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::setParameters(){
  18         try {
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(plist);
  20                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(prabund);
  21                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(psabund);
  22                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pshared);
  23         CommandParameter pcount("count", "InputTypes", "", "", "CountGroup", "none", "none",false,false); parameters.push_back(pcount);
  24                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none",false,false); parameters.push_back(pgroup);
  25                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  26                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  27                 CommandParameter pnseqs("nseqs", "Number", "", "0", "", "", "",false,true); parameters.push_back(pnseqs);
  28                 CommandParameter pbygroup("bygroup", "Boolean", "", "f", "", "", "",false,true); parameters.push_back(pbygroup);
  29                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  30                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  31
  32                 vector<string> myArray;
  33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "RemoveRareCommand", "setParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 string RemoveRareCommand::getHelpString(){
  43         try {
  44                 string helpString = "";
  45                 helpString += "The remove.rare command parameters are list, rabund, sabund, shared, group, count, label, groups, bygroup and nseqs.\n";
  46                 helpString += "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";
  47                 helpString += "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";
  48                 helpString += "The label parameter is used to analyze specific labels in your input. default=all. You may separate label names with dashes.\n";
  49                 helpString += "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";
  50                 helpString += "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";
  51                 helpString += "The nseqs parameter allows you to set the cutoff for an otu to be deemed rare. It is required.\n";
  52                 helpString += "The remove.rare command should be in the following format: remove.rare(shared=yourSharedFile, nseqs=yourRareCutoff).\n";
  53                 helpString += "Example remove.rare(shared=amazon.fn.shared, nseqs=2).\n";
  54                 helpString += "Note: No spaces between parameter labels (i.e. shared), '=' and parameters (i.e.yourSharedFile).\n";
  55                 return helpString;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "RemoveRareCommand", "getHelpString");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63 string RemoveRareCommand::getOutputFileNameTag(string type, string inputName=""){
  64         try {
  65         string outputFileName = "";
  66                 map<string, vector<string> >::iterator it;
  67
  68         //is this a type this command creates
  69         it = outputTypes.find(type);
  70         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  71         else {
  72             if (type == "rabund")            {   outputFileName =  "pick" + m->getExtension(inputName);   }
  73             else if (type == "sabund")    {   outputFileName =  "pick" + m->getExtension(inputName);   }
  74             else if (type == "shared")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  75             else if (type == "group")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  76             else if (type == "count")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  77             else if (type == "list")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  78             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  79         }
  80         return outputFileName;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "RemoveRareCommand", "getOutputFileNameTag");
  84                 exit(1);
  85         }
  86 }
  87
  88 //**********************************************************************************************************************
  89 RemoveRareCommand::RemoveRareCommand(){
  90         try {
  91                 abort = true; calledHelp = true;
  92                 setParameters();
  93                 vector<string> tempOutNames;
  94                 outputTypes["rabund"] = tempOutNames;
  95                 outputTypes["sabund"] = tempOutNames;
  96                 outputTypes["list"] = tempOutNames;
  97                 outputTypes["group"] = tempOutNames;
  98         outputTypes["count"] = tempOutNames;
  99                 outputTypes["shared"] = tempOutNames;
 100         }
 101         catch(exception& e) {
 102                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
 103                 exit(1);
 104         }
 105 }
 106 //**********************************************************************************************************************
 107 RemoveRareCommand::RemoveRareCommand(string option)  {
 108         try {
 109                 abort = false; calledHelp = false;
 110                 allLines = 1;
 111
 112                 //allow user to run help
 113                 if(option == "help") { help(); abort = true; calledHelp = true; }
 114                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 115
 116                 else {
 117                         vector<string> myArray = setParameters();
 118
 119                         OptionParser parser(option);
 120                         map<string,string> parameters = parser.getParameters();
 121
 122                         ValidParameters validParameter;
 123                         map<string,string>::iterator it;
 124
 125                         //check to make sure all parameters are valid for command
 126                         for (it = parameters.begin(); it != parameters.end(); it++) {
 127                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 128                         }
 129
 130                         //initialize outputTypes
 131                         vector<string> tempOutNames;
 132                         outputTypes["rabund"] = tempOutNames;
 133                         outputTypes["sabund"] = tempOutNames;
 134                         outputTypes["list"] = tempOutNames;
 135                         outputTypes["group"] = tempOutNames;
 136                         outputTypes["shared"] = tempOutNames;
 137             outputTypes["count"] = tempOutNames;
 138
 139                         //if the user changes the output directory command factory will send this info to us in the output parameter
 140                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 141
 142                         //if the user changes the input directory command factory will send this info to us in the output parameter
 143                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 144                         if (inputDir == "not found"){   inputDir = "";          }
 145                         else {
 146                                 string path;
 147                                 it = parameters.find("list");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 153                                 }
 154
 155                                 it = parameters.find("group");
 156                                 //user has given a template file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 161                                 }
 162
 163                                 it = parameters.find("sabund");
 164                                 //user has given a template file
 165                                 if(it != parameters.end()){
 166                                         path = m->hasPath(it->second);
 167                                         //if the user has not given a path then, add inputdir. else leave path alone.
 168                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 169                                 }
 170
 171                                 it = parameters.find("rabund");
 172                                 //user has given a template file
 173                                 if(it != parameters.end()){
 174                                         path = m->hasPath(it->second);
 175                                         //if the user has not given a path then, add inputdir. else leave path alone.
 176                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 177                                 }
 178
 179                                 it = parameters.find("shared");
 180                                 //user has given a template file
 181                                 if(it != parameters.end()){
 182                                         path = m->hasPath(it->second);
 183                                         //if the user has not given a path then, add inputdir. else leave path alone.
 184                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 185                                 }
 186
 187                 it = parameters.find("count");
 188                                 //user has given a template file
 189                                 if(it != parameters.end()){
 190                                         path = m->hasPath(it->second);
 191                                         //if the user has not given a path then, add inputdir. else leave path alone.
 192                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 193                                 }
 194                         }
 195
 196
 197                         //check for file parameters
 198                         listfile = validParameter.validFile(parameters, "list", true);
 199                         if (listfile == "not open") { abort = true; }
 200                         else if (listfile == "not found") {  listfile = "";  }
 201                         else { m->setListFile(listfile); }
 202
 203                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 204                         if (sabundfile == "not open") { abort = true; }
 205                         else if (sabundfile == "not found") {  sabundfile = "";  }
 206                         else { m->setSabundFile(sabundfile); }
 207
 208                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 209                         if (rabundfile == "not open") { abort = true; }
 210                         else if (rabundfile == "not found") {  rabundfile = "";  }
 211                         else { m->setRabundFile(rabundfile); }
 212
 213                         groupfile = validParameter.validFile(parameters, "group", true);
 214                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 215                         else if (groupfile == "not found") {  groupfile = "";  }
 216                         else { m->setGroupFile(groupfile); }
 217
 218                         sharedfile = validParameter.validFile(parameters, "shared", true);
 219                         if (sharedfile == "not open") { sharedfile = "";  abort = true; }
 220                         else if (sharedfile == "not found") {  sharedfile = "";  }
 221                         else { m->setSharedFile(sharedfile); }
 222
 223             countfile = validParameter.validFile(parameters, "count", true);
 224                         if (countfile == "not open") { countfile = ""; abort = true; }
 225                         else if (countfile == "not found") { countfile = "";  }
 226                         else { m->setCountTableFile(countfile); }
 227
 228             if ((groupfile != "") && (countfile != "")) {
 229                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 230             }
 231
 232                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 233                                 //is there are current file available for any of these?
 234                                 //give priority to shared, then list, then rabund, then sabund
 235                                 //if there is a current shared file, use it
 236                                 sharedfile = m->getSharedFile();
 237                                 if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 238                                 else {
 239                                         listfile = m->getListFile();
 240                                         if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 241                                         else {
 242                                                 rabundfile = m->getRabundFile();
 243                                                 if (rabundfile != "") {  m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 244                                                 else {
 245                                                         sabundfile = m->getSabundFile();
 246                                                         if (sabundfile != "") {  m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 247                                                         else {
 248                                                                 m->mothurOut("No valid current files. You must provide a list, sabund, rabund or shared file."); m->mothurOutEndLine();
 249                                                                 abort = true;
 250                                                         }
 251                                                 }
 252                                         }
 253                                 }
 254                         }
 255
 256                         groups = validParameter.validFile(parameters, "groups", false);
 257                         if (groups == "not found") { groups = "all"; }
 258                         m->splitAtDash(groups, Groups);
 259
 260                         label = validParameter.validFile(parameters, "label", false);
 261                         if (label == "not found") { label = ""; }
 262                         else {
 263                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 264                                 else { allLines = 1;  }
 265                         }
 266
 267                         string temp = validParameter.validFile(parameters, "nseqs", false);
 268                         if (temp == "not found") { m->mothurOut("nseqs is a required parameter."); m->mothurOutEndLine(); abort = true; }
 269                         else { m->mothurConvert(temp, nseqs); }
 270
 271                         temp = validParameter.validFile(parameters, "bygroup", false);   if (temp == "not found") { temp = "f"; }
 272                         byGroup = m->isTrue(temp);
 273
 274                         if (byGroup && (sharedfile == "")) { m->mothurOut("The byGroup parameter is only valid with a shared file."); m->mothurOutEndLine(); }
 275
 276                         if (((groupfile != "") || (countfile != "")) && (listfile == "")) { m->mothurOut("A group or count file is only valid with a list file."); m->mothurOutEndLine(); groupfile = ""; countfile = ""; }
 277                 }
 278
 279         }
 280         catch(exception& e) {
 281                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
 282                 exit(1);
 283         }
 284 }
 285 //**********************************************************************************************************************
 286
 287 int RemoveRareCommand::execute(){
 288         try {
 289
 290                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 291
 292                 if (m->control_pressed) { return 0; }
 293
 294                 //read through the correct file and output lines you want to keep
 295                 if (sabundfile != "")           {               processSabund();        }
 296                 if (rabundfile != "")           {               processRabund();        }
 297                 if (listfile != "")                     {               processList();          }
 298                 if (sharedfile != "")           {               processShared();        }
 299
 300                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 301
 302                 if (outputNames.size() != 0) {
 303                         m->mothurOutEndLine();
 304                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 305                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 306                         m->mothurOutEndLine();
 307
 308                         //set rabund file as new current rabundfile
 309                         string current = "";
 310                         itTypes = outputTypes.find("rabund");
 311                         if (itTypes != outputTypes.end()) {
 312                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 313                         }
 314
 315                         itTypes = outputTypes.find("sabund");
 316                         if (itTypes != outputTypes.end()) {
 317                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 318                         }
 319
 320                         itTypes = outputTypes.find("group");
 321                         if (itTypes != outputTypes.end()) {
 322                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 323                         }
 324
 325                         itTypes = outputTypes.find("list");
 326                         if (itTypes != outputTypes.end()) {
 327                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 328                         }
 329
 330                         itTypes = outputTypes.find("shared");
 331                         if (itTypes != outputTypes.end()) {
 332                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 333                         }
 334
 335             itTypes = outputTypes.find("count");
 336                         if (itTypes != outputTypes.end()) {
 337                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 338                         }
 339                 }
 340
 341                 return 0;
 342         }
 343
 344         catch(exception& e) {
 345                 m->errorOut(e, "RemoveRareCommand", "execute");
 346                 exit(1);
 347         }
 348 }
 349
 350 //**********************************************************************************************************************
 351 int RemoveRareCommand::processList(){
 352         try {
 353                 string thisOutputDir = outputDir;
 354                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 355                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + getOutputFileNameTag("list", listfile);
 356                 string outputGroupFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
 357         string outputCountFileName = thisOutputDir + m->getRootName(m->getSimpleName(countfile)) + getOutputFileNameTag("count", countfile);
 358
 359                 ofstream out, outGroup;
 360                 m->openOutputFile(outputFileName, out);
 361
 362                 bool wroteSomething = false;
 363
 364                 //you must provide a label because the names in the listfile need to be consistent
 365                 string thisLabel = "";
 366                 if (allLines) { m->mothurOut("For the listfile you must select one label, using first label in your listfile."); m->mothurOutEndLine(); }
 367                 else if (labels.size() > 1) { m->mothurOut("For the listfile you must select one label, using " + (*labels.begin()) + "."); m->mothurOutEndLine(); thisLabel = *labels.begin(); }
 368                 else { thisLabel = *labels.begin(); }
 369
 370                 InputData input(listfile, "list");
 371                 ListVector* list = input.getListVector();
 372
 373                 //get first one or the one we want
 374                 if (thisLabel != "") {
 375                         //use smart distancing
 376                         set<string> userLabels; userLabels.insert(thisLabel);
 377                         set<string> processedLabels;
 378                         string lastLabel = list->getLabel();
 379                         while((list != NULL) && (userLabels.size() != 0)) {
 380                                 if(userLabels.count(list->getLabel()) == 1){
 381                                         processedLabels.insert(list->getLabel());
 382                                         userLabels.erase(list->getLabel());
 383                                         break;
 384                                 }
 385
 386                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 387                                         processedLabels.insert(list->getLabel());
 388                                         userLabels.erase(list->getLabel());
 389                                         delete list;
 390                                         list = input.getListVector(lastLabel);
 391                                         break;
 392                                 }
 393                                 lastLabel = list->getLabel();
 394                                 delete list;
 395                                 list = input.getListVector();
 396                         }
 397                         if (userLabels.size() != 0) {
 398                                 m->mothurOut("Your file does not include the label " + thisLabel + ". I will use " + lastLabel + ".");  m->mothurOutEndLine();
 399                                 list = input.getListVector(lastLabel);
 400                         }
 401                 }
 402
 403                 //if groupfile is given then use it
 404                 GroupMap* groupMap;
 405         CountTable ct;
 406                 if (groupfile != "") {
 407                         groupMap = new GroupMap(groupfile); groupMap->readMap();
 408                         SharedUtil util;
 409                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 410                         util.setGroups(Groups, namesGroups);
 411                         m->openOutputFile(outputGroupFileName, outGroup);
 412                 }else if (countfile != "") {
 413             ct.readTable(countfile);
 414             if (ct.hasGroupInfo()) {
 415                 vector<string> namesGroups = ct.getNamesOfGroups();
 416                 SharedUtil util;
 417                 util.setGroups(Groups, namesGroups);
 418             }
 419         }
 420
 421
 422                 if (list != NULL) {
 423                         //make a new list vector
 424                         ListVector newList;
 425                         newList.setLabel(list->getLabel());
 426
 427                         //for each bin
 428                         for (int i = 0; i < list->getNumBins(); i++) {
 429                                 if (m->control_pressed) {  if (groupfile != "") { delete groupMap; outGroup.close(); m->mothurRemove(outputGroupFileName); } out.close();  m->mothurRemove(outputFileName);  return 0; }
 430
 431                                 //parse out names that are in accnos file
 432                                 string binnames = list->get(i);
 433                                 vector<string> names;
 434                                 string saveBinNames = binnames;
 435                                 m->splitAtComma(binnames, names);
 436                 int binsize = names.size();
 437
 438                                 vector<string> newGroupFile;
 439                                 if (groupfile != "") {
 440                                         vector<string> newNames;
 441                                         saveBinNames = "";
 442                                         for(int k = 0; k < names.size(); k++) {
 443                                                 string group = groupMap->getGroup(names[k]);
 444
 445                                                 if (m->inUsersGroups(group, Groups)) {
 446                                                         newGroupFile.push_back(names[k] + "\t" + group);
 447
 448                                                         newNames.push_back(names[k]);
 449                                                         saveBinNames += names[k] + ",";
 450                                                 }
 451                                         }
 452                                         names = newNames; binsize = names.size();
 453                                         saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
 454                                 }else if (countfile != "") {
 455                                         saveBinNames = "";
 456                     binsize = 0;
 457                                         for(int k = 0; k < names.size(); k++) {
 458                         if (ct.hasGroupInfo()) {
 459                             vector<string> thisSeqsGroups = ct.getGroups(names[k]);
 460
 461                             int thisSeqsCount = 0;
 462                             for (int n = 0; n < thisSeqsGroups.size(); n++) {
 463                                 if (m->inUsersGroups(thisSeqsGroups[n], Groups)) {
 464                                     thisSeqsCount += ct.getGroupCount(names[k], thisSeqsGroups[n]);
 465                                 }
 466                             }
 467                             binsize += thisSeqsCount;
 468                             //if you don't have any seqs from the groups the user wants, then remove you.
 469                             if (thisSeqsCount == 0) { newGroupFile.push_back(names[k]); }
 470                             else { saveBinNames += names[k] + ","; }
 471                         }else {
 472                             binsize += ct.getNumSeqs(names[k]);
 473                             saveBinNames += names[k] + ",";
 474                         }
 475                                         }
 476                                         saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
 477                 }
 478
 479                                 if (binsize > nseqs) { //keep bin
 480                                         newList.push_back(saveBinNames);
 481                                         if (groupfile != "") {  for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }  }
 482                     else if (countfile != "") { for(int k = 0; k < newGroupFile.size(); k++) {  ct.remove(newGroupFile[k]); } }
 483                                 }else {  if (countfile != "") {  for(int k = 0; k < names.size(); k++) {  ct.remove(names[k]); } }  }
 484                         }
 485
 486                         //print new listvector
 487                         if (newList.getNumBins() != 0) {
 488                                 wroteSomething = true;
 489                                 newList.print(out);
 490                         }
 491                 }
 492
 493                 out.close();
 494                 if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
 495         if (countfile != "") {
 496             if (ct.hasGroupInfo()) {
 497                 vector<string> allGroups = ct.getNamesOfGroups();
 498                 for (int i = 0; i < allGroups.size(); i++) {
 499                     if (!m->inUsersGroups(allGroups[i], Groups)) { ct.removeGroup(allGroups[i]); }
 500                 }
 501
 502             }
 503             ct.printTable(outputCountFileName);
 504             outputTypes["count"].push_back(outputCountFileName); outputNames.push_back(outputCountFileName);
 505         }
 506
 507                 if (wroteSomething == false) {  m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine();  }
 508                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 509
 510                 return 0;
 511         }
 512         catch(exception& e) {
 513                 m->errorOut(e, "RemoveRareCommand", "processList");
 514                 exit(1);
 515         }
 516 }
 517 //**********************************************************************************************************************
 518 int RemoveRareCommand::processSabund(){
 519         try {
 520                 string thisOutputDir = outputDir;
 521                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
 522                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + getOutputFileNameTag("sabund", sabundfile);
 523                 outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 524
 525                 ofstream out;
 526                 m->openOutputFile(outputFileName, out);
 527
 528                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 529                 InputData input(sabundfile, "sabund");
 530                 SAbundVector* sabund = input.getSAbundVector();
 531                 string lastLabel = sabund->getLabel();
 532                 set<string> processedLabels;
 533                 set<string> userLabels = labels;
 534
 535                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 536
 537                         if (m->control_pressed) { delete sabund; out.close(); return 0; }
 538
 539                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 540
 541                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 542                                 processedLabels.insert(sabund->getLabel());
 543                                 userLabels.erase(sabund->getLabel());
 544
 545                                 if (sabund->getMaxRank() > nseqs) {
 546                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 547                                 }else { sabund->clear(); }
 548
 549                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 550                         }
 551
 552                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 553                                 string saveLabel = sabund->getLabel();
 554
 555                                 delete sabund;
 556                                 sabund = input.getSAbundVector(lastLabel);
 557
 558                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 559                                 processedLabels.insert(sabund->getLabel());
 560                                 userLabels.erase(sabund->getLabel());
 561
 562                                 if (sabund->getMaxRank() > nseqs) {
 563                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 564                                 }else { sabund->clear(); }
 565
 566                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 567
 568                                 //restore real lastlabel to save below
 569                                 sabund->setLabel(saveLabel);
 570                         }
 571
 572                         lastLabel = sabund->getLabel();
 573
 574                         delete sabund;
 575                         sabund = input.getSAbundVector();
 576                 }
 577
 578                 if (m->control_pressed) {  out.close(); return 0; }
 579
 580                 //output error messages about any remaining user labels
 581                 set<string>::iterator it;
 582                 bool needToRun = false;
 583                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 584                         m->mothurOut("Your file does not include the label " + *it);
 585                         if (processedLabels.count(lastLabel) != 1) {
 586                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 587                                 needToRun = true;
 588                         }else {
 589                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 590                         }
 591                 }
 592
 593                 //run last label if you need to
 594                 if (needToRun == true)  {
 595                         if (sabund != NULL) {   delete sabund;  }
 596                         sabund = input.getSAbundVector(lastLabel);
 597
 598                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 599
 600                         if (sabund->getMaxRank() > nseqs) {
 601                                 for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 602                         }else { sabund->clear(); }
 603
 604                         if (sabund->getNumBins() > 0) { sabund->print(out); }
 605
 606                         delete sabund;
 607                 }
 608
 609                 return 0;
 610         }
 611         catch(exception& e) {
 612                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 613                 exit(1);
 614         }
 615 }
 616 //**********************************************************************************************************************
 617 int RemoveRareCommand::processRabund(){
 618         try {
 619                 string thisOutputDir = outputDir;
 620                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
 621                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + getOutputFileNameTag("rabund", rabundfile);
 622                 outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 623
 624                 ofstream out;
 625                 m->openOutputFile(outputFileName, out);
 626
 627                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 628                 InputData input(rabundfile, "rabund");
 629                 RAbundVector* rabund = input.getRAbundVector();
 630                 string lastLabel = rabund->getLabel();
 631                 set<string> processedLabels;
 632                 set<string> userLabels = labels;
 633
 634                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 635
 636                         if (m->control_pressed) { delete rabund; out.close(); return 0; }
 637
 638                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
 639
 640                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 641                                 processedLabels.insert(rabund->getLabel());
 642                                 userLabels.erase(rabund->getLabel());
 643
 644                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 645                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 646                                         if (rabund->get(i) > nseqs) {
 647                                                 newRabund.push_back(rabund->get(i));
 648                                         }
 649                                 }
 650                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 651                         }
 652
 653                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 654                                 string saveLabel = rabund->getLabel();
 655
 656                                 delete rabund;
 657                                 rabund = input.getRAbundVector(lastLabel);
 658
 659                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 660                                 processedLabels.insert(rabund->getLabel());
 661                                 userLabels.erase(rabund->getLabel());
 662
 663                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 664                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 665                                         if (rabund->get(i) > nseqs) {
 666                                                 newRabund.push_back(rabund->get(i));
 667                                         }
 668                                 }
 669                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 670
 671                                 //restore real lastlabel to save below
 672                                 rabund->setLabel(saveLabel);
 673                         }
 674
 675                         lastLabel = rabund->getLabel();
 676
 677                         delete rabund;
 678                         rabund = input.getRAbundVector();
 679                 }
 680
 681                 if (m->control_pressed) {  out.close(); return 0; }
 682
 683                 //output error messages about any remaining user labels
 684                 set<string>::iterator it;
 685                 bool needToRun = false;
 686                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 687                         m->mothurOut("Your file does not include the label " + *it);
 688                         if (processedLabels.count(lastLabel) != 1) {
 689                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 690                                 needToRun = true;
 691                         }else {
 692                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 693                         }
 694                 }
 695
 696                 //run last label if you need to
 697                 if (needToRun == true)  {
 698                         if (rabund != NULL) {   delete rabund;  }
 699                         rabund = input.getRAbundVector(lastLabel);
 700
 701                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 702
 703                         RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 704                         for (int i = 0; i < rabund->getNumBins(); i++) {
 705                                 if (rabund->get(i) > nseqs) {
 706                                         newRabund.push_back(rabund->get(i));
 707                                 }
 708                         }
 709                         if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 710
 711                         delete rabund;
 712                 }
 713
 714                 return 0;
 715         }
 716         catch(exception& e) {
 717                 m->errorOut(e, "RemoveRareCommand", "processRabund");
 718                 exit(1);
 719         }
 720 }
 721 //**********************************************************************************************************************
 722 int RemoveRareCommand::processShared(){
 723         try {
 724                 m->setGroups(Groups);
 725
 726                 string thisOutputDir = outputDir;
 727                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 728                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + getOutputFileNameTag("shared", sharedfile);
 729                 outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
 730
 731                 ofstream out;
 732                 m->openOutputFile(outputFileName, out);
 733
 734                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 735                 InputData input(sharedfile, "sharedfile");
 736                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 737                 string lastLabel = lookup[0]->getLabel();
 738                 set<string> processedLabels;
 739                 set<string> userLabels = labels;
 740
 741                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 742
 743                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  out.close(); return 0; }
 744
 745                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 746
 747                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 748                                 processedLabels.insert(lookup[0]->getLabel());
 749                                 userLabels.erase(lookup[0]->getLabel());
 750
 751                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 752                                 processLookup(lookup, out);
 753                         }
 754
 755                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 756                                 string saveLabel = lookup[0]->getLabel();
 757
 758                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 759                                 lookup = input.getSharedRAbundVectors(lastLabel);
 760
 761                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 762                                 processedLabels.insert(lookup[0]->getLabel());
 763                                 userLabels.erase(lookup[0]->getLabel());
 764
 765                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 766                                 processLookup(lookup, out);
 767
 768                                 //restore real lastlabel to save below
 769                                 lookup[0]->setLabel(saveLabel);
 770                         }
 771
 772                         lastLabel = lookup[0]->getLabel();
 773
 774                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 775                         lookup = input.getSharedRAbundVectors();
 776                 }
 777
 778                 if (m->control_pressed) {  out.close(); return 0; }
 779
 780                 //output error messages about any remaining user labels
 781                 set<string>::iterator it;
 782                 bool needToRun = false;
 783                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 784                         m->mothurOut("Your file does not include the label " + *it);
 785                         if (processedLabels.count(lastLabel) != 1) {
 786                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 787                                 needToRun = true;
 788                         }else {
 789                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 790                         }
 791                 }
 792
 793                 //run last label if you need to
 794                 if (needToRun == true)  {
 795                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       }  }
 796                         lookup = input.getSharedRAbundVectors(lastLabel);
 797
 798                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 799
 800                         if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 801                         processLookup(lookup, out);
 802
 803                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 804                 }
 805
 806                 return 0;
 807         }
 808         catch(exception& e) {
 809                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 810                 exit(1);
 811         }
 812 }
 813 //**********************************************************************************************************************
 814 int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup, ofstream& out){
 815         try {
 816
 817                 vector<SharedRAbundVector> newRabunds;  newRabunds.resize(lookup.size());
 818                 for (int i = 0; i < lookup.size(); i++) {
 819                         newRabunds[i].setGroup(lookup[i]->getGroup());
 820                         newRabunds[i].setLabel(lookup[i]->getLabel());
 821                 }
 822
 823                 if (byGroup) {
 824
 825                         //for each otu
 826                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 827                                 bool allZero = true;
 828
 829                                 if (m->control_pressed) { return 0; }
 830
 831                                 //for each group
 832                                 for (int j = 0; j < lookup.size(); j++) {
 833
 834                                         //are you rare?
 835                                         if (lookup[j]->getAbundance(i) > nseqs) {
 836                                                 newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 837                                                 allZero = false;
 838                                         }else {
 839                                                 newRabunds[j].push_back(0, newRabunds[j].getGroup());
 840                                         }
 841                                 }
 842
 843                                 //eliminates zero otus
 844                                 if (allZero) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 845                         }
 846                 }else {
 847                         //for each otu
 848                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 849
 850                                 if (m->control_pressed) { return 0; }
 851
 852                                 int totalAbund = 0;
 853                                 //get total otu abundance
 854                                 for (int j = 0; j < lookup.size(); j++) {
 855                                         newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 856                                         totalAbund += lookup[j]->getAbundance(i);
 857                                 }
 858
 859                                 //eliminates otus below rare cutoff
 860                                 if (totalAbund <= nseqs) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 861                         }
 862                 }
 863
 864                 //do we have any otus above the rare cutoff
 865                 if (newRabunds[0].getNumBins() != 0) {
 866                         for (int j = 0; j < newRabunds.size(); j++) {
 867                                 out << newRabunds[j].getLabel() << '\t' << newRabunds[j].getGroup() << '\t';
 868                                 newRabunds[j].print(out);
 869                         }
 870                 }
 871
 872                 return 0;
 873         }
 874         catch(exception& e) {
 875                 m->errorOut(e, "RemoveRareCommand", "processLookup");
 876                 exit(1);
 877         }
 878 }
 879 //**********************************************************************************************************************
 880
 881
 882
 883