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 vector<string> IndicatorCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"tree","shared","relabund","design","label","groups","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "IndicatorCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 vector<string> IndicatorCommand::getRequiredParameters(){
  27         try {
  28                 string Array[] =  {"tree"};
  29                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  30                 return myArray;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "IndicatorCommand", "getRequiredParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 IndicatorCommand::IndicatorCommand(){
  39         try {
  40                 abort = true; calledHelp = true;
  41                 vector<string> tempOutNames;
  42                 outputTypes["tree"] = tempOutNames;
  43                 outputTypes["summary"] = tempOutNames;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "IndicatorCommand", "IndicatorCommand");
  47                 exit(1);
  48         }
  49 }
  50
  51 //**********************************************************************************************************************
  52 vector<string> IndicatorCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "IndicatorCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63 IndicatorCommand::IndicatorCommand(string option)  {
  64         try {
  65                 globaldata = GlobalData::getInstance();
  66                 abort = false; calledHelp = false;
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; calledHelp = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string Array[] =  {"tree","shared","design","relabund","groups","label","outputdir","inputdir"};
  74                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string, string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string, string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         globaldata->newRead();
  88
  89                         vector<string> tempOutNames;
  90                         outputTypes["tree"] = tempOutNames;
  91                         outputTypes["summary"] = tempOutNames;
  92
  93                         //if the user changes the input directory command factory will send this info to us in the output parameter
  94                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  95                         if (inputDir == "not found"){   inputDir = "";          }
  96                         else {
  97                                 string path;
  98                                 it = parameters.find("tree");
  99                                 //user has given a template file
 100                                 if(it != parameters.end()){
 101                                         path = m->hasPath(it->second);
 102                                         //if the user has not given a path then, add inputdir. else leave path alone.
 103                                         if (path == "") {       parameters["tree"] = inputDir + it->second;             }
 104                                 }
 105
 106                                 it = parameters.find("shared");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 112                                 }
 113
 114                                 it = parameters.find("relabund");
 115                                 //user has given a template file
 116                                 if(it != parameters.end()){
 117                                         path = m->hasPath(it->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["relabund"] = inputDir + it->second;         }
 120                                 }
 121
 122                                 it = parameters.find("design");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 128                                 }
 129                         }
 130
 131                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 132
 133                         //check for required parameters
 134                         treefile = validParameter.validFile(parameters, "tree", true);
 135                         if (treefile == "not open") { abort = true; }
 136                         else if (treefile == "not found") { treefile = ""; m->mothurOut("tree is a required parameter for the indicator command."); m->mothurOutEndLine(); abort = true;  }
 137                         else {  globaldata->setTreeFile(treefile);  globaldata->setFormat("tree");      }
 138
 139                         sharedfile = validParameter.validFile(parameters, "shared", true);
 140                         if (sharedfile == "not open") { abort = true; }
 141                         else if (sharedfile == "not found") { sharedfile = ""; }
 142                         else { inputFileName = sharedfile; }
 143
 144                         relabundfile = validParameter.validFile(parameters, "relabund", true);
 145                         if (relabundfile == "not open") { abort = true; }
 146                         else if (relabundfile == "not found") { relabundfile = ""; }
 147                         else { inputFileName = relabundfile; }
 148
 149                         designfile = validParameter.validFile(parameters, "design", true);
 150                         if (designfile == "not open") { abort = true; }
 151                         else if (designfile == "not found") { designfile = ""; }
 152
 153                         groups = validParameter.validFile(parameters, "groups", false);
 154                         if (groups == "not found") { groups = "";  Groups.push_back("all"); }
 155                         else { m->splitAtDash(groups, Groups);  }
 156                         globaldata->Groups = Groups;
 157
 158                         label = validParameter.validFile(parameters, "label", false);
 159                         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=""; }
 160
 161                         if ((relabundfile == "") && (sharedfile == "")) { m->mothurOut("You must provide either a shared or relabund file."); m->mothurOutEndLine(); abort = true;  }
 162
 163                         if ((relabundfile != "") && (sharedfile != "")) { m->mothurOut("You may not use both a shared and relabund file."); m->mothurOutEndLine(); abort = true;  }
 164
 165                 }
 166         }
 167         catch(exception& e) {
 168                 m->errorOut(e, "IndicatorCommand", "IndicatorCommand");
 169                 exit(1);
 170         }
 171 }
 172 //**********************************************************************************************************************
 173
 174 void IndicatorCommand::help(){
 175         try {
 176                 m->mothurOut("The indicator command reads a shared or relabund file and a tree file, and outputs a .indicator.tre and .indicator.summary file. \n");
 177                 m->mothurOut("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");
 178                 m->mothurOut("The summary file lists the indicator value for each OTU for each node.\n");
 179                 m->mothurOut("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");
 180                 m->mothurOut("The design parameter allows you to provide a design file to relate the tree to the shared or relabund file.\n");
 181                 m->mothurOut("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");
 182                 m->mothurOut("The label parameter indicates at what distance your tree relates to the shared or relabund.\n");
 183                 m->mothurOut("The indicator command should be used in the following format: indicator(tree=test.tre, shared=test.shared, label=0.03)\n");
 184                 m->mothurOut("Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n\n");
 185         }
 186         catch(exception& e) {
 187                 m->errorOut(e, "IndicatorCommand", "help");
 188                 exit(1);
 189         }
 190 }
 191
 192 //**********************************************************************************************************************
 193
 194 IndicatorCommand::~IndicatorCommand(){}
 195
 196 //**********************************************************************************************************************
 197
 198 int IndicatorCommand::execute(){
 199         try {
 200
 201                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 202
 203                 //read designfile if given and set up globaldatas groups for read of sharedfiles
 204                 if (designfile != "") {
 205                         designMap = new GroupMap(designfile);
 206                         designMap->readDesignMap();
 207
 208                         //fill Groups - checks for "all" and for any typo groups
 209                         SharedUtil* util = new SharedUtil();
 210                         util->setGroups(Groups, designMap->namesOfGroups);
 211                         delete util;
 212
 213                         //loop through the Groups and fill Globaldata's Groups with the design file info
 214                         globaldata->Groups = designMap->getNamesSeqs(Groups);
 215                 }
 216
 217                 /***************************************************/
 218                 // use smart distancing to get right sharedRabund  //
 219                 /***************************************************/
 220                 if (sharedfile != "") {
 221                         getShared();
 222                         if (m->control_pressed) { if (designfile != "") { delete designMap; } for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } return 0; }
 223                         if (lookup[0] == NULL) { m->mothurOut("[ERROR] reading shared file."); m->mothurOutEndLine(); return 0; }
 224                 }else {
 225                         getSharedFloat();
 226                         if (m->control_pressed) {  if (designfile != "") { delete designMap; } for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } return 0; }
 227                         if (lookupFloat[0] == NULL) { m->mothurOut("[ERROR] reading relabund file."); m->mothurOutEndLine(); return 0; }
 228                 }
 229
 230                 //reset Globaldatas groups if needed
 231                 if (designfile != "") { globaldata->Groups = Groups; }
 232
 233                 /***************************************************/
 234                 //    reading tree info                                                    //
 235                 /***************************************************/
 236                 string groupfile = "";
 237                 Tree* tree = new Tree(treefile); delete tree;  //extracts names from tree to make faked out groupmap
 238
 239                 globaldata->setGroupFile(groupfile);
 240                 treeMap = new TreeMap();
 241                 bool mismatch = false;
 242
 243                 for (int i = 0; i < globaldata->Treenames.size(); i++) {
 244                         //sanity check - is this a group that is not in the sharedfile?
 245                         if (designfile == "") {
 246                                 if (!(m->inUsersGroups(globaldata->Treenames[i], globaldata->gGroupmap->namesOfGroups))) {
 247                                         m->mothurOut("[ERROR]: " + globaldata->Treenames[i] + " is not a group in your shared or relabund file."); m->mothurOutEndLine();
 248                                         mismatch = true;
 249                                 }
 250                                 treeMap->addSeq(globaldata->Treenames[i], "Group1");
 251                         }else{
 252                                 vector<string> myGroups; myGroups.push_back(globaldata->Treenames[i]);
 253                                 vector<string> myNames = designMap->getNamesSeqs(myGroups);
 254
 255                                 for(int k = 0; k < myNames.size(); k++) {
 256                                         if (!(m->inUsersGroups(myNames[k], globaldata->gGroupmap->namesOfGroups))) {
 257                                                 m->mothurOut("[ERROR]: " + myNames[k] + " is not a group in your shared or relabund file."); m->mothurOutEndLine();
 258                                                 mismatch = true;
 259                                         }
 260                                 }
 261                                 treeMap->addSeq(globaldata->Treenames[i], "Group1");
 262                         }
 263                 }
 264
 265                 if ((designfile != "") && (globaldata->Treenames.size() != Groups.size())) { m->mothurOut("[ERROR]: You design file does not match your tree, aborting."); m->mothurOutEndLine(); mismatch = true; }
 266
 267                 if (mismatch) { //cleanup and exit
 268                         if (designfile != "") { delete designMap; }
 269                         if (sharedfile != "") {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } }
 270                         else { for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } }
 271                         delete treeMap;
 272                         return 0;
 273                 }
 274
 275                 globaldata->gTreemap = treeMap;
 276
 277                 read = new ReadNewickTree(treefile);
 278                 int readOk = read->read();
 279
 280                 if (readOk != 0) { m->mothurOut("Read Terminated."); m->mothurOutEndLine(); globaldata->gTree.clear(); delete globaldata->gTreemap; delete read; return 0; }
 281
 282                 vector<Tree*> T = globaldata->gTree;
 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];  } globaldata->gTree.clear(); delete globaldata->gTreemap; return 0;
 291                 }
 292
 293                 T[0]->assembleTree();
 294
 295                 /***************************************************/
 296                 //    create ouptut tree - respecting pickedGroups //
 297                 /***************************************************/
 298                 Tree* outputTree = new Tree(globaldata->Groups.size());
 299
 300                 outputTree->getSubTree(T[0], globaldata->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];  } globaldata->gTree.clear();
 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 globaldata->gTreemap;  return 0;
 312                 }
 313
 314                 /***************************************************/
 315                 //              get indicator species values                       //
 316                 /***************************************************/
 317                 GetIndicatorSpecies(outputTree);
 318
 319                 if (designfile != "") { delete designMap; }
 320
 321                 if (m->control_pressed) {
 322                         if (sharedfile != "") {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } }
 323                         else { for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } }
 324                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 325                         delete outputTree; delete globaldata->gTreemap;  return 0;
 326                 }
 327
 328                 m->mothurOutEndLine();
 329                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 330                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 331                 m->mothurOutEndLine();
 332
 333                 return 0;
 334         }
 335         catch(exception& e) {
 336                 m->errorOut(e, "IndicatorCommand", "execute");
 337                 exit(1);
 338         }
 339 }
 340 //**********************************************************************************************************************
 341 //traverse tree finding indicator species values for each otu at each node
 342 //label node with otu number that has highest indicator value
 343 //report all otu values to file
 344 int IndicatorCommand::GetIndicatorSpecies(Tree*& T){
 345         try {
 346
 347                 string thisOutputDir = outputDir;
 348                 if (outputDir == "") {  thisOutputDir += m->hasPath(inputFileName);  }
 349                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(inputFileName)) + "indicator.summary";
 350                 outputNames.push_back(outputFileName); outputTypes["summary"].push_back(outputFileName);
 351
 352                 ofstream out;
 353                 m->openOutputFile(outputFileName, out);
 354                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 355
 356                 int numBins = 0;
 357                 if (sharedfile != "") { numBins = lookup[0]->getNumBins(); }
 358                 else { numBins = lookupFloat[0]->getNumBins(); }
 359
 360                 //print headings
 361                 out << "TreeNode\t";
 362                 for (int i = 0; i < numBins; i++) { out << "OTU-" << (i+1) << '\t'; }
 363                 out << endl;
 364
 365                 string treeOutputDir = outputDir;
 366                 if (outputDir == "") {  treeOutputDir += m->hasPath(treefile);  }
 367                 string outputTreeFileName = treeOutputDir + m->getRootName(m->getSimpleName(treefile)) + "indicator.tre";
 368
 369
 370                 //create a map from tree node index to names of descendants, save time later to know which sharedRabund you need
 371                 map<int, set<string> > nodeToDescendants;
 372                 map<int, set<int> > descendantNodes;
 373                 for (int i = 0; i < T->getNumNodes(); i++) {
 374                         if (m->control_pressed) { return 0; }
 375
 376                         nodeToDescendants[i] = getDescendantList(T, i, nodeToDescendants, descendantNodes);
 377                 }
 378
 379                 //you need the distances to leaf to decide grouping below
 380                 //this will also set branch lengths if the tree does not include them
 381                 map<int, float> distToRoot = getDistToRoot(T);
 382
 383                 //for each node
 384                 for (int i = T->getNumLeaves(); i < T->getNumNodes(); i++) {
 385
 386                         if (m->control_pressed) { out.close(); return 0; }
 387
 388                         /*****************************************************/
 389                         //create vectors containing rabund info                          //
 390                         /*****************************************************/
 391
 392                         vector<float> indicatorValues; //size of numBins
 393
 394                         if (sharedfile != "") {
 395                                 vector< vector<SharedRAbundVector*> > groupings;
 396
 397                                 //get nodes that will be a valid grouping
 398                                 //you are valid if you are not one of my descendants
 399                                 //AND your distToRoot is >= mine
 400                                 //AND you were not added as part of a larger grouping. Largest nodes are added first.
 401
 402                                 set<string> groupsAlreadyAdded;
 403                                 //create a grouping with my grouping
 404                                 vector<SharedRAbundVector*> subset;
 405                                 int count = 0;
 406                                 int doneCount = nodeToDescendants[i].size();
 407                                 for (int k = 0; k < lookup.size(); k++) {
 408                                         //is this descendant of i
 409                                         if ((nodeToDescendants[i].count(lookup[k]->getGroup()) != 0)) {
 410                                                 subset.push_back(lookup[k]);
 411                                                 groupsAlreadyAdded.insert(lookup[k]->getGroup());
 412                                                 count++;
 413                                         }
 414                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 415                                 }
 416                                 if (subset.size() != 0) { groupings.push_back(subset); }
 417
 418
 419                                 for (int j = (T->getNumNodes()-1); j >= 0; j--) {
 420
 421
 422                                         if ((descendantNodes[i].count(j) == 0) && (distToRoot[j] >= distToRoot[i])) {
 423                                                 vector<SharedRAbundVector*> subset;
 424                                                 int count = 0;
 425                                                 int doneCount = nodeToDescendants[j].size();
 426                                                 for (int k = 0; k < lookup.size(); k++) {
 427                                                         //is this descendant of j, and we didn't already add this as part of a larger grouping
 428                                                         if ((nodeToDescendants[j].count(lookup[k]->getGroup()) != 0) && (groupsAlreadyAdded.count(lookup[k]->getGroup()) == 0)) {
 429                                                                 subset.push_back(lookup[k]);
 430                                                                 groupsAlreadyAdded.insert(lookup[k]->getGroup());
 431                                                                 count++;
 432                                                         }
 433                                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 434                                                 }
 435
 436                                                 //if subset.size == 0 then the node was added as part of a larger grouping
 437                                                 if (subset.size() != 0) { groupings.push_back(subset); }
 438                                         }
 439                                 }
 440
 441                                 if (groupsAlreadyAdded.size() != lookup.size()) {  m->mothurOut("[ERROR]: could not make proper groupings."); m->mothurOutEndLine(); }
 442
 443                                 indicatorValues = getValues(groupings);
 444
 445                         }else {
 446                                 vector< vector<SharedRAbundFloatVector*> > groupings;
 447
 448                                 //get nodes that will be a valid grouping
 449                                 //you are valid if you are not one of my descendants
 450                                 //AND your distToRoot is >= mine
 451                                 //AND you were not added as part of a larger grouping. Largest nodes are added first.
 452
 453                                 set<string> groupsAlreadyAdded;
 454                                 //create a grouping with my grouping
 455                                 vector<SharedRAbundFloatVector*> subset;
 456                                 int count = 0;
 457                                 int doneCount = nodeToDescendants[i].size();
 458                                 for (int k = 0; k < lookupFloat.size(); k++) {
 459                                         //is this descendant of i
 460                                         if ((nodeToDescendants[i].count(lookupFloat[k]->getGroup()) != 0)) {
 461                                                 subset.push_back(lookupFloat[k]);
 462                                                 groupsAlreadyAdded.insert(lookupFloat[k]->getGroup());
 463                                                 count++;
 464                                         }
 465                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 466                                 }
 467                                 if (subset.size() != 0) { groupings.push_back(subset); }
 468
 469                                 for (int j = (T->getNumNodes()-1); j >= 0; j--) {
 470                                         if ((descendantNodes[i].count(j) == 0) && (distToRoot[j] >= distToRoot[i])) {
 471                                                 vector<SharedRAbundFloatVector*> subset;
 472                                                 int count = 0;
 473                                                 int doneCount = nodeToDescendants[j].size();
 474                                                 for (int k = 0; k < lookupFloat.size(); k++) {
 475                                                         //is this descendant of j, and we didn't already add this as part of a larger grouping
 476                                                         if ((nodeToDescendants[j].count(lookupFloat[k]->getGroup()) != 0) && (groupsAlreadyAdded.count(lookupFloat[k]->getGroup()) == 0)) {
 477                                                                 subset.push_back(lookupFloat[k]);
 478                                                                 groupsAlreadyAdded.insert(lookupFloat[k]->getGroup());
 479                                                                 count++;
 480                                                         }
 481                                                         if (count == doneCount) { break; } //quit once you get the rabunds for this grouping
 482                                                 }
 483
 484                                                 //if subset.size == 0 then the node was added as part of a larger grouping
 485                                                 if (subset.size() != 0) { groupings.push_back(subset); }
 486                                         }
 487                                 }
 488
 489                                 if (groupsAlreadyAdded.size() != lookupFloat.size()) { m->mothurOut("[ERROR]: could not make proper groupings."); m->mothurOutEndLine(); }
 490
 491                                 indicatorValues = getValues(groupings);
 492                         }
 493
 494                         if (m->control_pressed) { out.close(); return 0; }
 495
 496
 497                         /******************************************************/
 498                         //output indicator values to table form + label tree  //
 499                         /*****************************************************/
 500                         out << (i+1) << '\t';
 501                         for (int j = 0; j < indicatorValues.size(); j++) {
 502
 503                                 if (m->control_pressed) { out.close(); return 0; }
 504
 505                                 out << indicatorValues[j] << '\t';
 506                         }
 507                         out << endl;
 508
 509                         T->tree[i].setLabel((i+1));
 510
 511                 }
 512                 out.close();
 513
 514                 ofstream outTree;
 515                 m->openOutputFile(outputTreeFileName, outTree);
 516                 outputNames.push_back(outputTreeFileName); outputTypes["tree"].push_back(outputTreeFileName);
 517
 518                 T->print(outTree, "both");
 519                 outTree.close();
 520
 521                 return 0;
 522         }
 523         catch(exception& e) {
 524                 m->errorOut(e, "IndicatorCommand", "GetIndicatorSpecies");
 525                 exit(1);
 526         }
 527 }
 528 //**********************************************************************************************************************
 529 vector<float> IndicatorCommand::getValues(vector< vector<SharedRAbundFloatVector*> >& groupings){
 530         try {
 531                 vector<float> values;
 532
 533                 //for each otu
 534                 for (int i = 0; i < groupings[0][0]->getNumBins(); i++) {
 535
 536                         if (m->control_pressed) { return values; }
 537
 538                         vector<float> terms;
 539                         float AijDenominator = 0.0;
 540                         vector<float> Bij;
 541                         //get overall abundance of each grouping
 542                         for (int j = 0; j < groupings.size(); j++) {
 543
 544                                 float totalAbund = 0;
 545                                 int numNotZero = 0;
 546                                 for (int k = 0; k < groupings[j].size(); k++) {
 547                                         totalAbund += groupings[j][k]->getAbundance(i);
 548                                         if (groupings[j][k]->getAbundance(i) != 0) { numNotZero++; }
 549                                 }
 550
 551                                 float Aij = (totalAbund / (float) groupings[j].size());
 552                                 terms.push_back(Aij);
 553
 554                                 //percentage of sites represented
 555                                 Bij.push_back(numNotZero / (float) groupings[j].size());
 556
 557                                 AijDenominator += Aij;
 558                         }
 559
 560                         float maxIndVal = 0.0;
 561                         for (int j = 0; j < terms.size(); j++) {
 562                                 float thisAij = (terms[j] / AijDenominator);
 563                                 float thisValue = thisAij * Bij[j] * 100.0;
 564
 565                                 //save largest
 566                                 if (thisValue > maxIndVal) { maxIndVal = thisValue; }
 567                         }
 568
 569                         values.push_back(maxIndVal);
 570                 }
 571
 572                 return values;
 573         }
 574         catch(exception& e) {
 575                 m->errorOut(e, "IndicatorCommand", "getValues");
 576                 exit(1);
 577         }
 578 }
 579 //**********************************************************************************************************************
 580 //same as above, just data type difference
 581 vector<float> IndicatorCommand::getValues(vector< vector<SharedRAbundVector*> >& groupings){
 582         try {
 583                 vector<float> values;
 584
 585         /*for (int j = 0; j < groupings.size(); j++) {
 586                 cout << "grouping " << j << endl;
 587                 for (int k = 0; k < groupings[j].size(); k++) {
 588                         cout << groupings[j][k]->getGroup() << endl;
 589                 }
 590         }*/
 591                 //for each otu
 592                 for (int i = 0; i < groupings[0][0]->getNumBins(); i++) {
 593                         vector<float> terms;
 594                         float AijDenominator = 0.0;
 595                         vector<float> Bij;
 596                         //get overall abundance of each grouping
 597                         for (int j = 0; j < groupings.size(); j++) {
 598
 599                                 int totalAbund = 0.0;
 600                                 int numNotZero = 0;
 601                                 for (int k = 0; k < groupings[j].size(); k++) {
 602                                         totalAbund += groupings[j][k]->getAbundance(i);
 603                                         if (groupings[j][k]->getAbundance(i) != 0.0) { numNotZero++; }
 604
 605                                 }
 606
 607                                 float Aij = (totalAbund / (float) groupings[j].size());
 608                                 terms.push_back(Aij);
 609
 610                                 //percentage of sites represented
 611                                 Bij.push_back(numNotZero / (float) groupings[j].size());
 612
 613                                 AijDenominator += Aij;
 614                         }
 615
 616                         float maxIndVal = 0.0;
 617                         for (int j = 0; j < terms.size(); j++) {
 618                                 float thisAij = (terms[j] / AijDenominator);
 619                                 float thisValue = thisAij * Bij[j] * 100.0;
 620
 621                                 //save largest
 622                                 if (thisValue > maxIndVal) { maxIndVal = thisValue; }
 623                         }
 624
 625                         values.push_back(maxIndVal);
 626                 }
 627
 628                 return values;
 629         }
 630         catch(exception& e) {
 631                 m->errorOut(e, "IndicatorCommand", "getValues");
 632                 exit(1);
 633         }
 634 }
 635 //**********************************************************************************************************************
 636 //you need the distances to root to decide groupings
 637 //this will also set branch lengths if the tree does not include them
 638 map<int, float> IndicatorCommand::getDistToRoot(Tree*& T){
 639         try {
 640                 map<int, float> dists;
 641
 642                 bool hasBranchLengths = false;
 643                 for (int i = 0; i < T->getNumNodes(); i++) {
 644                         if (T->tree[i].getBranchLength() > 0.0) {  hasBranchLengths = true; break; }
 645                 }
 646
 647                 //set branchlengths if needed
 648                 if (!hasBranchLengths) {
 649                         for (int i = 0; i < T->getNumNodes(); i++) {
 650
 651                                 int lc = T->tree[i].getLChild();
 652                                 int rc = T->tree[i].getRChild();
 653
 654                                 if (lc == -1) { // you are a leaf
 655                                         //if you are a leaf set you priliminary length to 1.0, this may adjust later
 656                                         T->tree[i].setBranchLength(1.0);
 657                                         dists[i] = 1.0;
 658                                 }else{ // you are an internal node
 659                                         //look at your children's length to leaf
 660                                         float ldist = dists[lc];
 661                                         float rdist = dists[rc];
 662
 663                                         float greater = ldist;
 664                                         if (rdist > greater) { greater = rdist; dists[i] = ldist + 1.0;}
 665                                         else { dists[i] = rdist + 1.0; }
 666
 667
 668                                         //branch length = difference + 1
 669                                         T->tree[lc].setBranchLength((abs(ldist-greater) + 1.0));
 670                                         T->tree[rc].setBranchLength((abs(rdist-greater) + 1.0));
 671                                 }
 672                         }
 673                 }
 674
 675                 dists.clear();
 676
 677                 for (int i = 0; i < T->getNumNodes(); i++) {
 678
 679                         double sum = 0.0;
 680                         int index = i;
 681
 682                         while(T->tree[index].getParent() != -1){
 683                                 if (T->tree[index].getBranchLength() != -1) {
 684                                         sum += abs(T->tree[index].getBranchLength());
 685                                 }
 686                                 index = T->tree[index].getParent();
 687                         }
 688
 689                         dists[i] = sum;
 690                 }
 691
 692                 return dists;
 693         }
 694         catch(exception& e) {
 695                 m->errorOut(e, "IndicatorCommand", "getLengthToLeaf");
 696                 exit(1);
 697         }
 698 }
 699 //**********************************************************************************************************************
 700 set<string> IndicatorCommand::getDescendantList(Tree*& T, int i, map<int, set<string> > descendants, map<int, set<int> >& nodes){
 701         try {
 702                 set<string> names;
 703
 704                 set<string>::iterator it;
 705
 706                 int lc = T->tree[i].getLChild();
 707                 int rc = T->tree[i].getRChild();
 708
 709                 if (lc == -1) { //you are a leaf your only descendant is yourself
 710                         set<int> temp; temp.insert(i);
 711                         nodes[i] = temp;
 712
 713                         if (designfile == "") {
 714                                 names.insert(T->tree[i].getName());
 715                         }else {
 716                                 vector<string> myGroup; myGroup.push_back(T->tree[i].getName());
 717                                 vector<string> myReps = designMap->getNamesSeqs(myGroup);
 718                                 for (int k = 0; k < myReps.size(); k++) {
 719                                         names.insert(myReps[k]);
 720                                 }
 721                         }
 722
 723                 }else{ //your descedants are the combination of your childrens descendants
 724                         names = descendants[lc];
 725                         nodes[i] = nodes[lc];
 726                         for (it = descendants[rc].begin(); it != descendants[rc].end(); it++) {
 727                                 names.insert(*it);
 728                         }
 729                         for (set<int>::iterator itNum = nodes[rc].begin(); itNum != nodes[rc].end(); itNum++) {
 730                                 nodes[i].insert(*itNum);
 731                         }
 732                         //you are your own descendant
 733                         nodes[i].insert(i);
 734                 }
 735
 736                 return names;
 737         }
 738         catch(exception& e) {
 739                 m->errorOut(e, "IndicatorCommand", "getDescendantList");
 740                 exit(1);
 741         }
 742 }
 743 //**********************************************************************************************************************
 744 int IndicatorCommand::getShared(){
 745         try {
 746                 InputData* input = new InputData(sharedfile, "sharedfile");
 747                 lookup = input->getSharedRAbundVectors();
 748                 string lastLabel = lookup[0]->getLabel();
 749
 750                 if (label == "") { label = lastLabel; delete input; return 0; }
 751
 752                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 753                 set<string> labels; labels.insert(label);
 754                 set<string> processedLabels;
 755                 set<string> userLabels = labels;
 756
 757                 //as long as you are not at the end of the file or done wih the lines you want
 758                 while((lookup[0] != NULL) && (userLabels.size() != 0)) {
 759                         if (m->control_pressed) {  delete input; return 0;  }
 760
 761                         if(labels.count(lookup[0]->getLabel()) == 1){
 762                                 processedLabels.insert(lookup[0]->getLabel());
 763                                 userLabels.erase(lookup[0]->getLabel());
 764                                 break;
 765                         }
 766
 767                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 768                                 string saveLabel = lookup[0]->getLabel();
 769
 770                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 771                                 lookup = input->getSharedRAbundVectors(lastLabel);
 772
 773                                 processedLabels.insert(lookup[0]->getLabel());
 774                                 userLabels.erase(lookup[0]->getLabel());
 775
 776                                 //restore real lastlabel to save below
 777                                 lookup[0]->setLabel(saveLabel);
 778                                 break;
 779                         }
 780
 781                         lastLabel = lookup[0]->getLabel();
 782
 783                         //get next line to process
 784                         //prevent memory leak
 785                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 786                         lookup = input->getSharedRAbundVectors();
 787                 }
 788
 789
 790                 if (m->control_pressed) { delete input; return 0;  }
 791
 792                 //output error messages about any remaining user labels
 793                 set<string>::iterator it;
 794                 bool needToRun = false;
 795                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 796                         m->mothurOut("Your file does not include the label " + *it);
 797                         if (processedLabels.count(lastLabel) != 1) {
 798                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 799                                 needToRun = true;
 800                         }else {
 801                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 802                         }
 803                 }
 804
 805                 //run last label if you need to
 806                 if (needToRun == true)  {
 807                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       } }
 808                         lookup = input->getSharedRAbundVectors(lastLabel);
 809                 }
 810
 811                 delete input;
 812                 return 0;
 813         }
 814         catch(exception& e) {
 815                 m->errorOut(e, "IndicatorCommand", "getShared");
 816                 exit(1);
 817         }
 818 }
 819 //**********************************************************************************************************************
 820 int IndicatorCommand::getSharedFloat(){
 821         try {
 822                 InputData* input = new InputData(relabundfile, "relabund");
 823                 lookupFloat = input->getSharedRAbundFloatVectors();
 824                 string lastLabel = lookupFloat[0]->getLabel();
 825
 826                 if (label == "") { label = lastLabel; delete input; return 0; }
 827
 828                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 829                 set<string> labels; labels.insert(label);
 830                 set<string> processedLabels;
 831                 set<string> userLabels = labels;
 832
 833                 //as long as you are not at the end of the file or done wih the lines you want
 834                 while((lookupFloat[0] != NULL) && (userLabels.size() != 0)) {
 835
 836                         if (m->control_pressed) {  delete input; return 0;  }
 837
 838                         if(labels.count(lookupFloat[0]->getLabel()) == 1){
 839                                 processedLabels.insert(lookupFloat[0]->getLabel());
 840                                 userLabels.erase(lookupFloat[0]->getLabel());
 841                                 break;
 842                         }
 843
 844                         if ((m->anyLabelsToProcess(lookupFloat[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 845                                 string saveLabel = lookupFloat[0]->getLabel();
 846
 847                                 for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  }
 848                                 lookupFloat = input->getSharedRAbundFloatVectors(lastLabel);
 849
 850                                 processedLabels.insert(lookupFloat[0]->getLabel());
 851                                 userLabels.erase(lookupFloat[0]->getLabel());
 852
 853                                 //restore real lastlabel to save below
 854                                 lookupFloat[0]->setLabel(saveLabel);
 855                                 break;
 856                         }
 857
 858                         lastLabel = lookupFloat[0]->getLabel();
 859
 860                         //get next line to process
 861                         //prevent memory leak
 862                         for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  }
 863                         lookupFloat = input->getSharedRAbundFloatVectors();
 864                 }
 865
 866
 867                 if (m->control_pressed) { delete input; return 0;  }
 868
 869                 //output error messages about any remaining user labels
 870                 set<string>::iterator it;
 871                 bool needToRun = false;
 872                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 873                         m->mothurOut("Your file does not include the label " + *it);
 874                         if (processedLabels.count(lastLabel) != 1) {
 875                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 876                                 needToRun = true;
 877                         }else {
 878                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 879                         }
 880                 }
 881
 882                 //run last label if you need to
 883                 if (needToRun == true)  {
 884                         for (int i = 0; i < lookupFloat.size(); i++) {  if (lookupFloat[i] != NULL) {   delete lookupFloat[i];  } }
 885                         lookupFloat = input->getSharedRAbundFloatVectors(lastLabel);
 886                 }
 887
 888                 delete input;
 889                 return 0;
 890         }
 891         catch(exception& e) {
 892                 m->errorOut(e, "IndicatorCommand", "getShared");
 893         exit(1);
 894         }
 895 }
 896 /*****************************************************************/
 897
 898