removerarecommand.cpp

   1 /*
   2  *  removerarecommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 1/21/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removerarecommand.h"
  11 #include "sequence.hpp"
  12 #include "groupmap.h"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> RemoveRareCommand::getValidParameters(){
  18         try {
  19                 string Array[] =  {"rabund","sabund", "group", "list", "shared","bygroup","nseqs","groups","label","outputdir","inputdir"};
  20                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  21                 return myArray;
  22         }
  23         catch(exception& e) {
  24                 m->errorOut(e, "RemoveRareCommand", "getValidParameters");
  25                 exit(1);
  26         }
  27 }
  28 //**********************************************************************************************************************
  29 RemoveRareCommand::RemoveRareCommand(){
  30         try {
  31                 abort = true;
  32                 //initialize outputTypes
  33                 vector<string> tempOutNames;
  34                 outputTypes["rabund"] = tempOutNames;
  35                 outputTypes["sabund"] = tempOutNames;
  36                 outputTypes["list"] = tempOutNames;
  37                 outputTypes["group"] = tempOutNames;
  38                 outputTypes["shared"] = tempOutNames;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 vector<string> RemoveRareCommand::getRequiredParameters(){
  47         try {
  48                 string Array[] =  {"nseqs"};
  49                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "RemoveRareCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> RemoveRareCommand::getRequiredFiles(){
  59         try {
  60                 vector<string> myArray;
  61                 return myArray;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "RemoveRareCommand", "getRequiredFiles");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 RemoveRareCommand::RemoveRareCommand(string option)  {
  70         try {
  71                 globaldata = GlobalData::getInstance();
  72                 abort = false;
  73                 allLines = 1;
  74
  75                 //allow user to run help
  76                 if(option == "help") { help(); abort = true; }
  77
  78                 else {
  79                         //valid paramters for this command
  80                         string Array[] =  {"rabund","sabund", "group", "list", "shared", "bygroup", "nseqs","groups","label","outputdir","inputdir"};
  81                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  82
  83                         OptionParser parser(option);
  84                         map<string,string> parameters = parser.getParameters();
  85
  86                         ValidParameters validParameter;
  87                         map<string,string>::iterator it;
  88
  89                         //check to make sure all parameters are valid for command
  90                         for (it = parameters.begin(); it != parameters.end(); it++) {
  91                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  92                         }
  93
  94                         //initialize outputTypes
  95                         vector<string> tempOutNames;
  96                         outputTypes["rabund"] = tempOutNames;
  97                         outputTypes["sabund"] = tempOutNames;
  98                         outputTypes["list"] = tempOutNames;
  99                         outputTypes["group"] = tempOutNames;
 100                         outputTypes["shared"] = tempOutNames;
 101
 102                         //if the user changes the output directory command factory will send this info to us in the output parameter
 103                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("list");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 116                                 }
 117
 118                                 it = parameters.find("group");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 124                                 }
 125
 126                                 it = parameters.find("sabund");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 132                                 }
 133
 134                                 it = parameters.find("rabund");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 140                                 }
 141
 142                                 it = parameters.find("shared");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 148                                 }
 149                         }
 150
 151
 152                         //check for file parameters
 153                         listfile = validParameter.validFile(parameters, "list", true);
 154                         if (listfile == "not open") { abort = true; }
 155                         else if (listfile == "not found") {  listfile = "";  }
 156
 157                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 158                         if (sabundfile == "not open") { abort = true; }
 159                         else if (sabundfile == "not found") {  sabundfile = "";  }
 160
 161                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 162                         if (rabundfile == "not open") { abort = true; }
 163                         else if (rabundfile == "not found") {  rabundfile = "";  }
 164
 165                         groupfile = validParameter.validFile(parameters, "group", true);
 166                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 167                         else if (groupfile == "not found") {  groupfile = "";  }
 168
 169                         sharedfile = validParameter.validFile(parameters, "shared", true);
 170                         if (sharedfile == "not open") { sharedfile = "";  abort = true; }
 171                         else if (sharedfile == "not found") {  sharedfile = "";  }
 172
 173                         if ((sabundfile == "") && (rabundfile == "")  && (sharedfile == "") && (listfile == ""))  { m->mothurOut("You must provide at least one of the following: rabund, sabund, shared or list."); m->mothurOutEndLine(); abort = true; }
 174
 175                         groups = validParameter.validFile(parameters, "groups", false);
 176                         if (groups == "not found") { groups = "all"; }
 177                         m->splitAtDash(groups, Groups);
 178
 179                         label = validParameter.validFile(parameters, "label", false);
 180                         if (label == "not found") { label = ""; }
 181                         else {
 182                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 183                                 else { allLines = 1;  }
 184                         }
 185
 186                         string temp = validParameter.validFile(parameters, "nseqs", false);
 187                         if (temp == "not found") { m->mothurOut("nseqs is a required parameter."); m->mothurOutEndLine(); abort = true; }
 188                         else { convert(temp, nseqs); }
 189
 190                         temp = validParameter.validFile(parameters, "bygroup", false);   if (temp == "not found") { temp = "f"; }
 191                         byGroup = m->isTrue(temp);
 192
 193                         if (byGroup && (sharedfile == "")) { m->mothurOut("The byGroup parameter is only valid with a shared file."); m->mothurOutEndLine(); }
 194
 195                         if ((groupfile != "") && (listfile == "")) { m->mothurOut("A groupfile is only valid with a list file."); m->mothurOutEndLine(); groupfile = ""; }
 196                 }
 197
 198         }
 199         catch(exception& e) {
 200                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
 201                 exit(1);
 202         }
 203 }
 204 //**********************************************************************************************************************
 205
 206 void RemoveRareCommand::help(){
 207         try {
 208                 m->mothurOut("The remove.rare command parameters are list, rabund, sabund, shared, group, label, groups, bygroup and nseqs.\n");
 209                 m->mothurOut("The remove.rare command reads one of the following file types: list, rabund, sabund or shared file. It outputs a new file after removing the rare otus.\n");
 210                 m->mothurOut("The groups parameter allows you to specify which of the groups you would like analyzed.  Default=all. You may separate group names with dashes.\n");
 211                 m->mothurOut("The label parameter is used to analyze specific labels in your input. default=all. You may separate label names with dashes.\n");
 212                 m->mothurOut("The bygroup parameter is only valid with the shared file. default=f, meaning remove any OTU that has nseqs or fewer sequences across all groups.\n");
 213                 m->mothurOut("bygroups=T means remove any OTU that has nseqs or fewer sequences in each group (if groupA has 1 sequence and group B has 100 sequences in OTUZ and nseqs=1, then set the groupA count for OTUZ to 0 and keep groupB's count at 100.) \n");
 214                 m->mothurOut("The nseqs parameter allows you to set the cutoff for an otu to be deemed rare. It is required.\n");
 215                 m->mothurOut("The remove.rare command should be in the following format: remove.rare(shared=yourSharedFile, nseqs=yourRareCutoff).\n");
 216                 m->mothurOut("Example remove.rare(shared=amazon.fn.shared, nseqs=2).\n");
 217                 m->mothurOut("Note: No spaces between parameter labels (i.e. shared), '=' and parameters (i.e.yourSharedFile).\n\n");
 218         }
 219         catch(exception& e) {
 220                 m->errorOut(e, "RemoveRareCommand", "help");
 221                 exit(1);
 222         }
 223 }
 224
 225 //**********************************************************************************************************************
 226
 227 int RemoveRareCommand::execute(){
 228         try {
 229
 230                 if (abort == true) { return 0; }
 231
 232                 if (m->control_pressed) { return 0; }
 233
 234                 //read through the correct file and output lines you want to keep
 235                 if (sabundfile != "")           {               processSabund();        }
 236                 if (rabundfile != "")           {               processRabund();        }
 237                 if (listfile != "")                     {               processList();          }
 238                 if (sharedfile != "")           {               processShared();        }
 239
 240                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 241
 242                 if (outputNames.size() != 0) {
 243                         m->mothurOutEndLine();
 244                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 245                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 246                         m->mothurOutEndLine();
 247                 }
 248
 249                 return 0;
 250         }
 251
 252         catch(exception& e) {
 253                 m->errorOut(e, "RemoveRareCommand", "execute");
 254                 exit(1);
 255         }
 256 }
 257
 258 //**********************************************************************************************************************
 259 int RemoveRareCommand::processList(){
 260         try {
 261                 string thisOutputDir = outputDir;
 262                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 263                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 264                 string outputGroupFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" +  m->getExtension(groupfile);
 265
 266                 ofstream out, outGroup;
 267                 m->openOutputFile(outputFileName, out);
 268
 269                 bool wroteSomething = false;
 270
 271                 //you must provide a label because the names in the listfile need to be consistent
 272                 string thisLabel = "";
 273                 if (allLines) { m->mothurOut("For the listfile you must select one label, using first label in your listfile."); m->mothurOutEndLine(); }
 274                 else if (labels.size() > 1) { m->mothurOut("For the listfile you must select one label, using " + (*labels.begin()) + "."); m->mothurOutEndLine(); thisLabel = *labels.begin(); }
 275                 else { thisLabel = *labels.begin(); }
 276
 277                 InputData input(listfile, "list");
 278                 ListVector* list = input.getListVector();
 279
 280                 //get first one or the one we want
 281                 if (thisLabel != "") {
 282                         //use smart distancing
 283                         set<string> userLabels; userLabels.insert(thisLabel);
 284                         set<string> processedLabels;
 285                         string lastLabel = list->getLabel();
 286                         while((list != NULL) && (userLabels.size() != 0)) {
 287                                 if(userLabels.count(list->getLabel()) == 1){
 288                                         processedLabels.insert(list->getLabel());
 289                                         userLabels.erase(list->getLabel());
 290                                         break;
 291                                 }
 292
 293                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 294                                         processedLabels.insert(list->getLabel());
 295                                         userLabels.erase(list->getLabel());
 296                                         delete list;
 297                                         list = input.getListVector(lastLabel);
 298                                         break;
 299                                 }
 300                                 lastLabel = list->getLabel();
 301                                 delete list;
 302                                 list = input.getListVector();
 303                         }
 304                         if (userLabels.size() != 0) {
 305                                 m->mothurOut("Your file does not include the label " + thisLabel + ". I will use " + lastLabel + ".");  m->mothurOutEndLine();
 306                                 list = input.getListVector(lastLabel);
 307                         }
 308                 }
 309
 310                 //if groupfile is given then use it
 311                 GroupMap* groupMap;
 312                 if (groupfile != "") {
 313                         groupMap = new GroupMap(groupfile); groupMap->readMap();
 314                         SharedUtil util;
 315                         util.setGroups(Groups, groupMap->namesOfGroups);
 316                         m->openOutputFile(outputGroupFileName, outGroup);
 317                 }
 318
 319
 320                 if (list != NULL) {
 321                         //make a new list vector
 322                         ListVector newList;
 323                         newList.setLabel(list->getLabel());
 324
 325                         //for each bin
 326                         for (int i = 0; i < list->getNumBins(); i++) {
 327                                 if (m->control_pressed) {  if (groupfile != "") { delete groupMap; outGroup.close(); remove(outputGroupFileName.c_str()); } out.close();  remove(outputFileName.c_str());  return 0; }
 328
 329                                 //parse out names that are in accnos file
 330                                 string binnames = list->get(i);
 331                                 vector<string> names;
 332                                 string saveBinNames = binnames;
 333                                 m->splitAtComma(binnames, names);
 334
 335                                 vector<string> newGroupFile;
 336                                 if (groupfile != "") {
 337                                         vector<string> newNames;
 338                                         saveBinNames = "";
 339                                         for(int k = 0; k < names.size(); k++) {
 340                                                 string group = groupMap->getGroup(names[k]);
 341
 342                                                 if (m->inUsersGroups(group, Groups)) {
 343                                                         newGroupFile.push_back(names[k] + "\t" + group);
 344
 345                                                         newNames.push_back(names[k]);
 346                                                         saveBinNames += names[k] + ",";
 347                                                 }
 348                                         }
 349                                         names = newNames;
 350                                         saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
 351                                 }
 352
 353                                 if (names.size() > nseqs) { //keep bin
 354                                         newList.push_back(saveBinNames);
 355                                         for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }
 356                                 }
 357                         }
 358
 359                         //print new listvector
 360                         if (newList.getNumBins() != 0) {
 361                                 wroteSomething = true;
 362                                 newList.print(out);
 363                         }
 364                 }
 365
 366                 out.close();
 367                 if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
 368
 369                 if (wroteSomething == false) {  m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine();  }
 370                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 371
 372                 return 0;
 373         }
 374         catch(exception& e) {
 375                 m->errorOut(e, "RemoveRareCommand", "processList");
 376                 exit(1);
 377         }
 378 }
 379 //**********************************************************************************************************************
 380 int RemoveRareCommand::processSabund(){
 381         try {
 382                 string thisOutputDir = outputDir;
 383                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
 384                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "pick" +  m->getExtension(sabundfile);
 385                 outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 386
 387                 ofstream out;
 388                 m->openOutputFile(outputFileName, out);
 389
 390                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 391                 InputData input(sabundfile, "sabund");
 392                 SAbundVector* sabund = input.getSAbundVector();
 393                 string lastLabel = sabund->getLabel();
 394                 set<string> processedLabels;
 395                 set<string> userLabels = labels;
 396
 397                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 398
 399                         if (m->control_pressed) { delete sabund; out.close(); return 0; }
 400
 401                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 402
 403                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 404                                 processedLabels.insert(sabund->getLabel());
 405                                 userLabels.erase(sabund->getLabel());
 406
 407                                 if (sabund->getMaxRank() > nseqs) {
 408                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 409                                 }else { sabund->clear(); }
 410
 411                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 412                         }
 413
 414                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 415                                 string saveLabel = sabund->getLabel();
 416
 417                                 delete sabund;
 418                                 sabund = input.getSAbundVector(lastLabel);
 419
 420                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 421                                 processedLabels.insert(sabund->getLabel());
 422                                 userLabels.erase(sabund->getLabel());
 423
 424                                 if (sabund->getMaxRank() > nseqs) {
 425                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 426                                 }else { sabund->clear(); }
 427
 428                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 429
 430                                 //restore real lastlabel to save below
 431                                 sabund->setLabel(saveLabel);
 432                         }
 433
 434                         lastLabel = sabund->getLabel();
 435
 436                         delete sabund;
 437                         sabund = input.getSAbundVector();
 438                 }
 439
 440                 if (m->control_pressed) {  out.close(); return 0; }
 441
 442                 //output error messages about any remaining user labels
 443                 set<string>::iterator it;
 444                 bool needToRun = false;
 445                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 446                         m->mothurOut("Your file does not include the label " + *it);
 447                         if (processedLabels.count(lastLabel) != 1) {
 448                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 449                                 needToRun = true;
 450                         }else {
 451                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 452                         }
 453                 }
 454
 455                 //run last label if you need to
 456                 if (needToRun == true)  {
 457                         if (sabund != NULL) {   delete sabund;  }
 458                         sabund = input.getSAbundVector(lastLabel);
 459
 460                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 461
 462                         if (sabund->getMaxRank() > nseqs) {
 463                                 for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 464                         }else { sabund->clear(); }
 465
 466                         if (sabund->getNumBins() > 0) { sabund->print(out); }
 467
 468                         delete sabund;
 469                 }
 470
 471                 return 0;
 472         }
 473         catch(exception& e) {
 474                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 475                 exit(1);
 476         }
 477 }
 478 //**********************************************************************************************************************
 479 int RemoveRareCommand::processRabund(){
 480         try {
 481                 string thisOutputDir = outputDir;
 482                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
 483                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "pick" +  m->getExtension(rabundfile);
 484                 outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 485
 486                 ofstream out;
 487                 m->openOutputFile(outputFileName, out);
 488
 489                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 490                 InputData input(rabundfile, "rabund");
 491                 RAbundVector* rabund = input.getRAbundVector();
 492                 string lastLabel = rabund->getLabel();
 493                 set<string> processedLabels;
 494                 set<string> userLabels = labels;
 495
 496                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 497
 498                         if (m->control_pressed) { delete rabund; out.close(); return 0; }
 499
 500                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
 501
 502                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 503                                 processedLabels.insert(rabund->getLabel());
 504                                 userLabels.erase(rabund->getLabel());
 505
 506                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 507                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 508                                         if (rabund->get(i) > nseqs) {
 509                                                 newRabund.push_back(rabund->get(i));
 510                                         }
 511                                 }
 512                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 513                         }
 514
 515                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 516                                 string saveLabel = rabund->getLabel();
 517
 518                                 delete rabund;
 519                                 rabund = input.getRAbundVector(lastLabel);
 520
 521                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 522                                 processedLabels.insert(rabund->getLabel());
 523                                 userLabels.erase(rabund->getLabel());
 524
 525                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 526                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 527                                         if (rabund->get(i) > nseqs) {
 528                                                 newRabund.push_back(rabund->get(i));
 529                                         }
 530                                 }
 531                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 532
 533                                 //restore real lastlabel to save below
 534                                 rabund->setLabel(saveLabel);
 535                         }
 536
 537                         lastLabel = rabund->getLabel();
 538
 539                         delete rabund;
 540                         rabund = input.getRAbundVector();
 541                 }
 542
 543                 if (m->control_pressed) {  out.close(); return 0; }
 544
 545                 //output error messages about any remaining user labels
 546                 set<string>::iterator it;
 547                 bool needToRun = false;
 548                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 549                         m->mothurOut("Your file does not include the label " + *it);
 550                         if (processedLabels.count(lastLabel) != 1) {
 551                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 552                                 needToRun = true;
 553                         }else {
 554                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 555                         }
 556                 }
 557
 558                 //run last label if you need to
 559                 if (needToRun == true)  {
 560                         if (rabund != NULL) {   delete rabund;  }
 561                         rabund = input.getRAbundVector(lastLabel);
 562
 563                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 564
 565                         RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 566                         for (int i = 0; i < rabund->getNumBins(); i++) {
 567                                 if (rabund->get(i) > nseqs) {
 568                                         newRabund.push_back(rabund->get(i));
 569                                 }
 570                         }
 571                         if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 572
 573                         delete rabund;
 574                 }
 575
 576                 return 0;
 577         }
 578         catch(exception& e) {
 579                 m->errorOut(e, "RemoveRareCommand", "processRabund");
 580                 exit(1);
 581         }
 582 }
 583 //**********************************************************************************************************************
 584 int RemoveRareCommand::processShared(){
 585         try {
 586                 globaldata->Groups = Groups;
 587
 588                 string thisOutputDir = outputDir;
 589                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 590                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "pick" +  m->getExtension(sharedfile);
 591                 outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
 592
 593                 ofstream out;
 594                 m->openOutputFile(outputFileName, out);
 595
 596                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 597                 InputData input(sharedfile, "sharedfile");
 598                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 599                 string lastLabel = lookup[0]->getLabel();
 600                 set<string> processedLabels;
 601                 set<string> userLabels = labels;
 602
 603                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 604
 605                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  out.close(); return 0; }
 606
 607                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 608
 609                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 610                                 processedLabels.insert(lookup[0]->getLabel());
 611                                 userLabels.erase(lookup[0]->getLabel());
 612
 613                                 processLookup(lookup, out);
 614                         }
 615
 616                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 617                                 string saveLabel = lookup[0]->getLabel();
 618
 619                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 620                                 lookup = input.getSharedRAbundVectors(lastLabel);
 621
 622                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 623                                 processedLabels.insert(lookup[0]->getLabel());
 624                                 userLabels.erase(lookup[0]->getLabel());
 625
 626                                 processLookup(lookup, out);
 627
 628                                 //restore real lastlabel to save below
 629                                 lookup[0]->setLabel(saveLabel);
 630                         }
 631
 632                         lastLabel = lookup[0]->getLabel();
 633
 634                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 635                         lookup = input.getSharedRAbundVectors();
 636                 }
 637
 638                 if (m->control_pressed) {  out.close(); return 0; }
 639
 640                 //output error messages about any remaining user labels
 641                 set<string>::iterator it;
 642                 bool needToRun = false;
 643                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 644                         m->mothurOut("Your file does not include the label " + *it);
 645                         if (processedLabels.count(lastLabel) != 1) {
 646                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 647                                 needToRun = true;
 648                         }else {
 649                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 650                         }
 651                 }
 652
 653                 //run last label if you need to
 654                 if (needToRun == true)  {
 655                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       }  }
 656                         lookup = input.getSharedRAbundVectors(lastLabel);
 657
 658                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 659
 660                         processLookup(lookup, out);
 661
 662                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 663                 }
 664
 665                 return 0;
 666         }
 667         catch(exception& e) {
 668                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 669                 exit(1);
 670         }
 671 }
 672 //**********************************************************************************************************************
 673 int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup, ofstream& out){
 674         try {
 675
 676                 vector<SharedRAbundVector> newRabunds;  newRabunds.resize(lookup.size());
 677                 for (int i = 0; i < lookup.size(); i++) {
 678                         newRabunds[i].setGroup(lookup[i]->getGroup());
 679                         newRabunds[i].setLabel(lookup[i]->getLabel());
 680                 }
 681
 682                 if (byGroup) {
 683
 684                         //for each otu
 685                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 686                                 bool allZero = true;
 687
 688                                 if (m->control_pressed) { return 0; }
 689
 690                                 //for each group
 691                                 for (int j = 0; j < lookup.size(); j++) {
 692
 693                                         //are you rare?
 694                                         if (lookup[j]->getAbundance(i) > nseqs) {
 695                                                 newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 696                                                 allZero = false;
 697                                         }else {
 698                                                 newRabunds[j].push_back(0, newRabunds[j].getGroup());
 699                                         }
 700                                 }
 701
 702                                 //eliminates zero otus
 703                                 if (allZero) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 704                         }
 705                 }else {
 706                         //for each otu
 707                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 708
 709                                 if (m->control_pressed) { return 0; }
 710
 711                                 int totalAbund = 0;
 712                                 //get total otu abundance
 713                                 for (int j = 0; j < lookup.size(); j++) {
 714                                         newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 715                                         totalAbund += lookup[j]->getAbundance(i);
 716                                 }
 717
 718                                 //eliminates otus below rare cutoff
 719                                 if (totalAbund <= nseqs) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 720                         }
 721                 }
 722
 723                 //do we have any otus above the rare cutoff
 724                 if (newRabunds[0].getNumBins() != 0) {
 725                         for (int j = 0; j < newRabunds.size(); j++) {
 726                                 out << newRabunds[j].getLabel() << '\t' << newRabunds[j].getGroup() << '\t';
 727                                 newRabunds[j].print(out);
 728                         }
 729                 }
 730
 731                 return 0;
 732         }
 733         catch(exception& e) {
 734                 m->errorOut(e, "RemoveRareCommand", "processLookup");
 735                 exit(1);
 736         }
 737 }
 738 //**********************************************************************************************************************
 739
 740
 741
 742