indicatorcommand.cpp

   1 /*
   2  *  indicatorcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/12/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "indicatorcommand.h"
  11 #include "sharedutilities.h"
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> IndicatorCommand::setParameters(){
  16         try {
  17                 CommandParameter pdesign("design", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pdesign);
  18                 CommandParameter pshared("shared", "InputTypes", "", "", "SharedRel", "SharedRel", "none",false,false); parameters.push_back(pshared);
  19                 CommandParameter prelabund("relabund", "InputTypes", "", "", "SharedRel", "SharedRel", "none",false,false); parameters.push_back(prelabund);
  20                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  21                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  22                 CommandParameter ptree("tree", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptree);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "IndicatorCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string IndicatorCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The indicator command reads a shared or relabund file and a tree file, and outputs a .indicator.tre and .indicator.summary file. \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 summary file lists the indicator value for each OTU for each node.\n";
  42                 helpString += "The indicator command parameters are tree, groups, shared, relabund, design and label. The tree parameter is required as well as either shared or relabund.\n";
  43                 helpString += "The design parameter allows you to provide a design file to relate the tree to the shared or relabund file.\n";
  44                 helpString += "The groups parameter allows you to specify which of the groups in your shared or relabund you would like analyzed, or if you provide a design file the groups in your design file.  The groups may be entered separated by dashes.\n";
  45                 helpString += "The label parameter indicates at what distance your tree relates to the shared or relabund.\n";
  46                 helpString += "The indicator command should be used in the following format: indicator(tree=test.tre, shared=test.shared, label=0.03)\n";
  47                 helpString += "Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n";
  48                 return helpString;
  49         }
  50         catch(exception& e) {
  51                 m->errorOut(e, "IndicatorCommand", "getHelpString");
  52                 exit(1);
  53         }
  54 }
  55
  56 //**********************************************************************************************************************
  57 IndicatorCommand::IndicatorCommand(){
  58         try {
  59                 abort = true; calledHelp = true;
  60                 setParameters();
  61                 vector<string> tempOutNames;
  62                 outputTypes["tree"] = tempOutNames;
  63                 outputTypes["summary"] = tempOutNames;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "IndicatorCommand", "IndicatorCommand");
  67                 exit(1);
  68         }
  69 }
  70 //**********************************************************************************************************************
  71 IndicatorCommand::IndicatorCommand(string option)  {
  72         try {
  73                 abort = false; calledHelp = false;
  74
  75                 //allow user to run help
  76                 if(option == "help") { help(); abort = true; calledHelp = true; }
  77                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  78
  79                 else {
  80                         vector<string> myArray = setParameters();
  81
  82                         OptionParser parser(option);
  83                         map<string, string> parameters = parser.getParameters();
  84
  85                         ValidParameters validParameter;
  86                         map<string, string>::iterator it;
  87
  88                         //check to make sure all parameters are valid for command
  89                         for (it = parameters.begin(); it != parameters.end(); it++) {
  90                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  91                         }
  92
  93                         m->runParse = true;
  94                         m->Groups.clear();
  95                         m->namesOfGroups.clear();
  96                         m->Treenames.clear();
  97                         m->names.clear();
  98
  99                         vector<string> tempOutNames;
 100                         outputTypes["tree"] = tempOutNames;
 101                         outputTypes["summary"] = tempOutNames;
 102
 103                         //if the user changes the input directory command factory will send this info to us in the output parameter
 104                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 105                         if (inputDir == "not found"){   inputDir = "";          }
 106                         else {
 107                                 string path;
 108                                 it = parameters.find("tree");
 109                                 //user has given a template file
 110                                 if(it != parameters.end()){
 111                                         path = m->hasPath(it->second);
 112                                         //if the user has not given a path then, add inputdir. else leave path alone.
 113                                         if (path == "") {       parameters["tree"] = inputDir + it->second;             }
 114                                 }
 115
 116                                 it = parameters.find("shared");
 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["shared"] = inputDir + it->second;           }
 122                                 }
 123
 124                                 it = parameters.find("relabund");
 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["relabund"] = inputDir + it->second;         }
 130                                 }
 131
 132                                 it = parameters.find("design");
 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["design"] = inputDir + it->second;           }
 138                                 }
 139                         }
 140
 141                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 142
 143                         //check for required parameters
 144                         treefile = validParameter.validFile(parameters, "tree", true);
 145                         if (treefile == "not open") { abort = true; }
 146                         else if (treefile == "not found") {                             //if there is a current design file, use it
 147                                 treefile = m->getTreeFile();
 148                                 if (treefile != "") { m->mothurOut("Using " + treefile + " as input file for the tree parameter."); m->mothurOutEndLine(); }
 149                                 else {  m->mothurOut("You have no current tree file and the tree parameter is required."); m->mothurOutEndLine(); abort = true; }
 150                         }else { m->setTreeFile(treefile); }
 151
 152                         sharedfile = validParameter.validFile(parameters, "shared", true);
 153                         if (sharedfile == "not open") { abort = true; }
 154                         else if (sharedfile == "not found") { sharedfile = ""; }
 155                         else { inputFileName = sharedfile; m->setSharedFile(sharedfile); }
 156
 157                         relabundfile = validParameter.validFile(parameters, "relabund", true);
 158                         if (relabundfile == "not open") { abort = true; }
 159                         else if (relabundfile == "not found") { relabundfile = ""; }
 160                         else { inputFileName = relabundfile; m->setRelAbundFile(relabundfile); }
 161
 162                         designfile = validParameter.validFile(parameters, "design", true);
 163                         if (designfile == "not open") { abort = true; }
 164                         else if (designfile == "not found") { designfile = ""; }
 165                         else { m->setDesignFile(designfile); }
 166
 167                         groups = validParameter.validFile(parameters, "groups", false);
 168                         if (groups == "not found") { groups = "";  Groups.push_back("all"); }
 169                         else { m->splitAtDash(groups, Groups);  }
 170                         m->Groups = Groups;
 171
 172                         label = validParameter.validFile(parameters, "label", false);
 173                         if (label == "not found") { label = ""; m->mothurOut("You did not provide a label, I will use the first label in your inputfile."); m->mothurOutEndLine(); label=""; }
 174
 175                         if ((relabundfile == "") && (sharedfile == "")) {
 176                                 //is there are current file available for either of these?
 177                                 //give priority to shared, then relabund
 178                                 sharedfile = m->getSharedFile();
 179                                 if (sharedfile != "") {  inputFileName = sharedfile; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 180                                 else {
 181                                         relabundfile = m->getRelAbundFile();
 182                                         if (relabundfile != "") {  inputFileName = relabundfile; m->mothurOut("Using " + relabundfile + " as input file for the relabund parameter."); m->mothurOutEndLine(); }
 183                                         else {
 184                                                 m->mothurOut("No valid current files. You must provide a shared or relabund."); m->mothurOutEndLine();
 185                                                 abort = true;
 186                                         }
 187                                 }
 188                         }
 189
 190                         if ((relabundfile != "") && (sharedfile != "")) { m->mothurOut("You may not use both a shared and relabund file."); m->mothurOutEndLine(); abort = true;  }
 191
 192                 }
 193         }
 194         catch(exception& e) {
 195                 m->errorOut(e, "IndicatorCommand", "IndicatorCommand");
 196                 exit(1);
 197         }
 198 }
 199 //**********************************************************************************************************************
 200
 201 int IndicatorCommand::execute(){
 202         try {
 203
 204                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 205
 206                 //read designfile if given and set up globaldatas groups for read of sharedfiles
 207                 if (designfile != "") {
 208                         designMap = new GroupMap(designfile);
 209                         designMap->readDesignMap();
 210
 211                         //fill Groups - checks for "all" and for any typo groups
 212                         SharedUtil* util = new SharedUtil();
 213                         util->setGroups(Groups, designMap->namesOfGroups);
 214                         delete util;
 215
 216                         //loop through the Groups and fill Globaldata's Groups with the design file info
 217                         m->Groups = designMap->getNamesSeqs(Groups);
 218                 }
 219
 220                 /***************************************************/
 221                 // use smart distancing to get right sharedRabund  //
 222                 /***************************************************/
 223                 if (sharedfile != "") {
 224                         getShared();
 225                         if (m->control_pressed) { if (designfile != "") { delete designMap; } for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } return 0; }
 226                         if (lookup[0] == NULL) { m->mothurOut("[ERROR] reading shared file."); m->mothurOutEndLine(); return 0; }
 227                 }else {
 228                         getSharedFloat();
 229                         if (m->control_pressed) {  if (designfile != "") { delete designMap; } for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } return 0; }
 230                         if (lookupFloat[0] == NULL) { m->mothurOut("[ERROR] reading relabund file."); m->mothurOutEndLine(); return 0; }
 231                 }
 232
 233                 //reset Globaldatas groups if needed
 234                 if (designfile != "") { m->Groups = Groups; }
 235
 236                 /***************************************************/
 237                 //    reading tree info                                                    //
 238                 /***************************************************/
 239                 string groupfile = "";
 240                 m->setTreeFile(treefile);
 241                 Tree* tree = new Tree(treefile); delete tree;  //extracts names from tree to make faked out groupmap
 242                 treeMap = new TreeMap();
 243                 bool mismatch = false;
 244
 245                 for (int i = 0; i < m->Treenames.size(); i++) {
 246                         //sanity check - is this a group that is not in the sharedfile?
 247                         if (designfile == "") {
 248                                 if (!(m->inUsersGroups(m->Treenames[i], m->namesOfGroups))) {
 249                                         m->mothurOut("[ERROR]: " + m->Treenames[i] + " is not a group in your shared or relabund file."); m->mothurOutEndLine();
 250                                         mismatch = true;
 251                                 }
 252                                 treeMap->addSeq(m->Treenames[i], "Group1");
 253                         }else{
 254                                 vector<string> myGroups; myGroups.push_back(m->Treenames[i]);
 255                                 vector<string> myNames = designMap->getNamesSeqs(myGroups);
 256
 257                                 for(int k = 0; k < myNames.size(); k++) {
 258                                         if (!(m->inUsersGroups(myNames[k], m->namesOfGroups))) {
 259                                                 m->mothurOut("[ERROR]: " + myNames[k] + " is not a group in your shared or relabund file."); m->mothurOutEndLine();
 260                                                 mismatch = true;
 261                                         }
 262                                 }
 263                                 treeMap->addSeq(m->Treenames[i], "Group1");
 264                         }
 265                 }
 266
 267                 if ((designfile != "") && (m->Treenames.size() != Groups.size())) { m->mothurOut("[ERROR]: You design file does not match your tree, aborting."); m->mothurOutEndLine(); mismatch = true; }
 268
 269                 if (mismatch) { //cleanup and exit
 270                         if (designfile != "") { delete designMap; }
 271                         if (sharedfile != "") {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } }
 272                         else { for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } }
 273                         delete treeMap;
 274                         return 0;
 275                 }
 276
 277                 read = new ReadNewickTree(treefile);
 278                 int readOk = read->read(treeMap);
 279
 280                 if (readOk != 0) { m->mothurOut("Read Terminated."); m->mothurOutEndLine(); delete treeMap; delete read; return 0; }
 281
 282                 vector<Tree*> T = read->getTrees();
 283
 284                 delete read;
 285
 286                 if (m->control_pressed) {
 287                         if (designfile != "") { delete designMap; }
 288                         if (sharedfile != "") {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } }
 289                         else { for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } }
 290                         for (int i = 0; i < T.size(); i++) {  delete T[i];  }  delete treeMap; return 0;
 291                 }
 292
 293                 T[0]->assembleTree();
 294
 295                 /***************************************************/
 296                 //    create ouptut tree - respecting pickedGroups //
 297                 /***************************************************/
 298                 Tree* outputTree = new Tree(m->Groups.size(), treeMap);
 299
 300                 outputTree->getSubTree(T[0], m->Groups);
 301                 outputTree->assembleTree();
 302
 303                 //no longer need original tree, we have output tree to use and label
 304                 for (int i = 0; i < T.size(); i++) {  delete T[i];  }
 305
 306
 307                 if (m->control_pressed) {
 308                         if (designfile != "") { delete designMap; }
 309                         if (sharedfile != "") {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } }
 310                         else { for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } }
 311                         delete outputTree; delete treeMap;  return 0;
 312                 }
 313
 314                 /***************************************************/
 315                 //              get indicator species values                       //
 316                 /***************************************************/
 317                 GetIndicatorSpecies(outputTree);
 318
 319                 if (designfile != "") { delete designMap; }
 320                 if (sharedfile != "") {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } }
 321                 else { for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } }
 322                 delete outputTree; delete treeMap;
 323
 324
 325                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 326
 327                 //set tree file as new current treefile
 328                 string current = "";
 329                 itTypes = outputTypes.find("tree");
 330                 if (itTypes != outputTypes.end()) {
 331                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTreeFile(current); }
 332                 }
 333
 334                 m->mothurOutEndLine();
 335                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 336                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 337                 m->mothurOutEndLine();
 338
 339                 return 0;
 340         }
 341         catch(exception& e) {
 342                 m->errorOut(e, "IndicatorCommand", "execute");
 343                 exit(1);
 344         }
 345 }
 346 //**********************************************************************************************************************
 347 //traverse tree finding indicator species values for each otu at each node
 348 //label node with otu number that has highest indicator value
 349 //report all otu values to file
 350 int IndicatorCommand::GetIndicatorSpecies(Tree*& T){
 351         try {
 352
 353                 string thisOutputDir = outputDir;
 354                 if (outputDir == "") {  thisOutputDir += m->hasPath(inputFileName);  }
 355                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(inputFileName)) + "indicator.summary";
 356                 outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 357
 358                 ofstream out;
 359                 m->openOutputFile(outputFileName, out);
 360                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 361
 362                 int numBins = 0;
 363                 if (sharedfile != "") { numBins = lookup[0]->getNumBins(); }
 364                 else { numBins = lookupFloat[0]->getNumBins(); }
 365
 366                 //print headings
 367                 out << "TreeNode\t";
 368                 for (int i = 0; i < numBins; i++) { out << "OTU-" << (i+1) << '\t'; }
 369                 out << endl;
 370
 371                 string treeOutputDir = outputDir;
 372                 if (outputDir == "") {  treeOutputDir += m->hasPath(treefile);  }
 373                 string outputTreeFileName = treeOutputDir + m->getRootName(m->getSimpleName(treefile)) + "indicator.tre";
 374
 375
 376                 //create a map from tree node index to names of descendants, save time later to know which sharedRabund you need
 377                 map<int, set<string> > nodeToDescendants;
 378                 map<int, set<int> > descendantNodes;
 379                 for (int i = 0; i < T->getNumNodes(); i++) {
 380                         if (m->control_pressed) { return 0; }
 381
 382                         nodeToDescendants[i] = getDescendantList(T, i, nodeToDescendants, descendantNodes);
 383                 }
 384
 385                 //you need the distances to leaf to decide grouping below
 386                 //this will also set branch lengths if the tree does not include them
 387                 map<int, float> distToRoot = getDistToRoot(T);
 388
 389                 //for each node
 390                 for (int i = T->getNumLeaves(); i < T->getNumNodes(); i++) {
 391
 392                         if (m->control_pressed) { out.close(); return 0; }
 393
 394                         /*****************************************************/
 395                         //create vectors containing rabund info                          //
 396                         /*****************************************************/
 397
 398                         vector<float> indicatorValues; //size of numBins
 399
 400                         if (sharedfile != "") {
 401                                 vector< vector<SharedRAbundVector*> > groupings;
 402
 403                                 //get nodes that will be a valid grouping
 404                                 //you are valid if you are not one of my descendants
 405                                 //AND your distToRoot is >= mine
 406                                 //AND you were not added as part of a larger grouping. Largest nodes are added first.
 407
 408                                 set<string> groupsAlreadyAdded;
 409                                 //create a grouping with my grouping
 410                                 vector<SharedRAbundVector*> subset;
 411                                 int count = 0;
 412                                 int doneCount = nodeToDescendants[i].size();
 413                                 for (int k = 0; k < lookup.size(); k++) {
 414                                         //is this descendant of i
 415                                         if ((nodeToDescendants[i].count(lookup[k]->getGroup()) != 0)) {
 416                                                 subset.push_back(lookup[k]);
 417                                                 groupsAlreadyAdded.insert(lookup[k]->getGroup());
 418                                                 count++;
 419                                         }
 420                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 421                                 }
 422                                 if (subset.size() != 0) { groupings.push_back(subset); }
 423
 424
 425                                 for (int j = (T->getNumNodes()-1); j >= 0; j--) {
 426
 427
 428                                         if ((descendantNodes[i].count(j) == 0) && (distToRoot[j] >= distToRoot[i])) {
 429                                                 vector<SharedRAbundVector*> subset;
 430                                                 int count = 0;
 431                                                 int doneCount = nodeToDescendants[j].size();
 432                                                 for (int k = 0; k < lookup.size(); k++) {
 433                                                         //is this descendant of j, and we didn't already add this as part of a larger grouping
 434                                                         if ((nodeToDescendants[j].count(lookup[k]->getGroup()) != 0) && (groupsAlreadyAdded.count(lookup[k]->getGroup()) == 0)) {
 435                                                                 subset.push_back(lookup[k]);
 436                                                                 groupsAlreadyAdded.insert(lookup[k]->getGroup());
 437                                                                 count++;
 438                                                         }
 439                                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 440                                                 }
 441
 442                                                 //if subset.size == 0 then the node was added as part of a larger grouping
 443                                                 if (subset.size() != 0) { groupings.push_back(subset); }
 444                                         }
 445                                 }
 446
 447                                 if (groupsAlreadyAdded.size() != lookup.size()) {  m->mothurOut("[ERROR]: could not make proper groupings."); m->mothurOutEndLine(); }
 448
 449                                 indicatorValues = getValues(groupings);
 450
 451                         }else {
 452                                 vector< vector<SharedRAbundFloatVector*> > groupings;
 453
 454                                 //get nodes that will be a valid grouping
 455                                 //you are valid if you are not one of my descendants
 456                                 //AND your distToRoot is >= mine
 457                                 //AND you were not added as part of a larger grouping. Largest nodes are added first.
 458
 459                                 set<string> groupsAlreadyAdded;
 460                                 //create a grouping with my grouping
 461                                 vector<SharedRAbundFloatVector*> subset;
 462                                 int count = 0;
 463                                 int doneCount = nodeToDescendants[i].size();
 464                                 for (int k = 0; k < lookupFloat.size(); k++) {
 465                                         //is this descendant of i
 466                                         if ((nodeToDescendants[i].count(lookupFloat[k]->getGroup()) != 0)) {
 467                                                 subset.push_back(lookupFloat[k]);
 468                                                 groupsAlreadyAdded.insert(lookupFloat[k]->getGroup());
 469                                                 count++;
 470                                         }
 471                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 472                                 }
 473                                 if (subset.size() != 0) { groupings.push_back(subset); }
 474
 475                                 for (int j = (T->getNumNodes()-1); j >= 0; j--) {
 476                                         if ((descendantNodes[i].count(j) == 0) && (distToRoot[j] >= distToRoot[i])) {
 477                                                 vector<SharedRAbundFloatVector*> subset;
 478                                                 int count = 0;
 479                                                 int doneCount = nodeToDescendants[j].size();
 480                                                 for (int k = 0; k < lookupFloat.size(); k++) {
 481                                                         //is this descendant of j, and we didn't already add this as part of a larger grouping
 482                                                         if ((nodeToDescendants[j].count(lookupFloat[k]->getGroup()) != 0) && (groupsAlreadyAdded.count(lookupFloat[k]->getGroup()) == 0)) {
 483                                                                 subset.push_back(lookupFloat[k]);
 484                                                                 groupsAlreadyAdded.insert(lookupFloat[k]->getGroup());
 485                                                                 count++;
 486                                                         }
 487                                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 488                                                 }
 489
 490                                                 //if subset.size == 0 then the node was added as part of a larger grouping
 491                                                 if (subset.size() != 0) { groupings.push_back(subset); }
 492                                         }
 493                                 }
 494
 495                                 if (groupsAlreadyAdded.size() != lookupFloat.size()) { m->mothurOut("[ERROR]: could not make proper groupings."); m->mothurOutEndLine(); }
 496
 497                                 indicatorValues = getValues(groupings);
 498                         }
 499
 500                         if (m->control_pressed) { out.close(); return 0; }
 501
 502
 503                         /******************************************************/
 504                         //output indicator values to table form + label tree  //
 505                         /*****************************************************/
 506                         out << (i+1) << '\t';
 507                         for (int j = 0; j < indicatorValues.size(); j++) {
 508
 509                                 if (m->control_pressed) { out.close(); return 0; }
 510
 511                                 out << indicatorValues[j] << '\t';
 512                         }
 513                         out << endl;
 514
 515                         T->tree[i].setLabel((i+1));
 516
 517                 }
 518                 out.close();
 519
 520                 ofstream outTree;
 521                 m->openOutputFile(outputTreeFileName, outTree);
 522                 outputNames.push_back(outputTreeFileName); outputTypes["tree"].push_back(outputTreeFileName);
 523
 524                 T->print(outTree, "both");
 525                 outTree.close();
 526
 527                 return 0;
 528         }
 529         catch(exception& e) {
 530                 m->errorOut(e, "IndicatorCommand", "GetIndicatorSpecies");
 531                 exit(1);
 532         }
 533 }
 534 //**********************************************************************************************************************
 535 vector<float> IndicatorCommand::getValues(vector< vector<SharedRAbundFloatVector*> >& groupings){
 536         try {
 537                 vector<float> values;
 538
 539                 //for each otu
 540                 for (int i = 0; i < groupings[0][0]->getNumBins(); i++) {
 541
 542                         if (m->control_pressed) { return values; }
 543
 544                         vector<float> terms;
 545                         float AijDenominator = 0.0;
 546                         vector<float> Bij;
 547                         //get overall abundance of each grouping
 548                         for (int j = 0; j < groupings.size(); j++) {
 549
 550                                 float totalAbund = 0;
 551                                 int numNotZero = 0;
 552                                 for (int k = 0; k < groupings[j].size(); k++) {
 553                                         totalAbund += groupings[j][k]->getAbundance(i);
 554                                         if (groupings[j][k]->getAbundance(i) != 0) { numNotZero++; }
 555                                 }
 556
 557                                 float Aij = (totalAbund / (float) groupings[j].size());
 558                                 terms.push_back(Aij);
 559
 560                                 //percentage of sites represented
 561                                 Bij.push_back(numNotZero / (float) groupings[j].size());
 562
 563                                 AijDenominator += Aij;
 564                         }
 565
 566                         float maxIndVal = 0.0;
 567                         for (int j = 0; j < terms.size(); j++) {
 568                                 float thisAij = (terms[j] / AijDenominator);
 569                                 float thisValue = thisAij * Bij[j] * 100.0;
 570
 571                                 //save largest
 572                                 if (thisValue > maxIndVal) { maxIndVal = thisValue; }
 573                         }
 574
 575                         values.push_back(maxIndVal);
 576                 }
 577
 578                 return values;
 579         }
 580         catch(exception& e) {
 581                 m->errorOut(e, "IndicatorCommand", "getValues");
 582                 exit(1);
 583         }
 584 }
 585 //**********************************************************************************************************************
 586 //same as above, just data type difference
 587 vector<float> IndicatorCommand::getValues(vector< vector<SharedRAbundVector*> >& groupings){
 588         try {
 589                 vector<float> values;
 590
 591         /*for (int j = 0; j < groupings.size(); j++) {
 592                 cout << "grouping " << j << endl;
 593                 for (int k = 0; k < groupings[j].size(); k++) {
 594                         cout << groupings[j][k]->getGroup() << endl;
 595                 }
 596         }*/
 597                 //for each otu
 598                 for (int i = 0; i < groupings[0][0]->getNumBins(); i++) {
 599                         vector<float> terms;
 600                         float AijDenominator = 0.0;
 601                         vector<float> Bij;
 602                         //get overall abundance of each grouping
 603                         for (int j = 0; j < groupings.size(); j++) {
 604
 605                                 int totalAbund = 0.0;
 606                                 int numNotZero = 0;
 607                                 for (int k = 0; k < groupings[j].size(); k++) {
 608                                         totalAbund += groupings[j][k]->getAbundance(i);
 609                                         if (groupings[j][k]->getAbundance(i) != 0.0) { numNotZero++; }
 610
 611                                 }
 612
 613                                 float Aij = (totalAbund / (float) groupings[j].size());
 614                                 terms.push_back(Aij);
 615
 616                                 //percentage of sites represented
 617                                 Bij.push_back(numNotZero / (float) groupings[j].size());
 618
 619                                 AijDenominator += Aij;
 620                         }
 621
 622                         float maxIndVal = 0.0;
 623                         for (int j = 0; j < terms.size(); j++) {
 624                                 float thisAij = (terms[j] / AijDenominator);
 625                                 float thisValue = thisAij * Bij[j] * 100.0;
 626
 627                                 //save largest
 628                                 if (thisValue > maxIndVal) { maxIndVal = thisValue; }
 629                         }
 630
 631                         values.push_back(maxIndVal);
 632                 }
 633
 634                 return values;
 635         }
 636         catch(exception& e) {
 637                 m->errorOut(e, "IndicatorCommand", "getValues");
 638                 exit(1);
 639         }
 640 }
 641 //**********************************************************************************************************************
 642 //you need the distances to root to decide groupings
 643 //this will also set branch lengths if the tree does not include them
 644 map<int, float> IndicatorCommand::getDistToRoot(Tree*& T){
 645         try {
 646                 map<int, float> dists;
 647
 648                 bool hasBranchLengths = false;
 649                 for (int i = 0; i < T->getNumNodes(); i++) {
 650                         if (T->tree[i].getBranchLength() > 0.0) {  hasBranchLengths = true; break; }
 651                 }
 652
 653                 //set branchlengths if needed
 654                 if (!hasBranchLengths) {
 655                         for (int i = 0; i < T->getNumNodes(); i++) {
 656
 657                                 int lc = T->tree[i].getLChild();
 658                                 int rc = T->tree[i].getRChild();
 659
 660                                 if (lc == -1) { // you are a leaf
 661                                         //if you are a leaf set you priliminary length to 1.0, this may adjust later
 662                                         T->tree[i].setBranchLength(1.0);
 663                                         dists[i] = 1.0;
 664                                 }else{ // you are an internal node
 665                                         //look at your children's length to leaf
 666                                         float ldist = dists[lc];
 667                                         float rdist = dists[rc];
 668
 669                                         float greater = ldist;
 670                                         if (rdist > greater) { greater = rdist; dists[i] = ldist + 1.0;}
 671                                         else { dists[i] = rdist + 1.0; }
 672
 673
 674                                         //branch length = difference + 1
 675                                         T->tree[lc].setBranchLength((abs(ldist-greater) + 1.0));
 676                                         T->tree[rc].setBranchLength((abs(rdist-greater) + 1.0));
 677                                 }
 678                         }
 679                 }
 680
 681                 dists.clear();
 682
 683                 for (int i = 0; i < T->getNumNodes(); i++) {
 684
 685                         double sum = 0.0;
 686                         int index = i;
 687
 688                         while(T->tree[index].getParent() != -1){
 689                                 if (T->tree[index].getBranchLength() != -1) {
 690                                         sum += abs(T->tree[index].getBranchLength());
 691                                 }
 692                                 index = T->tree[index].getParent();
 693                         }
 694
 695                         dists[i] = sum;
 696                 }
 697
 698                 return dists;
 699         }
 700         catch(exception& e) {
 701                 m->errorOut(e, "IndicatorCommand", "getLengthToLeaf");
 702                 exit(1);
 703         }
 704 }
 705 //**********************************************************************************************************************
 706 set<string> IndicatorCommand::getDescendantList(Tree*& T, int i, map<int, set<string> > descendants, map<int, set<int> >& nodes){
 707         try {
 708                 set<string> names;
 709
 710                 set<string>::iterator it;
 711
 712                 int lc = T->tree[i].getLChild();
 713                 int rc = T->tree[i].getRChild();
 714
 715                 if (lc == -1) { //you are a leaf your only descendant is yourself
 716                         set<int> temp; temp.insert(i);
 717                         nodes[i] = temp;
 718
 719                         if (designfile == "") {
 720                                 names.insert(T->tree[i].getName());
 721                         }else {
 722                                 vector<string> myGroup; myGroup.push_back(T->tree[i].getName());
 723                                 vector<string> myReps = designMap->getNamesSeqs(myGroup);
 724                                 for (int k = 0; k < myReps.size(); k++) {
 725                                         names.insert(myReps[k]);
 726                                 }
 727                         }
 728
 729                 }else{ //your descedants are the combination of your childrens descendants
 730                         names = descendants[lc];
 731                         nodes[i] = nodes[lc];
 732                         for (it = descendants[rc].begin(); it != descendants[rc].end(); it++) {
 733                                 names.insert(*it);
 734                         }
 735                         for (set<int>::iterator itNum = nodes[rc].begin(); itNum != nodes[rc].end(); itNum++) {
 736                                 nodes[i].insert(*itNum);
 737                         }
 738                         //you are your own descendant
 739                         nodes[i].insert(i);
 740                 }
 741
 742                 return names;
 743         }
 744         catch(exception& e) {
 745                 m->errorOut(e, "IndicatorCommand", "getDescendantList");
 746                 exit(1);
 747         }
 748 }
 749 //**********************************************************************************************************************
 750 int IndicatorCommand::getShared(){
 751         try {
 752                 InputData* input = new InputData(sharedfile, "sharedfile");
 753                 lookup = input->getSharedRAbundVectors();
 754                 string lastLabel = lookup[0]->getLabel();
 755
 756                 if (label == "") { label = lastLabel; delete input; return 0; }
 757
 758                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 759                 set<string> labels; labels.insert(label);
 760                 set<string> processedLabels;
 761                 set<string> userLabels = labels;
 762
 763                 //as long as you are not at the end of the file or done wih the lines you want
 764                 while((lookup[0] != NULL) && (userLabels.size() != 0)) {
 765                         if (m->control_pressed) {  delete input; return 0;  }
 766
 767                         if(labels.count(lookup[0]->getLabel()) == 1){
 768                                 processedLabels.insert(lookup[0]->getLabel());
 769                                 userLabels.erase(lookup[0]->getLabel());
 770                                 break;
 771                         }
 772
 773                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 774                                 string saveLabel = lookup[0]->getLabel();
 775
 776                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 777                                 lookup = input->getSharedRAbundVectors(lastLabel);
 778
 779                                 processedLabels.insert(lookup[0]->getLabel());
 780                                 userLabels.erase(lookup[0]->getLabel());
 781
 782                                 //restore real lastlabel to save below
 783                                 lookup[0]->setLabel(saveLabel);
 784                                 break;
 785                         }
 786
 787                         lastLabel = lookup[0]->getLabel();
 788
 789                         //get next line to process
 790                         //prevent memory leak
 791                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 792                         lookup = input->getSharedRAbundVectors();
 793                 }
 794
 795
 796                 if (m->control_pressed) { delete input; return 0;  }
 797
 798                 //output error messages about any remaining user labels
 799                 set<string>::iterator it;
 800                 bool needToRun = false;
 801                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 802                         m->mothurOut("Your file does not include the label " + *it);
 803                         if (processedLabels.count(lastLabel) != 1) {
 804                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 805                                 needToRun = true;
 806                         }else {
 807                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 808                         }
 809                 }
 810
 811                 //run last label if you need to
 812                 if (needToRun == true)  {
 813                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       } }
 814                         lookup = input->getSharedRAbundVectors(lastLabel);
 815                 }
 816
 817                 delete input;
 818                 return 0;
 819         }
 820         catch(exception& e) {
 821                 m->errorOut(e, "IndicatorCommand", "getShared");
 822                 exit(1);
 823         }
 824 }
 825 //**********************************************************************************************************************
 826 int IndicatorCommand::getSharedFloat(){
 827         try {
 828                 InputData* input = new InputData(relabundfile, "relabund");
 829                 lookupFloat = input->getSharedRAbundFloatVectors();
 830                 string lastLabel = lookupFloat[0]->getLabel();
 831
 832                 if (label == "") { label = lastLabel; delete input; return 0; }
 833
 834                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 835                 set<string> labels; labels.insert(label);
 836                 set<string> processedLabels;
 837                 set<string> userLabels = labels;
 838
 839                 //as long as you are not at the end of the file or done wih the lines you want
 840                 while((lookupFloat[0] != NULL) && (userLabels.size() != 0)) {
 841
 842                         if (m->control_pressed) {  delete input; return 0;  }
 843
 844                         if(labels.count(lookupFloat[0]->getLabel()) == 1){
 845                                 processedLabels.insert(lookupFloat[0]->getLabel());
 846                                 userLabels.erase(lookupFloat[0]->getLabel());
 847                                 break;
 848                         }
 849
 850                         if ((m->anyLabelsToProcess(lookupFloat[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 851                                 string saveLabel = lookupFloat[0]->getLabel();
 852
 853                                 for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  }
 854                                 lookupFloat = input->getSharedRAbundFloatVectors(lastLabel);
 855
 856                                 processedLabels.insert(lookupFloat[0]->getLabel());
 857                                 userLabels.erase(lookupFloat[0]->getLabel());
 858
 859                                 //restore real lastlabel to save below
 860                                 lookupFloat[0]->setLabel(saveLabel);
 861                                 break;
 862                         }
 863
 864                         lastLabel = lookupFloat[0]->getLabel();
 865
 866                         //get next line to process
 867                         //prevent memory leak
 868                         for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  }
 869                         lookupFloat = input->getSharedRAbundFloatVectors();
 870                 }
 871
 872
 873                 if (m->control_pressed) { delete input; return 0;  }
 874
 875                 //output error messages about any remaining user labels
 876                 set<string>::iterator it;
 877                 bool needToRun = false;
 878                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 879                         m->mothurOut("Your file does not include the label " + *it);
 880                         if (processedLabels.count(lastLabel) != 1) {
 881                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 882                                 needToRun = true;
 883                         }else {
 884                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 885                         }
 886                 }
 887
 888                 //run last label if you need to
 889                 if (needToRun == true)  {
 890                         for (int i = 0; i < lookupFloat.size(); i++) {  if (lookupFloat[i] != NULL) {   delete lookupFloat[i];  } }
 891                         lookupFloat = input->getSharedRAbundFloatVectors(lastLabel);
 892                 }
 893
 894                 delete input;
 895                 return 0;
 896         }
 897         catch(exception& e) {
 898                 m->errorOut(e, "IndicatorCommand", "getShared");
 899         exit(1);
 900         }
 901 }
 902 /*****************************************************************/
 903
 904