removegroupscommand.cpp

   1 /*
   2  *  removegroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removegroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> RemoveGroupsCommand::setParameters(){
  18         try {
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(pfasta);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "sharedGroup", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  22                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "sharedGroup", "FNGLT",false,false); parameters.push_back(pgroup);
  23                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(plist);
  24                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(ptaxonomy);
  25                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  26                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "RemoveGroupsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string RemoveGroupsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The remove.groups command removes sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy or sharedfile.\n";
  44                 helpString += "It outputs a file containing the sequences NOT in the those specified groups, or with a sharedfile eliminates the groups you selected.\n";
  45                 helpString += "The remove.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared and groups. The group parameter is required, unless you have a current group file or are using a sharedfile.\n";
  46                 helpString += "You must also provide an accnos containing the list of groups to remove or set the groups parameter to the groups you wish to remove.\n";
  47                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like removed.  You can separate group names with dashes.\n";
  48                 helpString += "The remove.groups command should be in the following format: remove.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
  49                 helpString += "Example remove.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
  50                 helpString += "or remove.groups(groups=pasture, fasta=amazon.fasta, amazon.groups).\n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "RemoveGroupsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59
  60 //**********************************************************************************************************************
  61 RemoveGroupsCommand::RemoveGroupsCommand(){
  62         try {
  63                 abort = true; calledHelp = true;
  64                 setParameters();
  65                 vector<string> tempOutNames;
  66                 outputTypes["fasta"] = tempOutNames;
  67                 outputTypes["taxonomy"] = tempOutNames;
  68                 outputTypes["name"] = tempOutNames;
  69                 outputTypes["group"] = tempOutNames;
  70                 outputTypes["list"] = tempOutNames;
  71                 outputTypes["shared"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 RemoveGroupsCommand::RemoveGroupsCommand(string option)  {
  80         try {
  81                 abort = false; calledHelp = false;
  82
  83                 //allow user to run help
  84                 if(option == "help") { help(); abort = true; calledHelp = true; }
  85                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  86
  87                 else {
  88                         vector<string> myArray = setParameters();
  89
  90                         OptionParser parser(option);
  91                         map<string,string> parameters = parser.getParameters();
  92
  93                         ValidParameters validParameter;
  94                         map<string,string>::iterator it;
  95
  96                         //check to make sure all parameters are valid for command
  97                         for (it = parameters.begin(); it != parameters.end(); it++) {
  98                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  99                         }
 100
 101                         //initialize outputTypes
 102                         vector<string> tempOutNames;
 103                         outputTypes["fasta"] = tempOutNames;
 104                         outputTypes["taxonomy"] = tempOutNames;
 105                         outputTypes["name"] = tempOutNames;
 106                         outputTypes["group"] = tempOutNames;
 107                         outputTypes["list"] = tempOutNames;
 108                         outputTypes["shared"] = tempOutNames;
 109
 110
 111                         //if the user changes the output directory command factory will send this info to us in the output parameter
 112                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 113
 114                         //if the user changes the input directory command factory will send this info to us in the output parameter
 115                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 116                         if (inputDir == "not found"){   inputDir = "";          }
 117                         else {
 118                                 string path;
 119                                 it = parameters.find("fasta");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 125                                 }
 126
 127                                 it = parameters.find("accnos");
 128                                 //user has given a template file
 129                                 if(it != parameters.end()){
 130                                         path = m->hasPath(it->second);
 131                                         //if the user has not given a path then, add inputdir. else leave path alone.
 132                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 133                                 }
 134
 135                                 it = parameters.find("list");
 136                                 //user has given a template file
 137                                 if(it != parameters.end()){
 138                                         path = m->hasPath(it->second);
 139                                         //if the user has not given a path then, add inputdir. else leave path alone.
 140                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 141                                 }
 142
 143                                 it = parameters.find("name");
 144                                 //user has given a template file
 145                                 if(it != parameters.end()){
 146                                         path = m->hasPath(it->second);
 147                                         //if the user has not given a path then, add inputdir. else leave path alone.
 148                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 149                                 }
 150
 151                                 it = parameters.find("group");
 152                                 //user has given a template file
 153                                 if(it != parameters.end()){
 154                                         path = m->hasPath(it->second);
 155                                         //if the user has not given a path then, add inputdir. else leave path alone.
 156                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 157                                 }
 158
 159                                 it = parameters.find("taxonomy");
 160                                 //user has given a template file
 161                                 if(it != parameters.end()){
 162                                         path = m->hasPath(it->second);
 163                                         //if the user has not given a path then, add inputdir. else leave path alone.
 164                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 165                                 }
 166
 167                                 it = parameters.find("shared");
 168                                 //user has given a template file
 169                                 if(it != parameters.end()){
 170                                         path = m->hasPath(it->second);
 171                                         //if the user has not given a path then, add inputdir. else leave path alone.
 172                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 173                                 }
 174                         }
 175
 176
 177                         //check for required parameters
 178                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 179                         if (accnosfile == "not open") { accnosfile = ""; abort = true; }
 180                         else if (accnosfile == "not found") {  accnosfile = ""; }
 181                         else { m->setAccnosFile(accnosfile); }
 182
 183                         fastafile = validParameter.validFile(parameters, "fasta", true);
 184                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 185                         else if (fastafile == "not found") {  fastafile = "";  }
 186                         else { m->setFastaFile(fastafile); }
 187
 188                         namefile = validParameter.validFile(parameters, "name", true);
 189                         if (namefile == "not open") { namefile = ""; abort = true; }
 190                         else if (namefile == "not found") {  namefile = "";  }
 191                         else { m->setNameFile(namefile); }
 192
 193                         groupfile = validParameter.validFile(parameters, "group", true);
 194                         if (groupfile == "not open") { groupfile = "";  abort = true; }
 195                         else if (groupfile == "not found") {    groupfile = "";         }
 196                         else { m->setGroupFile(groupfile); }
 197
 198                         listfile = validParameter.validFile(parameters, "list", true);
 199                         if (listfile == "not open") { listfile = ""; abort = true; }
 200                         else if (listfile == "not found") {  listfile = "";  }
 201                         else { m->setListFile(listfile); }
 202
 203                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 204                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 205                         else if (taxfile == "not found") {  taxfile = "";  }
 206                         else { m->setTaxonomyFile(taxfile); }
 207
 208                         groups = validParameter.validFile(parameters, "groups", false);
 209                         if (groups == "not found") { groups = ""; }
 210                         else {
 211                                 m->splitAtDash(groups, Groups);
 212                                 m->setGroups(Groups);
 213                         }
 214
 215                         sharedfile = validParameter.validFile(parameters, "shared", true);
 216                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 217                         else if (sharedfile == "not found") {  sharedfile = "";  }
 218                         else { m->setSharedFile(sharedfile); }
 219
 220                         groupfile = validParameter.validFile(parameters, "group", true);
 221                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 222                         else if (groupfile == "not found") {    groupfile = ""; }
 223                         else { m->setGroupFile(groupfile); }
 224
 225                         if ((sharedfile == "") && (groupfile == "")) {
 226                                 //is there are current file available for any of these?
 227                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 228                                         //give priority to group, then shared
 229                                         groupfile = m->getGroupFile();
 230                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 231                                         else {
 232                                                 sharedfile = m->getSharedFile();
 233                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 234                                                 else {
 235                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 236                                                 }
 237                                         }
 238                                 }else {
 239                                         //give priority to shared, then group
 240                                         sharedfile = m->getSharedFile();
 241                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 242                                         else {
 243                                                 groupfile = m->getGroupFile();
 244                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 245                                                 else {
 246                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 247                                                 }
 248                                         }
 249                                 }
 250                         }
 251
 252                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file containing group names or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 253
 254                         if ((fastafile == "") && (namefile == "") && (groupfile == "")  && (sharedfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared or list."); m->mothurOutEndLine(); abort = true; }
 255                         if ((groupfile == "") && ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")))  { m->mothurOut("If using a fasta, name, taxonomy, group or list, then you must provide a group file."); m->mothurOutEndLine(); abort = true; }
 256
 257                         if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 258                                 vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 259                                 parser.getNameFile(files);
 260                         }
 261
 262                 }
 263
 264         }
 265         catch(exception& e) {
 266                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 267                 exit(1);
 268         }
 269 }
 270 //**********************************************************************************************************************
 271
 272 int RemoveGroupsCommand::execute(){
 273         try {
 274
 275                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 276
 277                 //get groups you want to remove
 278                 if (accnosfile != "") { readAccnos(); }
 279
 280                 if (groupfile != "") {
 281                         groupMap = new GroupMap(groupfile);
 282                         groupMap->readMap();
 283
 284                         //make sure groups are valid
 285                         //takes care of user setting groupNames that are invalid or setting groups=all
 286                         SharedUtil* util = new SharedUtil();
 287                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 288                         util->setGroups(Groups, namesGroups);
 289                         delete util;
 290
 291                         //fill names with names of sequences that are from the groups we want to remove
 292                         fillNames();
 293
 294                         delete groupMap;
 295                 }
 296
 297                 if (m->control_pressed) { return 0; }
 298
 299                 //read through the correct file and output lines you want to keep
 300                 if (namefile != "")                     {               readName();             }
 301                 if (fastafile != "")            {               readFasta();    }
 302                 if (groupfile != "")            {               readGroup();    }
 303                 if (listfile != "")                     {               readList();             }
 304                 if (taxfile != "")                      {               readTax();              }
 305                 if (sharedfile != "")           {               readShared();   }
 306
 307                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 308
 309                 if (outputNames.size() != 0) {
 310                         m->mothurOutEndLine();
 311                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 312                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 313                         m->mothurOutEndLine();
 314
 315                         //set fasta file as new current fastafile
 316                         string current = "";
 317                         itTypes = outputTypes.find("fasta");
 318                         if (itTypes != outputTypes.end()) {
 319                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 320                         }
 321
 322                         itTypes = outputTypes.find("name");
 323                         if (itTypes != outputTypes.end()) {
 324                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 325                         }
 326
 327                         itTypes = outputTypes.find("group");
 328                         if (itTypes != outputTypes.end()) {
 329                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 330                         }
 331
 332                         itTypes = outputTypes.find("list");
 333                         if (itTypes != outputTypes.end()) {
 334                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 335                         }
 336
 337                         itTypes = outputTypes.find("taxonomy");
 338                         if (itTypes != outputTypes.end()) {
 339                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 340                         }
 341
 342                         itTypes = outputTypes.find("shared");
 343                         if (itTypes != outputTypes.end()) {
 344                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 345                         }
 346                 }
 347
 348                 return 0;
 349         }
 350
 351         catch(exception& e) {
 352                 m->errorOut(e, "RemoveGroupsCommand", "execute");
 353                 exit(1);
 354         }
 355 }
 356
 357 //**********************************************************************************************************************
 358 int RemoveGroupsCommand::readFasta(){
 359         try {
 360                 string thisOutputDir = outputDir;
 361                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 362                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 363
 364                 ofstream out;
 365                 m->openOutputFile(outputFileName, out);
 366
 367                 ifstream in;
 368                 m->openInputFile(fastafile, in);
 369                 string name;
 370
 371                 bool wroteSomething = false;
 372                 int removedCount = 0;
 373
 374                 while(!in.eof()){
 375                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 376
 377                         Sequence currSeq(in);
 378                         name = currSeq.getName();
 379
 380                         if (name != "") {
 381                                 //if this name is in the accnos file
 382                                 if (names.count(name) == 0) {
 383                                         wroteSomething = true;
 384                                         currSeq.printSequence(out);
 385                                 }else {
 386                                         //if you are not in the accnos file check if you are a name that needs to be changed
 387                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 388                                         if (it != uniqueToRedundant.end()) {
 389                                                 wroteSomething = true;
 390                                                 currSeq.setName(it->second);
 391                                                 currSeq.printSequence(out);
 392                                         }else { removedCount++; }
 393                                 }
 394                         }
 395                         m->gobble(in);
 396                 }
 397                 in.close();
 398                 out.close();
 399
 400                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 401                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 402
 403                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 404
 405                 return 0;
 406
 407         }
 408         catch(exception& e) {
 409                 m->errorOut(e, "RemoveGroupsCommand", "readFasta");
 410                 exit(1);
 411         }
 412 }
 413 //**********************************************************************************************************************
 414 int RemoveGroupsCommand::readShared(){
 415         try {
 416                 string thisOutputDir = outputDir;
 417                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 418
 419                 //get group names from sharedfile so we can set Groups to the groupNames we want to keep
 420                 //that way we can take advantage of the reads in inputdata and sharedRabundVector
 421                 InputData* tempInput = new InputData(sharedfile, "sharedfile");
 422                 vector<SharedRAbundVector*> lookup = tempInput->getSharedRAbundVectors();
 423
 424                 //save m->Groups
 425                 vector<string> allGroupsNames = m->getAllGroups();
 426                 vector<string> mothurOutGroups = m->getGroups();
 427
 428                 vector<string> groupsToKeep;
 429                 for (int i = 0; i < allGroupsNames.size(); i++) {
 430                         if (!m->inUsersGroups(allGroupsNames[i], m->getGroups())) {
 431                                 groupsToKeep.push_back(allGroupsNames[i]);
 432                         }
 433                 }
 434
 435                 if (allGroupsNames.size() == groupsToKeep.size()) { m->mothurOut("Your file does not contain any groups you wish to remove."); m->mothurOutEndLine(); m->setGroups(mothurOutGroups); delete tempInput; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  return 0; }
 436
 437                 //reset read
 438                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 439                 delete tempInput;
 440                 m->setGroups(groupsToKeep);
 441                 m->clearAllGroups();
 442                 m->names.clear();
 443                 m->saveNextLabel = "";
 444                 m->printedHeaders = false;
 445                 m->currentBinLabels.clear();
 446                 m->binLabelsInFile.clear();
 447
 448                 InputData input(sharedfile, "sharedfile");
 449                 lookup = input.getSharedRAbundVectors();
 450
 451                 bool wroteSomething = false;
 452
 453                 while(lookup[0] != NULL) {
 454
 455                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 456                         ofstream out;
 457                         m->openOutputFile(outputFileName, out);
 458                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 459
 460                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 461
 462                         lookup[0]->printHeaders(out);
 463
 464                         for (int i = 0; i < lookup.size(); i++) {
 465                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 466                                 lookup[i]->print(out);
 467                                 wroteSomething = true;
 468
 469                         }
 470
 471                         //get next line to process
 472                         //prevent memory leak
 473                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 474                         lookup = input.getSharedRAbundVectors();
 475
 476                         out.close();
 477                 }
 478
 479
 480                 m->setGroups(mothurOutGroups);
 481
 482                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 483
 484                 string groupsString = "";
 485                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 486                 groupsString += Groups[Groups.size()-1];
 487
 488                 m->mothurOut("Removed groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 489
 490                 return 0;
 491
 492         }
 493         catch(exception& e) {
 494                 m->errorOut(e, "RemoveGroupsCommand", "readShared");
 495                 exit(1);
 496         }
 497 }
 498 //**********************************************************************************************************************
 499 int RemoveGroupsCommand::readList(){
 500         try {
 501                 string thisOutputDir = outputDir;
 502                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 503                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 504
 505                 ofstream out;
 506                 m->openOutputFile(outputFileName, out);
 507
 508                 ifstream in;
 509                 m->openInputFile(listfile, in);
 510
 511                 bool wroteSomething = false;
 512                 int removedCount = 0;
 513
 514                 while(!in.eof()){
 515
 516                         removedCount = 0;
 517
 518                         //read in list vector
 519                         ListVector list(in);
 520
 521                         //make a new list vector
 522                         ListVector newList;
 523                         newList.setLabel(list.getLabel());
 524
 525                         //for each bin
 526                         for (int i = 0; i < list.getNumBins(); i++) {
 527                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 528
 529                                 //parse out names that are in accnos file
 530                                 string binnames = list.get(i);
 531
 532                                 string newNames = "";
 533                                 while (binnames.find_first_of(',') != -1) {
 534                                         string name = binnames.substr(0,binnames.find_first_of(','));
 535                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 536
 537                                         //if that name is in the .accnos file, add it
 538                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 539                                         else {
 540                                                 //if you are not in the accnos file check if you are a name that needs to be changed
 541                                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 542                                                 if (it != uniqueToRedundant.end()) {
 543                                                         newNames += it->second + ",";
 544                                                 }else { removedCount++; }
 545                                         }
 546                                 }
 547
 548                                 //get last name
 549                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 550                                 else { //if you are not in the accnos file check if you are a name that needs to be changed
 551                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 552                                         if (it != uniqueToRedundant.end()) {
 553                                                 newNames += it->second + ",";
 554                                         }else { removedCount++; }
 555                                 }
 556
 557                                 //if there are names in this bin add to new list
 558                                 if (newNames != "") {
 559                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 560                                         newList.push_back(newNames);
 561                                 }
 562                         }
 563
 564                         //print new listvector
 565                         if (newList.getNumBins() != 0) {
 566                                 wroteSomething = true;
 567                                 newList.print(out);
 568                         }
 569
 570                         m->gobble(in);
 571                 }
 572                 in.close();
 573                 out.close();
 574
 575                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 576                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 577
 578                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 579
 580                 return 0;
 581
 582         }
 583         catch(exception& e) {
 584                 m->errorOut(e, "RemoveGroupsCommand", "readList");
 585                 exit(1);
 586         }
 587 }
 588 //**********************************************************************************************************************
 589 int RemoveGroupsCommand::readName(){
 590         try {
 591                 string thisOutputDir = outputDir;
 592                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 593                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 594
 595                 ofstream out;
 596                 m->openOutputFile(outputFileName, out);
 597
 598                 ifstream in;
 599                 m->openInputFile(namefile, in);
 600                 string name, firstCol, secondCol;
 601
 602                 bool wroteSomething = false;
 603                 int removedCount = 0;
 604
 605                 while(!in.eof()){
 606                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 607
 608                         in >> firstCol;         m->gobble(in);
 609                         in >> secondCol;
 610
 611                         vector<string> parsedNames;
 612                         m->splitAtComma(secondCol, parsedNames);
 613
 614                         vector<string> validSecond;  validSecond.clear();
 615                         for (int i = 0; i < parsedNames.size(); i++) {
 616                                 if (names.count(parsedNames[i]) == 0) {
 617                                         validSecond.push_back(parsedNames[i]);
 618                                 }
 619                         }
 620
 621                         removedCount += parsedNames.size()-validSecond.size();
 622
 623                         //if the name in the first column is in the set then print it and any other names in second column also in set
 624                         if (names.count(firstCol) == 0) {
 625
 626                                 wroteSomething = true;
 627
 628                                 out << firstCol << '\t';
 629
 630                                 //you know you have at least one valid second since first column is valid
 631                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 632                                 out << validSecond[validSecond.size()-1] << endl;
 633
 634                                 //make first name in set you come to first column and then add the remaining names to second column
 635                         }else {
 636
 637                                 //you want part of this row
 638                                 if (validSecond.size() != 0) {
 639
 640                                         wroteSomething = true;
 641
 642                                         out << validSecond[0] << '\t';
 643
 644                                         //you know you have at least one valid second since first column is valid
 645                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 646                                         out << validSecond[validSecond.size()-1] << endl;
 647                                         uniqueToRedundant[firstCol] = validSecond[0];
 648                                 }
 649                         }
 650
 651                         m->gobble(in);
 652                 }
 653                 in.close();
 654                 out.close();
 655
 656                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 657                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 658
 659                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 660
 661                 return 0;
 662         }
 663         catch(exception& e) {
 664                 m->errorOut(e, "RemoveGroupsCommand", "readName");
 665                 exit(1);
 666         }
 667 }
 668
 669 //**********************************************************************************************************************
 670 int RemoveGroupsCommand::readGroup(){
 671         try {
 672                 string thisOutputDir = outputDir;
 673                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 674                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 675
 676                 ofstream out;
 677                 m->openOutputFile(outputFileName, out);
 678
 679                 ifstream in;
 680                 m->openInputFile(groupfile, in);
 681                 string name, group;
 682
 683                 bool wroteSomething = false;
 684                 int removedCount = 0;
 685
 686                 while(!in.eof()){
 687                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 688
 689                         in >> name;                             //read from first column
 690                         in >> group;                    //read from second column
 691
 692                         //if this name is in the accnos file
 693                         if (names.count(name) == 0) {
 694                                 wroteSomething = true;
 695                                 out << name << '\t' << group << endl;
 696                         }else {  removedCount++;  }
 697
 698                         m->gobble(in);
 699                 }
 700                 in.close();
 701                 out.close();
 702
 703                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 704                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 705
 706                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 707
 708
 709                 return 0;
 710         }
 711         catch(exception& e) {
 712                 m->errorOut(e, "RemoveGroupsCommand", "readGroup");
 713                 exit(1);
 714         }
 715 }
 716 //**********************************************************************************************************************
 717 int RemoveGroupsCommand::readTax(){
 718         try {
 719                 string thisOutputDir = outputDir;
 720                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 721                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 722                 ofstream out;
 723                 m->openOutputFile(outputFileName, out);
 724
 725                 ifstream in;
 726                 m->openInputFile(taxfile, in);
 727                 string name, tax;
 728
 729                 bool wroteSomething = false;
 730                 int removedCount = 0;
 731
 732                 while(!in.eof()){
 733                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 734
 735                         in >> name;                             //read from first column
 736                         in >> tax;                      //read from second column
 737
 738                         //if this name is in the accnos file
 739                         if (names.count(name) == 0) {
 740                                 wroteSomething = true;
 741                                 out << name << '\t' << tax << endl;
 742                         }else {  //if you are not in the accnos file check if you are a name that needs to be changed
 743                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 744                                 if (it != uniqueToRedundant.end()) {
 745                                         wroteSomething = true;
 746                                         out << it->second << '\t' << tax << endl;
 747                                 }else { removedCount++; }  }
 748
 749                         m->gobble(in);
 750                 }
 751                 in.close();
 752                 out.close();
 753
 754                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 755                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 756
 757                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 758
 759                 return 0;
 760         }
 761         catch(exception& e) {
 762                 m->errorOut(e, "RemoveGroupsCommand", "readTax");
 763                 exit(1);
 764         }
 765 }
 766 //**********************************************************************************************************************
 767 void RemoveGroupsCommand::readAccnos(){
 768         try {
 769                 Groups.clear();
 770
 771                 ifstream in;
 772                 m->openInputFile(accnosfile, in);
 773                 string name;
 774
 775                 while(!in.eof()){
 776                         in >> name;
 777
 778                         Groups.push_back(name);
 779
 780                         m->gobble(in);
 781                 }
 782                 in.close();
 783
 784                 m->setGroups(Groups);
 785
 786         }
 787         catch(exception& e) {
 788                 m->errorOut(e, "RemoveGroupsCommand", "readAccnos");
 789                 exit(1);
 790         }
 791 }
 792 //**********************************************************************************************************************
 793 int RemoveGroupsCommand::fillNames(){
 794         try {
 795                 vector<string> seqs = groupMap->getNamesSeqs();
 796
 797                 for (int i = 0; i < seqs.size(); i++) {
 798
 799                         if (m->control_pressed) { return 0; }
 800
 801                         string group = groupMap->getGroup(seqs[i]);
 802
 803                         if (m->inUsersGroups(group, Groups)) {
 804                                 names.insert(seqs[i]);
 805                         }
 806                 }
 807
 808                 return 0;
 809         }
 810         catch(exception& e) {
 811                 m->errorOut(e, "RemoveGroupsCommand", "fillNames");
 812                 exit(1);
 813         }
 814 }
 815
 816 //**********************************************************************************************************************
 817
 818
 819