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", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
  22                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pgroup);
  23                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  24                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",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                         listfile = validParameter.validFile(parameters, "list", true);
 194                         if (listfile == "not open") { listfile = ""; abort = true; }
 195                         else if (listfile == "not found") {  listfile = "";  }
 196                         else { m->setListFile(listfile); }
 197
 198                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 199                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 200                         else if (taxfile == "not found") {  taxfile = "";  }
 201                         else { m->setTaxonomyFile(taxfile); }
 202
 203                         groups = validParameter.validFile(parameters, "groups", false);
 204                         if (groups == "not found") { groups = ""; }
 205                         else {
 206                                 m->splitAtDash(groups, Groups);
 207                                 m->Groups = Groups;
 208                         }
 209
 210                         sharedfile = validParameter.validFile(parameters, "shared", true);
 211                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 212                         else if (sharedfile == "not found") {  sharedfile = "";  }
 213                         else { m->setSharedFile(sharedfile); }
 214
 215                         groupfile = validParameter.validFile(parameters, "group", true);
 216                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 217                         else if (groupfile == "not found") {    groupfile = ""; }
 218                         else { m->setGroupFile(groupfile); }
 219
 220                         if ((sharedfile == "") && (groupfile == "")) {
 221                                 //is there are current file available for any of these?
 222                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 223                                         //give priority to group, then shared
 224                                         groupfile = m->getGroupFile();
 225                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 226                                         else {
 227                                                 sharedfile = m->getSharedFile();
 228                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 229                                                 else {
 230                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 231                                                 }
 232                                         }
 233                                 }else {
 234                                         //give priority to shared, then group
 235                                         sharedfile = m->getSharedFile();
 236                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 237                                         else {
 238                                                 groupfile = m->getGroupFile();
 239                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 240                                                 else {
 241                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 242                                                 }
 243                                         }
 244                                 }
 245                         }
 246
 247                         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; }
 248
 249                         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; }
 250                         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; }
 251                 }
 252
 253         }
 254         catch(exception& e) {
 255                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 256                 exit(1);
 257         }
 258 }
 259 //**********************************************************************************************************************
 260
 261 int RemoveGroupsCommand::execute(){
 262         try {
 263
 264                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 265
 266                 //get groups you want to remove
 267                 if (accnosfile != "") { readAccnos(); }
 268
 269                 if (groupfile != "") {
 270                         groupMap = new GroupMap(groupfile);
 271                         groupMap->readMap();
 272
 273                         //make sure groups are valid
 274                         //takes care of user setting groupNames that are invalid or setting groups=all
 275                         SharedUtil* util = new SharedUtil();
 276                         util->setGroups(Groups, groupMap->namesOfGroups);
 277                         delete util;
 278
 279                         //fill names with names of sequences that are from the groups we want to remove
 280                         fillNames();
 281
 282                         delete groupMap;
 283                 }
 284
 285                 if (m->control_pressed) { return 0; }
 286
 287                 //read through the correct file and output lines you want to keep
 288                 if (namefile != "")                     {               readName();             }
 289                 if (fastafile != "")            {               readFasta();    }
 290                 if (groupfile != "")            {               readGroup();    }
 291                 if (listfile != "")                     {               readList();             }
 292                 if (taxfile != "")                      {               readTax();              }
 293                 if (sharedfile != "")           {               readShared();   }
 294
 295                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 296
 297                 if (outputNames.size() != 0) {
 298                         m->mothurOutEndLine();
 299                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 300                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 301                         m->mothurOutEndLine();
 302
 303                         //set fasta file as new current fastafile
 304                         string current = "";
 305                         itTypes = outputTypes.find("fasta");
 306                         if (itTypes != outputTypes.end()) {
 307                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 308                         }
 309
 310                         itTypes = outputTypes.find("name");
 311                         if (itTypes != outputTypes.end()) {
 312                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 313                         }
 314
 315                         itTypes = outputTypes.find("group");
 316                         if (itTypes != outputTypes.end()) {
 317                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 318                         }
 319
 320                         itTypes = outputTypes.find("list");
 321                         if (itTypes != outputTypes.end()) {
 322                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 323                         }
 324
 325                         itTypes = outputTypes.find("taxonomy");
 326                         if (itTypes != outputTypes.end()) {
 327                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(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
 336                 return 0;
 337         }
 338
 339         catch(exception& e) {
 340                 m->errorOut(e, "RemoveGroupsCommand", "execute");
 341                 exit(1);
 342         }
 343 }
 344
 345 //**********************************************************************************************************************
 346 int RemoveGroupsCommand::readFasta(){
 347         try {
 348                 string thisOutputDir = outputDir;
 349                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 350                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 351
 352                 ofstream out;
 353                 m->openOutputFile(outputFileName, out);
 354
 355                 ifstream in;
 356                 m->openInputFile(fastafile, in);
 357                 string name;
 358
 359                 bool wroteSomething = false;
 360                 int removedCount = 0;
 361
 362                 while(!in.eof()){
 363                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 364
 365                         Sequence currSeq(in);
 366                         name = currSeq.getName();
 367
 368                         if (name != "") {
 369                                 //if this name is in the accnos file
 370                                 if (names.count(name) == 0) {
 371                                         wroteSomething = true;
 372
 373                                         currSeq.printSequence(out);
 374                                 }else { removedCount++; }
 375                         }
 376                         m->gobble(in);
 377                 }
 378                 in.close();
 379                 out.close();
 380
 381                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 382                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 383
 384                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 385
 386                 return 0;
 387
 388         }
 389         catch(exception& e) {
 390                 m->errorOut(e, "RemoveGroupsCommand", "readFasta");
 391                 exit(1);
 392         }
 393 }
 394 //**********************************************************************************************************************
 395 int RemoveGroupsCommand::readShared(){
 396         try {
 397                 string thisOutputDir = outputDir;
 398                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 399
 400                 //get group names from sharedfile so we can set Groups to the groupNames we want to keep
 401                 //that way we can take advantage of the reads in inputdata and sharedRabundVector
 402                 InputData* tempInput = new InputData(sharedfile, "sharedfile");
 403                 vector<SharedRAbundVector*> lookup = tempInput->getSharedRAbundVectors();
 404
 405                 //save m->Groups
 406                 vector<string> allGroupsNames = m->namesOfGroups;
 407                 vector<string> mothurOutGroups = m->Groups;
 408
 409                 vector<string> groupsToKeep;
 410                 for (int i = 0; i < allGroupsNames.size(); i++) {
 411                         if (!m->inUsersGroups(allGroupsNames[i], m->Groups)) {
 412                                 groupsToKeep.push_back(allGroupsNames[i]);
 413                         }
 414                 }
 415
 416                 if (allGroupsNames.size() == groupsToKeep.size()) { m->mothurOut("Your file does not contain any groups you wish to remove."); m->mothurOutEndLine(); m->Groups = mothurOutGroups; delete tempInput; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  return 0; }
 417
 418                 //reset read
 419                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 420                 delete tempInput;
 421                 m->Groups = groupsToKeep;
 422                 m->namesOfGroups.clear();
 423                 m->names.clear();
 424                 m->saveNextLabel = "";
 425                 m->printedHeaders = false;
 426                 m->currentBinLabels.clear();
 427                 m->binLabelsInFile.clear();
 428
 429                 InputData input(sharedfile, "sharedfile");
 430                 lookup = input.getSharedRAbundVectors();
 431
 432                 bool wroteSomething = false;
 433
 434                 while(lookup[0] != NULL) {
 435
 436                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 437                         ofstream out;
 438                         m->openOutputFile(outputFileName, out);
 439                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 440
 441                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 442
 443                         lookup[0]->printHeaders(out);
 444
 445                         for (int i = 0; i < lookup.size(); i++) {
 446                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 447                                 lookup[i]->print(out);
 448                                 wroteSomething = true;
 449
 450                         }
 451
 452                         //get next line to process
 453                         //prevent memory leak
 454                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 455                         lookup = input.getSharedRAbundVectors();
 456
 457                         out.close();
 458                 }
 459
 460
 461                 m->Groups = mothurOutGroups;
 462
 463                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 464
 465                 string groupsString = "";
 466                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 467                 groupsString += Groups[Groups.size()-1];
 468
 469                 m->mothurOut("Removed groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 470
 471                 return 0;
 472
 473         }
 474         catch(exception& e) {
 475                 m->errorOut(e, "RemoveGroupsCommand", "readShared");
 476                 exit(1);
 477         }
 478 }
 479 //**********************************************************************************************************************
 480 int RemoveGroupsCommand::readList(){
 481         try {
 482                 string thisOutputDir = outputDir;
 483                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 484                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 485
 486                 ofstream out;
 487                 m->openOutputFile(outputFileName, out);
 488
 489                 ifstream in;
 490                 m->openInputFile(listfile, in);
 491
 492                 bool wroteSomething = false;
 493                 int removedCount = 0;
 494
 495                 while(!in.eof()){
 496
 497                         removedCount = 0;
 498
 499                         //read in list vector
 500                         ListVector list(in);
 501
 502                         //make a new list vector
 503                         ListVector newList;
 504                         newList.setLabel(list.getLabel());
 505
 506                         //for each bin
 507                         for (int i = 0; i < list.getNumBins(); i++) {
 508                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 509
 510                                 //parse out names that are in accnos file
 511                                 string binnames = list.get(i);
 512
 513                                 string newNames = "";
 514                                 while (binnames.find_first_of(',') != -1) {
 515                                         string name = binnames.substr(0,binnames.find_first_of(','));
 516                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 517
 518                                         //if that name is in the .accnos file, add it
 519                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 520                                         else { removedCount++; }
 521                                 }
 522
 523                                 //get last name
 524                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 525                                 else { removedCount++; }
 526
 527                                 //if there are names in this bin add to new list
 528                                 if (newNames != "") {
 529                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 530                                         newList.push_back(newNames);
 531                                 }
 532                         }
 533
 534                         //print new listvector
 535                         if (newList.getNumBins() != 0) {
 536                                 wroteSomething = true;
 537                                 newList.print(out);
 538                         }
 539
 540                         m->gobble(in);
 541                 }
 542                 in.close();
 543                 out.close();
 544
 545                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 546                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 547
 548                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 549
 550                 return 0;
 551
 552         }
 553         catch(exception& e) {
 554                 m->errorOut(e, "RemoveGroupsCommand", "readList");
 555                 exit(1);
 556         }
 557 }
 558 //**********************************************************************************************************************
 559 int RemoveGroupsCommand::readName(){
 560         try {
 561                 string thisOutputDir = outputDir;
 562                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 563                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 564
 565                 ofstream out;
 566                 m->openOutputFile(outputFileName, out);
 567
 568                 ifstream in;
 569                 m->openInputFile(namefile, in);
 570                 string name, firstCol, secondCol;
 571
 572                 bool wroteSomething = false;
 573                 int removedCount = 0;
 574
 575                 while(!in.eof()){
 576                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 577
 578                         in >> firstCol;         m->gobble(in);
 579                         in >> secondCol;
 580
 581                         vector<string> parsedNames;
 582                         m->splitAtComma(secondCol, parsedNames);
 583
 584                         vector<string> validSecond;  validSecond.clear();
 585                         for (int i = 0; i < parsedNames.size(); i++) {
 586                                 if (names.count(parsedNames[i]) == 0) {
 587                                         validSecond.push_back(parsedNames[i]);
 588                                 }
 589                         }
 590
 591                         removedCount += parsedNames.size()-validSecond.size();
 592
 593                         //if the name in the first column is in the set then print it and any other names in second column also in set
 594                         if (names.count(firstCol) == 0) {
 595
 596                                 wroteSomething = true;
 597
 598                                 out << firstCol << '\t';
 599
 600                                 //you know you have at least one valid second since first column is valid
 601                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 602                                 out << validSecond[validSecond.size()-1] << endl;
 603
 604                                 //make first name in set you come to first column and then add the remaining names to second column
 605                         }else {
 606
 607                                 //you want part of this row
 608                                 if (validSecond.size() != 0) {
 609
 610                                         wroteSomething = true;
 611
 612                                         out << validSecond[0] << '\t';
 613
 614                                         //you know you have at least one valid second since first column is valid
 615                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 616                                         out << validSecond[validSecond.size()-1] << endl;
 617                                 }
 618                         }
 619
 620                         m->gobble(in);
 621                 }
 622                 in.close();
 623                 out.close();
 624
 625                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 626                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 627
 628                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 629
 630                 return 0;
 631         }
 632         catch(exception& e) {
 633                 m->errorOut(e, "RemoveGroupsCommand", "readName");
 634                 exit(1);
 635         }
 636 }
 637
 638 //**********************************************************************************************************************
 639 int RemoveGroupsCommand::readGroup(){
 640         try {
 641                 string thisOutputDir = outputDir;
 642                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 643                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 644
 645                 ofstream out;
 646                 m->openOutputFile(outputFileName, out);
 647
 648                 ifstream in;
 649                 m->openInputFile(groupfile, in);
 650                 string name, group;
 651
 652                 bool wroteSomething = false;
 653                 int removedCount = 0;
 654
 655                 while(!in.eof()){
 656                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 657
 658                         in >> name;                             //read from first column
 659                         in >> group;                    //read from second column
 660
 661                         //if this name is in the accnos file
 662                         if (names.count(name) == 0) {
 663                                 wroteSomething = true;
 664                                 out << name << '\t' << group << endl;
 665                         }else {  removedCount++;  }
 666
 667                         m->gobble(in);
 668                 }
 669                 in.close();
 670                 out.close();
 671
 672                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 673                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 674
 675                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 676
 677
 678                 return 0;
 679         }
 680         catch(exception& e) {
 681                 m->errorOut(e, "RemoveGroupsCommand", "readGroup");
 682                 exit(1);
 683         }
 684 }
 685 //**********************************************************************************************************************
 686 int RemoveGroupsCommand::readTax(){
 687         try {
 688                 string thisOutputDir = outputDir;
 689                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 690                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 691                 ofstream out;
 692                 m->openOutputFile(outputFileName, out);
 693
 694                 ifstream in;
 695                 m->openInputFile(taxfile, in);
 696                 string name, tax;
 697
 698                 bool wroteSomething = false;
 699                 int removedCount = 0;
 700
 701                 while(!in.eof()){
 702                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 703
 704                         in >> name;                             //read from first column
 705                         in >> tax;                      //read from second column
 706
 707                         //if this name is in the accnos file
 708                         if (names.count(name) == 0) {
 709                                 wroteSomething = true;
 710                                 out << name << '\t' << tax << endl;
 711                         }else {  removedCount++;  }
 712
 713                         m->gobble(in);
 714                 }
 715                 in.close();
 716                 out.close();
 717
 718                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 719                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 720
 721                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 722
 723                 return 0;
 724         }
 725         catch(exception& e) {
 726                 m->errorOut(e, "RemoveGroupsCommand", "readTax");
 727                 exit(1);
 728         }
 729 }
 730 //**********************************************************************************************************************
 731 void RemoveGroupsCommand::readAccnos(){
 732         try {
 733                 Groups.clear();
 734
 735                 ifstream in;
 736                 m->openInputFile(accnosfile, in);
 737                 string name;
 738
 739                 while(!in.eof()){
 740                         in >> name;
 741
 742                         Groups.push_back(name);
 743
 744                         m->gobble(in);
 745                 }
 746                 in.close();
 747
 748                 m->Groups = Groups;
 749
 750         }
 751         catch(exception& e) {
 752                 m->errorOut(e, "RemoveGroupsCommand", "readAccnos");
 753                 exit(1);
 754         }
 755 }
 756 //**********************************************************************************************************************
 757 int RemoveGroupsCommand::fillNames(){
 758         try {
 759                 vector<string> seqs = groupMap->getNamesSeqs();
 760
 761                 for (int i = 0; i < seqs.size(); i++) {
 762
 763                         if (m->control_pressed) { return 0; }
 764
 765                         string group = groupMap->getGroup(seqs[i]);
 766
 767                         if (m->inUsersGroups(group, Groups)) {
 768                                 names.insert(seqs[i]);
 769                         }
 770                 }
 771
 772                 return 0;
 773         }
 774         catch(exception& e) {
 775                 m->errorOut(e, "RemoveGroupsCommand", "fillNames");
 776                 exit(1);
 777         }
 778 }
 779
 780 //**********************************************************************************************************************
 781
 782
 783