = 1;
 268                         }
 269
 270                         if(rooted != 1){
 271                                 rc = readNewickInt(filehandle, n, T);
 272                                 if (rc == -1) { m->mothurOut("error with rc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
 273                                 if(filehandle.peek() == ')'){
 274                                         readSpecialChar(filehandle,')',"right parenthesis");
 275                                 }
 276                         }
 277                 }
 278                 //note: treeclimber had the code below added - not sure why?
 279                 else{
 280                         filehandle.putback(ch);
 281                         char name[MAX_LINE];
 282                         filehandle.get(name, MAX_LINE,'\n');
 283                         SKIPLINE(filehandle, ch);
 284
 285                         n = T->getIndex(name);
 286
 287                         if(n!=0){
 288                                 m->mothurOut("Internal error: The only taxon is not taxon 0.\n");
 289                                 //exit(1);
 290                                 readOk = -1; return -1;
 291                         }
 292                         lc = rc = -1;
 293                 }
 294
 295                 while(((ch=filehandle.get())!=';') && (filehandle.eof() != true)){;}
 296
 297                 if(rooted != 1){
 298                         T->tree[n].setChildren(lc,rc);
 299                         T->tree[n].setBranchLength(0);
 300                         T->tree[n].setParent(-1);
 301                         if(lc!=-1){             T->tree[lc].setParent(n);               }
 302                         if(rc!=-1){             T->tree[rc].setParent(n);               }
 303                 }
 304
 305                 //T->printTree(); cout << endl;
 306                 return 0;
 307
 308         }
 309         catch(exception& e) {
 310                 m->errorOut(e, "ReadNewickTree", "readTreeString");
 311                 exit(1);
 312         }
 313 }
 314 /**************************************************************************************************/
 315
 316 int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
 317         try {
 318
 319                 if (m->control_pressed) { return -1; }
 320
 321                 int c = readNodeChar(f);
 322
 323                 if(c == '('){
 324
 325                         //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating
 326                         //read all children
 327                         vector<int> childrenNodes;
 328                         while(f.peek() != ')'){
 329                                 int child = readNewickInt(f, n, T);
 330                                 if (child == -1) { return -1; } //reports an error in reading
 331                 //cout << "child = " << child << endl;
 332                                 childrenNodes.push_back(child);
 333
 334                                 //after a child you either have , or ), check for both
 335                                 if(f.peek()==')'){  break;  }
 336                                 else if (f.peek()==',') {   readSpecialChar(f,',',"comma");  }
 337                                 else {;}
 338                         }
 339         //cout << childrenNodes.size() << endl;
 340                         if (childrenNodes.size() < 2) {  m->mothurOut("Error in tree, please correct."); m->mothurOutEndLine(); return -1; }
 341
 342                         //then force into 2 node structure
 343                         for (int i = 1; i < childrenNodes.size(); i++) {
 344
 345                                 int lc, rc;
 346                                 if (i == 1) { lc = childrenNodes[i-1]; rc = childrenNodes[i]; }
 347                                 else { lc = n-1; rc = childrenNodes[i]; }
 348                         //cout << i << '\t' << lc << '\t' << rc << endl;
 349                                 T->tree[n].setChildren(lc,rc);
 350                                 T->tree[lc].setParent(n);
 351                                 T->tree[rc].setParent(n);
 352
 353                                 //T->printTree(); cout << endl;
 354                                 n++;
 355                         }
 356
 357                         //to account for extra ++ in looping
 358                         n--;
 359
 360                         if(f.peek()==')'){
 361                                 readSpecialChar(f,')',"right parenthesis");
 362                                 //to pass over labels in trees
 363                                 c=filehandle.get();
 364                                 while((c!=',') && (c != -1) && (c!= ':') && (c!=';')&& (c!=')')){ c=filehandle.get(); }
 365                                 filehandle.putback(c);
 366                         }
 367
 368                         if(f.peek() == ':'){
 369                                 readSpecialChar(f,':',"colon");
 370
 371                                 if(n >= numNodes){ m->mothurOut("Error: Too many nodes in input tree\n");  readOk = -1; return -1; }
 372
 373                                 T->tree[n].setBranchLength(readBranchLength(f));
 374                         }else{
 375                                 T->tree[n].setBranchLength(0.0);
 376                         }
 377
 378                         //T->tree[n].setChildren(lc,rc);
 379                         //T->tree[lc].setParent(n);
 380                         //T->tree[rc].setParent(n);
 381                         //T->printTree();  cout << endl;
 382
 383                         return n++;
 384
 385                 }else{
 386                         f.putback(c);
 387                         string name = "";
 388                         char d=f.get();
 389                         while(d != ':' && d != ',' && d!=')' && d!='\n'){
 390                                 name += d;
 391                                 d=f.get();
 392                         }
 393 //cout << name << endl;
 394                         int blen = 0;
 395                         if(d == ':')    {               blen = 1;       }
 396
 397                         f.putback(d);
 398
 399                         //set group info
 400                         string group = globaldata->gTreemap->getGroup(name);
 401
 402                         //find index in tree of name
 403                         int n1 = T->getIndex(name);
 404
 405                         //adds sequence names that are not in group file to the "xxx" group
 406                         if(group == "not found") {
 407                                 m->mothurOut("Name: " + name + " is not in your groupfile, and will be disregarded. \n");  //readOk = -1; return n1;
 408
 409                                 globaldata->gTreemap->namesOfSeqs.push_back(name);
 410                                 globaldata->gTreemap->treemap[name].groupname = "xxx";
 411
 412                                 map<string, int>::iterator it;
 413                                 it = globaldata->gTreemap->seqsPerGroup.find("xxx");
 414                                 if (it == globaldata->gTreemap->seqsPerGroup.end()) { //its a new group
 415                                         globaldata->gTreemap->namesOfGroups.push_back("xxx");
 416                                         globaldata->gTreemap->seqsPerGroup["xxx"] = 1;
 417                                 }else {
 418                                         globaldata->gTreemap->seqsPerGroup["xxx"]++;
 419                                 }
 420
 421                                 group = "xxx";
 422                         }
 423
 424                         vector<string> tempGroup; tempGroup.push_back(group);
 425
 426                         T->tree[n1].setGroup(tempGroup);
 427                         T->tree[n1].setChildren(-1,-1);
 428
 429                         if(blen == 1){
 430                                 f.get();
 431                                 T->tree[n1].setBranchLength(readBranchLength(f));
 432                         }else{
 433                                 T->tree[n1].setBranchLength(0.0);
 434                         }
 435
 436                         while((c=f.get())!=0 && (c != ':' && c != ',' && c!=')') )              {;}
 437
 438                         f.putback(c);
 439
 440                         return n1;
 441                 }
 442         }
 443         catch(exception& e) {
 444                 m->errorOut(e, "ReadNewickTree", "readNewickInt");
 445                 exit(1);
 446         }
 447 }
 448 /**************************************************************************************************/
 449 /**************************************************************************************************/
 450