classifytreecommand.cpp

   1 //
   2 //  classifytreecommand.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 2/20/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "classifytreecommand.h"
  10 #include "phylotree.h"
  11 #include "treereader.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ClassifyTreeCommand::setParameters(){
  15         try {
  16                 CommandParameter ptree("tree", "InputTypes", "", "", "", "", "none",false,true); parameters.push_back(ptree);
  17         CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "", "", "none",false,true); parameters.push_back(ptaxonomy);
  18         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "none",false,false); parameters.push_back(pname);
  19         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "none", "none",false,false); parameters.push_back(pcount);
  20                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none",false,false); parameters.push_back(pgroup);
  21         CommandParameter pcutoff("cutoff", "Number", "", "51", "", "", "",false,true); parameters.push_back(pcutoff);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ClassifyTreeCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ClassifyTreeCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The classify.tree command reads a tree and taxonomy file and output the consensus taxonomy for each node on the tree. \n";
  39                 helpString += "If you provide a group file, the concensus for each group will also be provided. \n";
  40                 helpString += "The new tree contains labels at each internal node.  The label is the node number so you can relate the tree to the summary file.\n";
  41         helpString += "The count parameter allows you add a count file so you can have the summary totals broken up by group.\n";
  42                 helpString += "The summary file lists the concensus taxonomy for the descendants of each node.\n";
  43                 helpString += "The classify.tree command parameters are tree, group, name, count and taxonomy. The tree and taxonomy files are required.\n";
  44         helpString += "The cutoff parameter allows you to specify a consensus confidence threshold for your taxonomy.  The default is 51, meaning 51%. Cutoff cannot be below 51.\n";
  45         helpString += "The classify.tree command should be used in the following format: classify.tree(tree=test.tre, group=test.group, taxonomy=test.taxonomy)\n";
  46                 helpString += "Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n";
  47                 return helpString;
  48         }
  49         catch(exception& e) {
  50                 m->errorOut(e, "ClassifyTreeCommand", "getHelpString");
  51                 exit(1);
  52         }
  53 }
  54 //**********************************************************************************************************************
  55 string ClassifyTreeCommand::getOutputFileNameTag(string type, string inputName=""){
  56         try {
  57         string outputFileName = "";
  58                 map<string, vector<string> >::iterator it;
  59
  60         //is this a type this command creates
  61         it = outputTypes.find(type);
  62         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  63         else {
  64             if (type == "tree") {  outputFileName =  "taxonomy.tre"; }
  65             else if (type == "summary") {  outputFileName =  "taxonomy.summary"; }
  66             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  67         }
  68         return outputFileName;
  69         }
  70         catch(exception& e) {
  71                 m->errorOut(e, "ClassifyTreeCommand", "getOutputFileNameTag");
  72                 exit(1);
  73         }
  74 }
  75 //**********************************************************************************************************************
  76 ClassifyTreeCommand::ClassifyTreeCommand(){
  77         try {
  78                 abort = true; calledHelp = true;
  79                 setParameters();
  80                 vector<string> tempOutNames;
  81                 outputTypes["tree"] = tempOutNames;
  82                 outputTypes["summary"] = tempOutNames;
  83         }
  84         catch(exception& e) {
  85                 m->errorOut(e, "ClassifyTreeCommand", "ClassifyTreeCommand");
  86                 exit(1);
  87         }
  88 }
  89 //**********************************************************************************************************************
  90 ClassifyTreeCommand::ClassifyTreeCommand(string option)  {
  91         try {
  92                 abort = false; calledHelp = false;
  93
  94                 //allow user to run help
  95                 if(option == "help") { help(); abort = true; calledHelp = true; }
  96                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  97
  98                 else {
  99                         vector<string> myArray = setParameters();
 100
 101                         OptionParser parser(option);
 102                         map<string, string> parameters = parser.getParameters();
 103
 104                         ValidParameters validParameter;
 105                         map<string, string>::iterator it;
 106
 107                         //check to make sure all parameters are valid for command
 108                         for (it = parameters.begin(); it != parameters.end(); it++) {
 109                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 110                         }
 111
 112                         vector<string> tempOutNames;
 113                         outputTypes["tree"] = tempOutNames;
 114                         outputTypes["summary"] = tempOutNames;
 115
 116                         //if the user changes the input directory command factory will send this info to us in the output parameter
 117                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 118                         if (inputDir == "not found"){   inputDir = "";          }
 119                         else {
 120                                 string path;
 121                                 it = parameters.find("tree");
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["tree"] = inputDir + it->second;             }
 127                                 }
 128
 129                                 it = parameters.find("name");
 130                                 //user has given a template file
 131                                 if(it != parameters.end()){
 132                                         path = m->hasPath(it->second);
 133                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 135                                 }
 136
 137                                 it = parameters.find("group");
 138                                 //user has given a template file
 139                                 if(it != parameters.end()){
 140                                         path = m->hasPath(it->second);
 141                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 143                                 }
 144
 145                                 it = parameters.find("taxonomy");
 146                                 //user has given a template file
 147                                 if(it != parameters.end()){
 148                                         path = m->hasPath(it->second);
 149                                         //if the user has not given a path then, add inputdir. else leave path alone.
 150                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 151                                 }
 152
 153                 it = parameters.find("count");
 154                                 //user has given a template file
 155                                 if(it != parameters.end()){
 156                                         path = m->hasPath(it->second);
 157                                         //if the user has not given a path then, add inputdir. else leave path alone.
 158                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 159                                 }
 160                         }
 161
 162                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 163
 164                         //check for required parameters
 165                         treefile = validParameter.validFile(parameters, "tree", true);
 166                         if (treefile == "not open") { treefile = ""; abort = true; }
 167                         else if (treefile == "not found") { treefile = "";
 168                 treefile = m->getTreeFile();
 169                 if (treefile != "") {  m->mothurOut("Using " + treefile + " as input file for the tree parameter."); m->mothurOutEndLine(); }
 170                 else { m->mothurOut("No valid current files. You must provide a tree file."); m->mothurOutEndLine(); abort = true; }
 171             }else { m->setTreeFile(treefile); }
 172
 173             taxonomyfile = validParameter.validFile(parameters, "taxonomy", true);
 174                         if (taxonomyfile == "not open") { taxonomyfile = ""; abort = true; }
 175                         else if (taxonomyfile == "not found") { taxonomyfile = "";
 176                 taxonomyfile = m->getTaxonomyFile();
 177                 if (taxonomyfile != "") {  m->mothurOut("Using " + taxonomyfile + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
 178                 else { m->mothurOut("No valid current files. You must provide a taxonomy file."); m->mothurOutEndLine(); abort = true; }
 179             }else { m->setTaxonomyFile(taxonomyfile); }
 180
 181                         namefile = validParameter.validFile(parameters, "name", true);
 182                         if (namefile == "not open") { namefile = ""; abort = true; }
 183                         else if (namefile == "not found") { namefile = ""; }
 184                         else { m->setNameFile(namefile); }
 185
 186                         groupfile = validParameter.validFile(parameters, "group", true);
 187                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 188                         else if (groupfile == "not found") { groupfile = ""; }
 189                         else { m->setGroupFile(groupfile); }
 190
 191             countfile = validParameter.validFile(parameters, "count", true);
 192                         if (countfile == "not open") { countfile = ""; abort = true; }
 193                         else if (countfile == "not found") { countfile = "";  }
 194                         else { m->setCountTableFile(countfile); }
 195
 196             if ((namefile != "") && (countfile != "")) {
 197                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 198             }
 199
 200             if ((groupfile != "") && (countfile != "")) {
 201                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 202             }
 203
 204             string temp = validParameter.validFile(parameters, "cutoff", false);                        if (temp == "not found") { temp = "51"; }
 205                         m->mothurConvert(temp, cutoff);
 206
 207                         if ((cutoff < 51) || (cutoff > 100)) { m->mothurOut("cutoff must be above 50, and no greater than 100."); m->mothurOutEndLine(); abort = true;  }
 208
 209             if (countfile == "") {
 210                 if (namefile == "") {
 211                     vector<string> files; files.push_back(treefile);
 212                     parser.getNameFile(files);
 213                 }
 214                         }
 215                 }
 216         }
 217         catch(exception& e) {
 218                 m->errorOut(e, "ClassifyTreeCommand", "ClassifyTreeCommand");
 219                 exit(1);
 220         }
 221 }
 222 //**********************************************************************************************************************
 223
 224 int ClassifyTreeCommand::execute(){
 225         try {
 226
 227                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 228
 229                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 230
 231                 int start = time(NULL);
 232
 233                 /***************************************************/
 234                 //    reading tree info                                                    //
 235                 /***************************************************/
 236         m->setTreeFile(treefile);
 237
 238         TreeReader* reader = new TreeReader(treefile, groupfile, namefile);
 239         vector<Tree*> T = reader->getTrees();
 240         CountTable* tmap = T[0]->getCountTable();
 241         Tree* outputTree = T[0];
 242         delete reader;
 243
 244         if (namefile != "") { m->readNames(namefile, nameMap, nameCount); }
 245
 246         if (m->control_pressed) { delete tmap;  delete outputTree;  return 0; }
 247
 248         m->readTax(taxonomyfile, taxMap);
 249
 250         /***************************************************/
 251         //              get concensus taxonomies                    //
 252         /***************************************************/
 253         getClassifications(outputTree);
 254         delete outputTree; delete tmap;
 255
 256                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 257
 258                 //set tree file as new current treefile
 259                 if (treefile != "") {
 260                         string current = "";
 261                         itTypes = outputTypes.find("tree");
 262                         if (itTypes != outputTypes.end()) {
 263                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 264                         }
 265                 }
 266
 267                 m->mothurOutEndLine(); m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to find the concensus taxonomies."); m->mothurOutEndLine();
 268                 m->mothurOutEndLine();
 269                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 270                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 271                 m->mothurOutEndLine();
 272
 273                 return 0;
 274         }
 275         catch(exception& e) {
 276                 m->errorOut(e, "ClassifyTreeCommand", "execute");
 277                 exit(1);
 278         }
 279 }
 280 //**********************************************************************************************************************
 281 //traverse tree finding concensus taxonomy at each node
 282 //label node with a number to relate to output summary file
 283 //report all concensus taxonomies to file
 284 int ClassifyTreeCommand::getClassifications(Tree*& T){
 285         try {
 286
 287                 string thisOutputDir = outputDir;
 288                 if (outputDir == "") {  thisOutputDir += m->hasPath(treefile);  }
 289                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(treefile)) + getOutputFileNameTag("summary");
 290                 outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 291
 292                 ofstream out;
 293                 m->openOutputFile(outputFileName, out);
 294                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 295
 296                 //print headings
 297                 out << "TreeNode\t";
 298                 if (groupfile != "") { out << "Group\t"; }
 299         out << "NumRep\tTaxonomy" << endl;
 300
 301                 string treeOutputDir = outputDir;
 302                 if (outputDir == "") {  treeOutputDir += m->hasPath(treefile);  }
 303                 string outputTreeFileName = treeOutputDir + m->getRootName(m->getSimpleName(treefile)) + getOutputFileNameTag("tree");
 304
 305                 //create a map from tree node index to names of descendants, save time later
 306                 map<int, map<string, set<string> > > nodeToDescendants; //node# -> (groupName -> groupMembers)
 307                 for (int i = 0; i < T->getNumNodes(); i++) {
 308                         if (m->control_pressed) { return 0; }
 309
 310                         nodeToDescendants[i] = getDescendantList(T, i, nodeToDescendants);
 311                 }
 312
 313                 //for each node
 314                 for (int i = T->getNumLeaves(); i < T->getNumNodes(); i++) {
 315
 316                         if (m->control_pressed) { out.close(); return 0; }
 317
 318                         string tax = "not classifed";
 319             int size;
 320             if (groupfile != "") {
 321                 for (map<string, set<string> >::iterator itGroups = nodeToDescendants[i].begin(); itGroups != nodeToDescendants[i].end(); itGroups++) {
 322                     if (itGroups->first != "AllGroups") {
 323                         tax = getTaxonomy(itGroups->second, size);
 324                         out << (i+1) << '\t' << itGroups->first << '\t' << size << '\t' << tax << endl;
 325                     }
 326                 }
 327             }else {
 328                 string group = "AllGroups";
 329                 tax = getTaxonomy(nodeToDescendants[i][group], size);
 330                 out << (i+1) << '\t' << size << '\t' << tax << endl;
 331             }
 332
 333                         T->tree[i].setLabel((i+1));
 334                 }
 335                 out.close();
 336
 337                 ofstream outTree;
 338                 m->openOutputFile(outputTreeFileName, outTree);
 339                 outputNames.push_back(outputTreeFileName); outputTypes["tree"].push_back(outputTreeFileName);
 340                 T->print(outTree, "both");
 341                 outTree.close();
 342
 343                 return 0;
 344         }
 345         catch(exception& e) {
 346                 m->errorOut(e, "ClassifyTreeCommand", "GetConcensusTaxonomies");
 347                 exit(1);
 348         }
 349 }
 350 //**********************************************************************************************************************
 351 string ClassifyTreeCommand::getTaxonomy(set<string> names, int& size) {
 352         try{
 353                 string conTax = "";
 354         size = 0;
 355
 356                 //create a tree containing sequences from this bin
 357                 PhyloTree* phylo = new PhyloTree();
 358
 359                 for (set<string>::iterator it = names.begin(); it != names.end(); it++) {
 360
 361
 362                         //if namesfile include the names
 363                         if (namefile != "") {
 364
 365                                 //is this sequence in the name file - namemap maps seqName -> repSeqName
 366                                 map<string, string>::iterator it2 = nameMap.find(*it);
 367
 368                                 if (it2 == nameMap.end()) { //this name is not in name file, skip it
 369                                         m->mothurOut((*it) + " is not in your name file.  I will not include it in the consensus."); m->mothurOutEndLine();
 370                                 }else{
 371
 372                                         //is this sequence in the taxonomy file - look for repSeqName since we are assuming the taxonomy file is unique
 373                                         map<string, string>::iterator itTax = taxMap.find((it2->second));
 374
 375                                         if (itTax == taxMap.end()) { //this name is not in taxonomy file, skip it
 376
 377                                                 if ((*it) != (it2->second)) { m->mothurOut((*it) + " is represented by " +  it2->second + " and is not in your taxonomy file.  I will not include it in the consensus."); m->mothurOutEndLine(); }
 378                                                 else {  m->mothurOut((*it) + " is not in your taxonomy file.  I will not include it in the consensus."); m->mothurOutEndLine(); }
 379                                         }else{
 380                                                 //add seq to tree
 381                         int num = nameCount[(*it)]; // we know its there since we found it in nameMap
 382                                                 for (int i = 0; i < num; i++) {  phylo->addSeqToTree((*it)+toString(i), it2->second);  }
 383                         size += num;
 384                                         }
 385                                 }
 386
 387                         }else{
 388                                 //is this sequence in the taxonomy file - look for repSeqName since we are assuming the taxonomy file is unique
 389                                 map<string, string>::iterator itTax = taxMap.find((*it));
 390
 391                                 if (itTax == taxMap.end()) { //this name is not in taxonomy file, skip it
 392                                         m->mothurOut((*it) + " is not in your taxonomy file.  I will not include it in the consensus."); m->mothurOutEndLine();
 393                                 }else{
 394                                         if (countfile != "") {
 395                         int numDups = ct->getNumSeqs((*it));
 396                         for (int j = 0; j < numDups; j++) {  phylo->addSeqToTree((*it), itTax->second);  }
 397                         size += numDups;
 398                     }else{
 399                         //add seq to tree
 400                         phylo->addSeqToTree((*it), itTax->second);
 401                         size++;
 402                     }                           }
 403                         }
 404
 405                         if (m->control_pressed) { delete phylo; return conTax; }
 406
 407                 }
 408
 409                 //build tree
 410                 phylo->assignHeirarchyIDs(0);
 411
 412                 TaxNode currentNode = phylo->get(0);
 413                 int myLevel = 0;
 414                 //at each level
 415                 while (currentNode.children.size() != 0) { //you still have more to explore
 416
 417                         TaxNode bestChild;
 418                         int bestChildSize = 0;
 419
 420                         //go through children
 421                         for (map<string, int>::iterator itChild = currentNode.children.begin(); itChild != currentNode.children.end(); itChild++) {
 422
 423                                 TaxNode temp = phylo->get(itChild->second);
 424
 425                                 //select child with largest accesions - most seqs assigned to it
 426                                 if (temp.accessions.size() > bestChildSize) {
 427                                         bestChild = phylo->get(itChild->second);
 428                                         bestChildSize = temp.accessions.size();
 429                                 }
 430
 431                         }
 432
 433                         //is this taxonomy above cutoff
 434                         int consensusConfidence = ceil((bestChildSize / (float) size) * 100);
 435
 436                         if (consensusConfidence >= cutoff) { //if yes, add it
 437                 conTax += bestChild.name + "(" + toString(consensusConfidence) + ");";
 438                                 myLevel++;
 439                         }else{ //if no, quit
 440                                 break;
 441                         }
 442
 443                         //move down a level
 444                         currentNode = bestChild;
 445                 }
 446
 447                 if (myLevel != phylo->getMaxLevel()) {
 448                         while (myLevel != phylo->getMaxLevel()) {
 449                                 conTax += "unclassified;";
 450                                 myLevel++;
 451                         }
 452                 }
 453                 if (conTax == "") {  conTax = "no_consensus;";  }
 454
 455                 delete phylo;
 456
 457         return conTax;
 458
 459         }
 460         catch(exception& e) {
 461                 m->errorOut(e, "ClassifyTreeCommand", "getTaxonomy");
 462                 exit(1);
 463         }
 464 }
 465
 466 //**********************************************************************************************************************
 467 map<string, set<string> > ClassifyTreeCommand::getDescendantList(Tree*& T, int i, map<int, map<string, set<string> > > descendants){
 468         try {
 469                 map<string ,set<string> > names;
 470
 471                 map<string ,set<string> >::iterator it;
 472         map<string ,set<string> >::iterator it2;
 473
 474                 int lc = T->tree[i].getLChild();
 475                 int rc = T->tree[i].getRChild();
 476        // TreeMap* tmap = T->getTreeMap();
 477
 478                 if (lc == -1) { //you are a leaf your only descendant is yourself
 479             vector<string> groups = T->tree[i].getGroup();
 480             set<string> mynames; mynames.insert(T->tree[i].getName());
 481             for (int j = 0; j < groups.size(); j++) { names[groups[j]] = mynames;   } //mygroup -> me
 482             names["AllGroups"] = mynames;
 483                 }else{ //your descedants are the combination of your childrens descendants
 484                         names = descendants[lc];
 485                         for (it = descendants[rc].begin(); it != descendants[rc].end(); it++) {
 486                 it2 = names.find(it->first); //do we already have this group
 487                 if (it2 == names.end()) { //nope, so add it
 488                     names[it->first] = it->second;
 489                 }else {
 490                     for (set<string>::iterator it3 = (it->second).begin(); it3 != (it->second).end(); it3++) {
 491                         names[it->first].insert(*it3);
 492                     }
 493                 }
 494
 495                         }
 496                 }
 497
 498                 return names;
 499         }
 500         catch(exception& e) {
 501                 m->errorOut(e, "ClassifyTreeCommand", "getDescendantList");
 502                 exit(1);
 503         }
 504 }
 505 /*****************************************************************/
 506
 507