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","tree-summary",false,true,true); parameters.push_back(ptree);
  17         CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "", "", "none","",false,true,true); parameters.push_back(ptaxonomy);
  18         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "none","",false,false,true); parameters.push_back(pname);
  19         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "none", "none","",false,false,true); parameters.push_back(pcount);
  20                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none","",false,false,true); 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::getOutputPattern(string type) {
  56     try {
  57         string pattern = "";
  58
  59         if (type == "summary") {  pattern = "[filename],taxonomy.summary"; } //makes file like: amazon.0.03.fasta
  60         else if (type == "tree") {  pattern = "[filename],taxonomy.tre"; }
  61         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  62
  63         return pattern;
  64     }
  65     catch(exception& e) {
  66         m->errorOut(e, "ClassifyTreeCommand", "getOutputPattern");
  67         exit(1);
  68     }
  69 }
  70 //**********************************************************************************************************************
  71 ClassifyTreeCommand::ClassifyTreeCommand(){
  72         try {
  73                 abort = true; calledHelp = true;
  74                 setParameters();
  75                 vector<string> tempOutNames;
  76                 outputTypes["tree"] = tempOutNames;
  77                 outputTypes["summary"] = tempOutNames;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "ClassifyTreeCommand", "ClassifyTreeCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85 ClassifyTreeCommand::ClassifyTreeCommand(string option)  {
  86         try {
  87                 abort = false; calledHelp = false;
  88
  89                 //allow user to run help
  90                 if(option == "help") { help(); abort = true; calledHelp = true; }
  91                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  92
  93                 else {
  94                         vector<string> myArray = setParameters();
  95
  96                         OptionParser parser(option);
  97                         map<string, string> parameters = parser.getParameters();
  98
  99                         ValidParameters validParameter;
 100                         map<string, string>::iterator it;
 101
 102                         //check to make sure all parameters are valid for command
 103                         for (it = parameters.begin(); it != parameters.end(); it++) {
 104                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 105                         }
 106
 107                         vector<string> tempOutNames;
 108                         outputTypes["tree"] = tempOutNames;
 109                         outputTypes["summary"] = tempOutNames;
 110
 111                         //if the user changes the input directory command factory will send this info to us in the output parameter
 112                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 113                         if (inputDir == "not found"){   inputDir = "";          }
 114                         else {
 115                                 string path;
 116                                 it = parameters.find("tree");
 117                                 //user has given a template file
 118                                 if(it != parameters.end()){
 119                                         path = m->hasPath(it->second);
 120                                         //if the user has not given a path then, add inputdir. else leave path alone.
 121                                         if (path == "") {       parameters["tree"] = inputDir + it->second;             }
 122                                 }
 123
 124                                 it = parameters.find("name");
 125                                 //user has given a template file
 126                                 if(it != parameters.end()){
 127                                         path = m->hasPath(it->second);
 128                                         //if the user has not given a path then, add inputdir. else leave path alone.
 129                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 130                                 }
 131
 132                                 it = parameters.find("group");
 133                                 //user has given a template file
 134                                 if(it != parameters.end()){
 135                                         path = m->hasPath(it->second);
 136                                         //if the user has not given a path then, add inputdir. else leave path alone.
 137                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 138                                 }
 139
 140                                 it = parameters.find("taxonomy");
 141                                 //user has given a template file
 142                                 if(it != parameters.end()){
 143                                         path = m->hasPath(it->second);
 144                                         //if the user has not given a path then, add inputdir. else leave path alone.
 145                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 146                                 }
 147
 148                 it = parameters.find("count");
 149                                 //user has given a template file
 150                                 if(it != parameters.end()){
 151                                         path = m->hasPath(it->second);
 152                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 154                                 }
 155                         }
 156
 157                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 158
 159                         //check for required parameters
 160                         treefile = validParameter.validFile(parameters, "tree", true);
 161                         if (treefile == "not open") { treefile = ""; abort = true; }
 162                         else if (treefile == "not found") { treefile = "";
 163                 treefile = m->getTreeFile();
 164                 if (treefile != "") {  m->mothurOut("Using " + treefile + " as input file for the tree parameter."); m->mothurOutEndLine(); }
 165                 else { m->mothurOut("No valid current files. You must provide a tree file."); m->mothurOutEndLine(); abort = true; }
 166             }else { m->setTreeFile(treefile); }
 167
 168             taxonomyfile = validParameter.validFile(parameters, "taxonomy", true);
 169                         if (taxonomyfile == "not open") { taxonomyfile = ""; abort = true; }
 170                         else if (taxonomyfile == "not found") { taxonomyfile = "";
 171                 taxonomyfile = m->getTaxonomyFile();
 172                 if (taxonomyfile != "") {  m->mothurOut("Using " + taxonomyfile + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
 173                 else { m->mothurOut("No valid current files. You must provide a taxonomy file."); m->mothurOutEndLine(); abort = true; }
 174             }else { m->setTaxonomyFile(taxonomyfile); }
 175
 176                         namefile = validParameter.validFile(parameters, "name", true);
 177                         if (namefile == "not open") { namefile = ""; abort = true; }
 178                         else if (namefile == "not found") { namefile = ""; }
 179                         else { m->setNameFile(namefile); }
 180
 181                         groupfile = validParameter.validFile(parameters, "group", true);
 182                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 183                         else if (groupfile == "not found") { groupfile = ""; }
 184                         else { m->setGroupFile(groupfile); }
 185
 186             countfile = validParameter.validFile(parameters, "count", true);
 187                         if (countfile == "not open") { countfile = ""; abort = true; }
 188                         else if (countfile == "not found") { countfile = "";  }
 189                         else { m->setCountTableFile(countfile); }
 190
 191             if ((namefile != "") && (countfile != "")) {
 192                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 193             }
 194
 195             if ((groupfile != "") && (countfile != "")) {
 196                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 197             }
 198
 199             string temp = validParameter.validFile(parameters, "cutoff", false);                        if (temp == "not found") { temp = "51"; }
 200                         m->mothurConvert(temp, cutoff);
 201
 202                         if ((cutoff < 51) || (cutoff > 100)) { m->mothurOut("cutoff must be above 50, and no greater than 100."); m->mothurOutEndLine(); abort = true;  }
 203
 204             if (countfile == "") {
 205                 if (namefile == "") {
 206                     vector<string> files; files.push_back(treefile);
 207                     parser.getNameFile(files);
 208                 }
 209                         }
 210                 }
 211         }
 212         catch(exception& e) {
 213                 m->errorOut(e, "ClassifyTreeCommand", "ClassifyTreeCommand");
 214                 exit(1);
 215         }
 216 }
 217 //**********************************************************************************************************************
 218
 219 int ClassifyTreeCommand::execute(){
 220         try {
 221
 222                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 223
 224                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 225
 226                 int start = time(NULL);
 227
 228                 /***************************************************/
 229                 //    reading tree info                                                    //
 230                 /***************************************************/
 231         m->setTreeFile(treefile);
 232
 233         TreeReader* reader = new TreeReader(treefile, groupfile, namefile);
 234         vector<Tree*> T = reader->getTrees();
 235         CountTable* tmap = T[0]->getCountTable();
 236         Tree* outputTree = T[0];
 237         delete reader;
 238
 239         if (namefile != "") { m->readNames(namefile, nameMap, nameCount); }
 240
 241         if (m->control_pressed) { delete tmap;  delete outputTree;  return 0; }
 242
 243         m->readTax(taxonomyfile, taxMap);
 244
 245         /***************************************************/
 246         //              get concensus taxonomies                    //
 247         /***************************************************/
 248         getClassifications(outputTree);
 249         delete outputTree; delete tmap;
 250
 251                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 252
 253                 //set tree file as new current treefile
 254                 if (treefile != "") {
 255                         string current = "";
 256                         itTypes = outputTypes.find("tree");
 257                         if (itTypes != outputTypes.end()) {
 258                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 259                         }
 260                 }
 261
 262                 m->mothurOutEndLine(); m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to find the concensus taxonomies."); m->mothurOutEndLine();
 263                 m->mothurOutEndLine();
 264                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 265                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 266                 m->mothurOutEndLine();
 267
 268                 return 0;
 269         }
 270         catch(exception& e) {
 271                 m->errorOut(e, "ClassifyTreeCommand", "execute");
 272                 exit(1);
 273         }
 274 }
 275 //**********************************************************************************************************************
 276 //traverse tree finding concensus taxonomy at each node
 277 //label node with a number to relate to output summary file
 278 //report all concensus taxonomies to file
 279 int ClassifyTreeCommand::getClassifications(Tree*& T){
 280         try {
 281
 282                 string thisOutputDir = outputDir;
 283                 if (outputDir == "") {  thisOutputDir += m->hasPath(treefile);  }
 284         map<string, string> variables;
 285         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(treefile));
 286                 string outputFileName = getOutputFileName("summary", variables);
 287                 outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 288
 289                 ofstream out;
 290                 m->openOutputFile(outputFileName, out);
 291                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 292
 293                 //print headings
 294                 out << "TreeNode\t";
 295                 if (groupfile != "") { out << "Group\t"; }
 296         out << "NumRep\tTaxonomy" << endl;
 297
 298                 string treeOutputDir = outputDir;
 299                 if (outputDir == "") {  treeOutputDir += m->hasPath(treefile);  }
 300         variables["[filename]"] = treeOutputDir + m->getRootName(m->getSimpleName(treefile));
 301                 string outputTreeFileName = getOutputFileName("tree", variables);
 302
 303                 //create a map from tree node index to names of descendants, save time later
 304                 map<int, map<string, set<string> > > nodeToDescendants; //node# -> (groupName -> groupMembers)
 305                 for (int i = 0; i < T->getNumNodes(); i++) {
 306                         if (m->control_pressed) { return 0; }
 307
 308                         nodeToDescendants[i] = getDescendantList(T, i, nodeToDescendants);
 309                 }
 310
 311                 //for each node
 312                 for (int i = T->getNumLeaves(); i < T->getNumNodes(); i++) {
 313
 314                         if (m->control_pressed) { out.close(); return 0; }
 315
 316                         string tax = "not classifed";
 317             int size;
 318             if (groupfile != "") {
 319                 for (map<string, set<string> >::iterator itGroups = nodeToDescendants[i].begin(); itGroups != nodeToDescendants[i].end(); itGroups++) {
 320                     if (itGroups->first != "AllGroups") {
 321                         tax = getTaxonomy(itGroups->second, size);
 322                         out << (i+1) << '\t' << itGroups->first << '\t' << size << '\t' << tax << endl;
 323                     }
 324                 }
 325             }else {
 326                 string group = "AllGroups";
 327                 tax = getTaxonomy(nodeToDescendants[i][group], size);
 328                 out << (i+1) << '\t' << size << '\t' << tax << endl;
 329             }
 330
 331                         T->tree[i].setLabel((i+1));
 332                 }
 333                 out.close();
 334
 335                 ofstream outTree;
 336                 m->openOutputFile(outputTreeFileName, outTree);
 337                 outputNames.push_back(outputTreeFileName); outputTypes["tree"].push_back(outputTreeFileName);
 338                 T->print(outTree, "both");
 339                 outTree.close();
 340
 341                 return 0;
 342         }
 343         catch(exception& e) {
 344                 m->errorOut(e, "ClassifyTreeCommand", "GetConcensusTaxonomies");
 345                 exit(1);
 346         }
 347 }
 348 //**********************************************************************************************************************
 349 string ClassifyTreeCommand::getTaxonomy(set<string> names, int& size) {
 350         try{
 351                 string conTax = "";
 352         size = 0;
 353
 354                 //create a tree containing sequences from this bin
 355                 PhyloTree* phylo = new PhyloTree();
 356
 357                 for (set<string>::iterator it = names.begin(); it != names.end(); it++) {
 358
 359
 360                         //if namesfile include the names
 361                         if (namefile != "") {
 362
 363                                 //is this sequence in the name file - namemap maps seqName -> repSeqName
 364                                 map<string, string>::iterator it2 = nameMap.find(*it);
 365
 366                                 if (it2 == nameMap.end()) { //this name is not in name file, skip it
 367                                         m->mothurOut((*it) + " is not in your name file.  I will not include it in the consensus."); m->mothurOutEndLine();
 368                                 }else{
 369
 370                                         //is this sequence in the taxonomy file - look for repSeqName since we are assuming the taxonomy file is unique
 371                                         map<string, string>::iterator itTax = taxMap.find((it2->second));
 372
 373                                         if (itTax == taxMap.end()) { //this name is not in taxonomy file, skip it
 374
 375                                                 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(); }
 376                                                 else {  m->mothurOut((*it) + " is not in your taxonomy file.  I will not include it in the consensus."); m->mothurOutEndLine(); }
 377                                         }else{
 378                                                 //add seq to tree
 379                         int num = nameCount[(*it)]; // we know its there since we found it in nameMap
 380                                                 for (int i = 0; i < num; i++) {  phylo->addSeqToTree((*it)+toString(i), it2->second);  }
 381                         size += num;
 382                                         }
 383                                 }
 384
 385                         }else{
 386                                 //is this sequence in the taxonomy file - look for repSeqName since we are assuming the taxonomy file is unique
 387                                 map<string, string>::iterator itTax = taxMap.find((*it));
 388
 389                                 if (itTax == taxMap.end()) { //this name is not in taxonomy file, skip it
 390                                         m->mothurOut((*it) + " is not in your taxonomy file.  I will not include it in the consensus."); m->mothurOutEndLine();
 391                                 }else{
 392                                         if (countfile != "") {
 393                         int numDups = ct->getNumSeqs((*it));
 394                         for (int j = 0; j < numDups; j++) {  phylo->addSeqToTree((*it), itTax->second);  }
 395                         size += numDups;
 396                     }else{
 397                         //add seq to tree
 398                         phylo->addSeqToTree((*it), itTax->second);
 399                         size++;
 400                     }                           }
 401                         }
 402
 403                         if (m->control_pressed) { delete phylo; return conTax; }
 404
 405                 }
 406
 407                 //build tree
 408                 phylo->assignHeirarchyIDs(0);
 409
 410                 TaxNode currentNode = phylo->get(0);
 411                 int myLevel = 0;
 412                 //at each level
 413                 while (currentNode.children.size() != 0) { //you still have more to explore
 414
 415                         TaxNode bestChild;
 416                         int bestChildSize = 0;
 417
 418                         //go through children
 419                         for (map<string, int>::iterator itChild = currentNode.children.begin(); itChild != currentNode.children.end(); itChild++) {
 420
 421                                 TaxNode temp = phylo->get(itChild->second);
 422
 423                                 //select child with largest accesions - most seqs assigned to it
 424                                 if (temp.accessions.size() > bestChildSize) {
 425                                         bestChild = phylo->get(itChild->second);
 426                                         bestChildSize = temp.accessions.size();
 427                                 }
 428
 429                         }
 430
 431                         //is this taxonomy above cutoff
 432                         int consensusConfidence = ceil((bestChildSize / (float) size) * 100);
 433
 434                         if (consensusConfidence >= cutoff) { //if yes, add it
 435                 conTax += bestChild.name + "(" + toString(consensusConfidence) + ");";
 436                                 myLevel++;
 437                         }else{ //if no, quit
 438                                 break;
 439                         }
 440
 441                         //move down a level
 442                         currentNode = bestChild;
 443                 }
 444
 445                 if (myLevel != phylo->getMaxLevel()) {
 446                         while (myLevel != phylo->getMaxLevel()) {
 447                                 conTax += "unclassified;";
 448                                 myLevel++;
 449                         }
 450                 }
 451                 if (conTax == "") {  conTax = "no_consensus;";  }
 452
 453                 delete phylo;
 454
 455         return conTax;
 456
 457         }
 458         catch(exception& e) {
 459                 m->errorOut(e, "ClassifyTreeCommand", "getTaxonomy");
 460                 exit(1);
 461         }
 462 }
 463
 464 //**********************************************************************************************************************
 465 map<string, set<string> > ClassifyTreeCommand::getDescendantList(Tree*& T, int i, map<int, map<string, set<string> > > descendants){
 466         try {
 467                 map<string ,set<string> > names;
 468
 469                 map<string ,set<string> >::iterator it;
 470         map<string ,set<string> >::iterator it2;
 471
 472                 int lc = T->tree[i].getLChild();
 473                 int rc = T->tree[i].getRChild();
 474        // TreeMap* tmap = T->getTreeMap();
 475
 476                 if (lc == -1) { //you are a leaf your only descendant is yourself
 477             vector<string> groups = T->tree[i].getGroup();
 478             set<string> mynames; mynames.insert(T->tree[i].getName());
 479             for (int j = 0; j < groups.size(); j++) { names[groups[j]] = mynames;   } //mygroup -> me
 480             names["AllGroups"] = mynames;
 481                 }else{ //your descedants are the combination of your childrens descendants
 482                         names = descendants[lc];
 483                         for (it = descendants[rc].begin(); it != descendants[rc].end(); it++) {
 484                 it2 = names.find(it->first); //do we already have this group
 485                 if (it2 == names.end()) { //nope, so add it
 486                     names[it->first] = it->second;
 487                 }else {
 488                     for (set<string>::iterator it3 = (it->second).begin(); it3 != (it->second).end(); it3++) {
 489                         names[it->first].insert(*it3);
 490                     }
 491                 }
 492
 493                         }
 494                 }
 495
 496                 return names;
 497         }
 498         catch(exception& e) {
 499                 m->errorOut(e, "ClassifyTreeCommand", "getDescendantList");
 500                 exit(1);
 501         }
 502 }
 503 /*****************************************************************/
 504
 505