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 pdesign("design", "InputTypes", "", "", "none", "sharedGroup", "FNGLT",false,false); parameters.push_back(pdesign);
  24                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(plist);
  25                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(ptaxonomy);
  26                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  27                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "RemoveGroupsCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string RemoveGroupsCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 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, design or sharedfile.\n";
  45                 helpString += "It outputs a file containing the sequences NOT in the those specified groups, or with a sharedfile eliminates the groups you selected.\n";
  46                 helpString += "The remove.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared, design and groups. The group parameter is required, unless you have a current group file or are using a sharedfile.\n";
  47                 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";
  48                 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";
  49                 helpString += "The remove.groups command should be in the following format: remove.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
  50                 helpString += "Example remove.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
  51                 helpString += "or remove.groups(groups=pasture, fasta=amazon.fasta, amazon.groups).\n";
  52                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  53                 return helpString;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "RemoveGroupsCommand", "getHelpString");
  57                 exit(1);
  58         }
  59 }
  60
  61 //**********************************************************************************************************************
  62 RemoveGroupsCommand::RemoveGroupsCommand(){
  63         try {
  64                 abort = true; calledHelp = true;
  65                 setParameters();
  66                 vector<string> tempOutNames;
  67                 outputTypes["fasta"] = tempOutNames;
  68                 outputTypes["taxonomy"] = tempOutNames;
  69                 outputTypes["name"] = tempOutNames;
  70                 outputTypes["group"] = tempOutNames;
  71                 outputTypes["list"] = tempOutNames;
  72                 outputTypes["shared"] = tempOutNames;
  73         outputTypes["design"] = tempOutNames;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81 RemoveGroupsCommand::RemoveGroupsCommand(string option)  {
  82         try {
  83                 abort = false; calledHelp = false;
  84
  85                 //allow user to run help
  86                 if(option == "help") { help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string,string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string,string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["fasta"] = tempOutNames;
 106                         outputTypes["taxonomy"] = tempOutNames;
 107                         outputTypes["name"] = tempOutNames;
 108                         outputTypes["group"] = tempOutNames;
 109                         outputTypes["list"] = tempOutNames;
 110                         outputTypes["shared"] = tempOutNames;
 111             outputTypes["design"] = tempOutNames;
 112
 113
 114                         //if the user changes the output directory command factory will send this info to us in the output parameter
 115                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 116
 117                         //if the user changes the input directory command factory will send this info to us in the output parameter
 118                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 119                         if (inputDir == "not found"){   inputDir = "";          }
 120                         else {
 121                                 string path;
 122                                 it = parameters.find("fasta");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 128                                 }
 129
 130                                 it = parameters.find("accnos");
 131                                 //user has given a template file
 132                                 if(it != parameters.end()){
 133                                         path = m->hasPath(it->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 136                                 }
 137
 138                                 it = parameters.find("list");
 139                                 //user has given a template file
 140                                 if(it != parameters.end()){
 141                                         path = m->hasPath(it->second);
 142                                         //if the user has not given a path then, add inputdir. else leave path alone.
 143                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 144                                 }
 145
 146                                 it = parameters.find("name");
 147                                 //user has given a template file
 148                                 if(it != parameters.end()){
 149                                         path = m->hasPath(it->second);
 150                                         //if the user has not given a path then, add inputdir. else leave path alone.
 151                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 152                                 }
 153
 154                                 it = parameters.find("group");
 155                                 //user has given a template file
 156                                 if(it != parameters.end()){
 157                                         path = m->hasPath(it->second);
 158                                         //if the user has not given a path then, add inputdir. else leave path alone.
 159                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 160                                 }
 161
 162                                 it = parameters.find("taxonomy");
 163                                 //user has given a template file
 164                                 if(it != parameters.end()){
 165                                         path = m->hasPath(it->second);
 166                                         //if the user has not given a path then, add inputdir. else leave path alone.
 167                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 168                                 }
 169
 170                                 it = parameters.find("shared");
 171                                 //user has given a template file
 172                                 if(it != parameters.end()){
 173                                         path = m->hasPath(it->second);
 174                                         //if the user has not given a path then, add inputdir. else leave path alone.
 175                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 176                                 }
 177
 178                 it = parameters.find("design");
 179                                 //user has given a template file
 180                                 if(it != parameters.end()){
 181                                         path = m->hasPath(it->second);
 182                                         //if the user has not given a path then, add inputdir. else leave path alone.
 183                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 184                                 }
 185                         }
 186
 187
 188                         //check for required parameters
 189                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 190                         if (accnosfile == "not open") { accnosfile = ""; abort = true; }
 191                         else if (accnosfile == "not found") {  accnosfile = ""; }
 192                         else { m->setAccnosFile(accnosfile); }
 193
 194                         fastafile = validParameter.validFile(parameters, "fasta", true);
 195                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 196                         else if (fastafile == "not found") {  fastafile = "";  }
 197                         else { m->setFastaFile(fastafile); }
 198
 199                         namefile = validParameter.validFile(parameters, "name", true);
 200                         if (namefile == "not open") { namefile = ""; abort = true; }
 201                         else if (namefile == "not found") {  namefile = "";  }
 202                         else { m->setNameFile(namefile); }
 203
 204                         groupfile = validParameter.validFile(parameters, "group", true);
 205                         if (groupfile == "not open") { groupfile = "";  abort = true; }
 206                         else if (groupfile == "not found") {    groupfile = "";         }
 207                         else { m->setGroupFile(groupfile); }
 208
 209                         listfile = validParameter.validFile(parameters, "list", true);
 210                         if (listfile == "not open") { listfile = ""; abort = true; }
 211                         else if (listfile == "not found") {  listfile = "";  }
 212                         else { m->setListFile(listfile); }
 213
 214                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 215                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 216                         else if (taxfile == "not found") {  taxfile = "";  }
 217                         else { m->setTaxonomyFile(taxfile); }
 218
 219             designfile = validParameter.validFile(parameters, "design", true);
 220                         if (designfile == "not open") { designfile = ""; abort = true; }
 221                         else if (designfile == "not found") {  designfile = "";  }
 222                         else { m->setDesignFile(designfile); }
 223
 224                         groups = validParameter.validFile(parameters, "groups", false);
 225                         if (groups == "not found") { groups = ""; }
 226                         else {
 227                                 m->splitAtDash(groups, Groups);
 228                                 m->setGroups(Groups);
 229                         }
 230
 231                         sharedfile = validParameter.validFile(parameters, "shared", true);
 232                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 233                         else if (sharedfile == "not found") {  sharedfile = "";  }
 234                         else { m->setSharedFile(sharedfile); }
 235
 236                         groupfile = validParameter.validFile(parameters, "group", true);
 237                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 238                         else if (groupfile == "not found") {    groupfile = ""; }
 239                         else { m->setGroupFile(groupfile); }
 240
 241                         if ((sharedfile == "") && (groupfile == "") && (designfile == "")) {
 242                                 //is there are current file available for any of these?
 243                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 244                                         //give priority to group, then shared
 245                                         groupfile = m->getGroupFile();
 246                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 247                                         else {
 248                                                 sharedfile = m->getSharedFile();
 249                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 250                                                 else {
 251                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 252                                                 }
 253                                         }
 254                                 }else {
 255                                         //give priority to shared, then group
 256                                         sharedfile = m->getSharedFile();
 257                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 258                                         else {
 259                                                 groupfile = m->getGroupFile();
 260                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 261                                                 else {
 262                                                         designfile = m->getDesignFile();
 263                             if (designfile != "") { m->mothurOut("Using " + designfile + " as input file for the design parameter."); m->mothurOutEndLine(); }
 264                             else {
 265                                 m->mothurOut("You have no current groupfile or sharedfile or designfile and one is required."); m->mothurOutEndLine(); abort = true;
 266                             }
 267                                                 }
 268                                         }
 269                                 }
 270                         }
 271
 272                         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; }
 273
 274                         if ((fastafile == "") && (namefile == "") && (groupfile == "")  && (sharedfile == "") && (designfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared, design or list."); m->mothurOutEndLine(); abort = true; }
 275                         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; }
 276
 277                         if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 278                                 vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 279                                 parser.getNameFile(files);
 280                         }
 281
 282                 }
 283
 284         }
 285         catch(exception& e) {
 286                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 287                 exit(1);
 288         }
 289 }
 290 //**********************************************************************************************************************
 291
 292 int RemoveGroupsCommand::execute(){
 293         try {
 294
 295                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 296
 297                 //get groups you want to remove
 298                 if (accnosfile != "") { readAccnos(); }
 299
 300                 if (groupfile != "") {
 301                         groupMap = new GroupMap(groupfile);
 302                         groupMap->readMap();
 303
 304                         //make sure groups are valid
 305                         //takes care of user setting groupNames that are invalid or setting groups=all
 306                         SharedUtil* util = new SharedUtil();
 307                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 308                         util->setGroups(Groups, namesGroups);
 309                         delete util;
 310
 311                         //fill names with names of sequences that are from the groups we want to remove
 312                         fillNames();
 313
 314                         delete groupMap;
 315                 }
 316
 317                 if (m->control_pressed) { return 0; }
 318
 319                 //read through the correct file and output lines you want to keep
 320                 if (namefile != "")                     {               readName();             }
 321                 if (fastafile != "")            {               readFasta();    }
 322                 if (groupfile != "")            {               readGroup();    }
 323                 if (listfile != "")                     {               readList();             }
 324                 if (taxfile != "")                      {               readTax();              }
 325                 if (sharedfile != "")           {               readShared();   }
 326         if (designfile != "")           {               readDesign();   }
 327
 328                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 329
 330                 if (outputNames.size() != 0) {
 331                         m->mothurOutEndLine();
 332                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 333                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 334                         m->mothurOutEndLine();
 335
 336                         //set fasta file as new current fastafile
 337                         string current = "";
 338                         itTypes = outputTypes.find("fasta");
 339                         if (itTypes != outputTypes.end()) {
 340                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 341                         }
 342
 343                         itTypes = outputTypes.find("name");
 344                         if (itTypes != outputTypes.end()) {
 345                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 346                         }
 347
 348                         itTypes = outputTypes.find("group");
 349                         if (itTypes != outputTypes.end()) {
 350                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 351                         }
 352
 353                         itTypes = outputTypes.find("list");
 354                         if (itTypes != outputTypes.end()) {
 355                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 356                         }
 357
 358                         itTypes = outputTypes.find("taxonomy");
 359                         if (itTypes != outputTypes.end()) {
 360                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 361                         }
 362
 363                         itTypes = outputTypes.find("shared");
 364                         if (itTypes != outputTypes.end()) {
 365                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 366                         }
 367
 368             itTypes = outputTypes.find("design");
 369                         if (itTypes != outputTypes.end()) {
 370                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setDesignFile(current); }
 371                         }
 372                 }
 373
 374                 return 0;
 375         }
 376
 377         catch(exception& e) {
 378                 m->errorOut(e, "RemoveGroupsCommand", "execute");
 379                 exit(1);
 380         }
 381 }
 382
 383 //**********************************************************************************************************************
 384 int RemoveGroupsCommand::readFasta(){
 385         try {
 386                 string thisOutputDir = outputDir;
 387                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 388                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 389
 390                 ofstream out;
 391                 m->openOutputFile(outputFileName, out);
 392
 393                 ifstream in;
 394                 m->openInputFile(fastafile, in);
 395                 string name;
 396
 397                 bool wroteSomething = false;
 398                 int removedCount = 0;
 399
 400                 while(!in.eof()){
 401                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 402
 403                         Sequence currSeq(in);
 404                         name = currSeq.getName();
 405
 406                         if (name != "") {
 407                                 //if this name is in the accnos file
 408                                 if (names.count(name) == 0) {
 409                                         wroteSomething = true;
 410                                         currSeq.printSequence(out);
 411                                 }else {
 412                                         //if you are not in the accnos file check if you are a name that needs to be changed
 413                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 414                                         if (it != uniqueToRedundant.end()) {
 415                                                 wroteSomething = true;
 416                                                 currSeq.setName(it->second);
 417                                                 currSeq.printSequence(out);
 418                                         }else { removedCount++; }
 419                                 }
 420                         }
 421                         m->gobble(in);
 422                 }
 423                 in.close();
 424                 out.close();
 425
 426                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 427                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 428
 429                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 430
 431                 return 0;
 432
 433         }
 434         catch(exception& e) {
 435                 m->errorOut(e, "RemoveGroupsCommand", "readFasta");
 436                 exit(1);
 437         }
 438 }
 439 //**********************************************************************************************************************
 440 int RemoveGroupsCommand::readShared(){
 441         try {
 442                 string thisOutputDir = outputDir;
 443                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 444
 445                 //get group names from sharedfile so we can set Groups to the groupNames we want to keep
 446                 //that way we can take advantage of the reads in inputdata and sharedRabundVector
 447                 InputData* tempInput = new InputData(sharedfile, "sharedfile");
 448                 vector<SharedRAbundVector*> lookup = tempInput->getSharedRAbundVectors();
 449
 450                 //save m->Groups
 451                 vector<string> allGroupsNames = m->getAllGroups();
 452                 vector<string> mothurOutGroups = m->getGroups();
 453
 454                 vector<string> groupsToKeep;
 455                 for (int i = 0; i < allGroupsNames.size(); i++) {
 456                         if (!m->inUsersGroups(allGroupsNames[i], m->getGroups())) {
 457                                 groupsToKeep.push_back(allGroupsNames[i]);
 458                         }
 459                 }
 460
 461                 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; }
 462
 463                 //reset read
 464                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 465                 delete tempInput;
 466                 m->setGroups(groupsToKeep);
 467                 m->clearAllGroups();
 468                 m->saveNextLabel = "";
 469                 m->printedHeaders = false;
 470                 m->currentBinLabels.clear();
 471                 m->binLabelsInFile.clear();
 472
 473                 InputData input(sharedfile, "sharedfile");
 474                 lookup = input.getSharedRAbundVectors();
 475
 476                 bool wroteSomething = false;
 477
 478                 while(lookup[0] != NULL) {
 479
 480                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 481                         ofstream out;
 482                         m->openOutputFile(outputFileName, out);
 483                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 484
 485                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 486
 487                         lookup[0]->printHeaders(out);
 488
 489                         for (int i = 0; i < lookup.size(); i++) {
 490                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 491                                 lookup[i]->print(out);
 492                                 wroteSomething = true;
 493
 494                         }
 495
 496                         //get next line to process
 497                         //prevent memory leak
 498                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 499                         lookup = input.getSharedRAbundVectors();
 500
 501                         out.close();
 502                 }
 503
 504
 505                 m->setGroups(mothurOutGroups);
 506
 507                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 508
 509                 string groupsString = "";
 510                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 511                 groupsString += Groups[Groups.size()-1];
 512
 513                 m->mothurOut("Removed groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 514
 515                 return 0;
 516
 517         }
 518         catch(exception& e) {
 519                 m->errorOut(e, "RemoveGroupsCommand", "readShared");
 520                 exit(1);
 521         }
 522 }
 523 //**********************************************************************************************************************
 524 int RemoveGroupsCommand::readList(){
 525         try {
 526                 string thisOutputDir = outputDir;
 527                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 528                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 529
 530                 ofstream out;
 531                 m->openOutputFile(outputFileName, out);
 532
 533                 ifstream in;
 534                 m->openInputFile(listfile, in);
 535
 536                 bool wroteSomething = false;
 537                 int removedCount = 0;
 538
 539                 while(!in.eof()){
 540
 541                         removedCount = 0;
 542
 543                         //read in list vector
 544                         ListVector list(in);
 545
 546                         //make a new list vector
 547                         ListVector newList;
 548                         newList.setLabel(list.getLabel());
 549
 550                         //for each bin
 551                         for (int i = 0; i < list.getNumBins(); i++) {
 552                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 553
 554                                 //parse out names that are in accnos file
 555                                 string binnames = list.get(i);
 556
 557                                 string newNames = "";
 558                                 while (binnames.find_first_of(',') != -1) {
 559                                         string name = binnames.substr(0,binnames.find_first_of(','));
 560                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 561
 562                                         //if that name is in the .accnos file, add it
 563                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 564                                         else {
 565                                                 //if you are not in the accnos file check if you are a name that needs to be changed
 566                                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 567                                                 if (it != uniqueToRedundant.end()) {
 568                                                         newNames += it->second + ",";
 569                                                 }else { removedCount++; }
 570                                         }
 571                                 }
 572
 573                                 //get last name
 574                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 575                                 else { //if you are not in the accnos file check if you are a name that needs to be changed
 576                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 577                                         if (it != uniqueToRedundant.end()) {
 578                                                 newNames += it->second + ",";
 579                                         }else { removedCount++; }
 580                                 }
 581
 582                                 //if there are names in this bin add to new list
 583                                 if (newNames != "") {
 584                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 585                                         newList.push_back(newNames);
 586                                 }
 587                         }
 588
 589                         //print new listvector
 590                         if (newList.getNumBins() != 0) {
 591                                 wroteSomething = true;
 592                                 newList.print(out);
 593                         }
 594
 595                         m->gobble(in);
 596                 }
 597                 in.close();
 598                 out.close();
 599
 600                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 601                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 602
 603                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 604
 605                 return 0;
 606
 607         }
 608         catch(exception& e) {
 609                 m->errorOut(e, "RemoveGroupsCommand", "readList");
 610                 exit(1);
 611         }
 612 }
 613 //**********************************************************************************************************************
 614 int RemoveGroupsCommand::readName(){
 615         try {
 616                 string thisOutputDir = outputDir;
 617                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 618                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 619
 620                 ofstream out;
 621                 m->openOutputFile(outputFileName, out);
 622
 623                 ifstream in;
 624                 m->openInputFile(namefile, in);
 625                 string name, firstCol, secondCol;
 626
 627                 bool wroteSomething = false;
 628                 int removedCount = 0;
 629
 630                 while(!in.eof()){
 631                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 632
 633                         in >> firstCol;         m->gobble(in);
 634                         in >> secondCol;
 635
 636                         vector<string> parsedNames;
 637                         m->splitAtComma(secondCol, parsedNames);
 638
 639                         vector<string> validSecond;  validSecond.clear();
 640                         for (int i = 0; i < parsedNames.size(); i++) {
 641                                 if (names.count(parsedNames[i]) == 0) {
 642                                         validSecond.push_back(parsedNames[i]);
 643                                 }
 644                         }
 645
 646                         removedCount += parsedNames.size()-validSecond.size();
 647
 648                         //if the name in the first column is in the set then print it and any other names in second column also in set
 649                         if (names.count(firstCol) == 0) {
 650
 651                                 wroteSomething = true;
 652
 653                                 out << firstCol << '\t';
 654
 655                                 //you know you have at least one valid second since first column is valid
 656                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 657                                 out << validSecond[validSecond.size()-1] << endl;
 658
 659                                 //make first name in set you come to first column and then add the remaining names to second column
 660                         }else {
 661
 662                                 //you want part of this row
 663                                 if (validSecond.size() != 0) {
 664
 665                                         wroteSomething = true;
 666
 667                                         out << validSecond[0] << '\t';
 668
 669                                         //you know you have at least one valid second since first column is valid
 670                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 671                                         out << validSecond[validSecond.size()-1] << endl;
 672                                         uniqueToRedundant[firstCol] = validSecond[0];
 673                                 }
 674                         }
 675
 676                         m->gobble(in);
 677                 }
 678                 in.close();
 679                 out.close();
 680
 681                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 682                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 683
 684                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 685
 686                 return 0;
 687         }
 688         catch(exception& e) {
 689                 m->errorOut(e, "RemoveGroupsCommand", "readName");
 690                 exit(1);
 691         }
 692 }
 693
 694 //**********************************************************************************************************************
 695 int RemoveGroupsCommand::readGroup(){
 696         try {
 697                 string thisOutputDir = outputDir;
 698                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 699                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 700
 701                 ofstream out;
 702                 m->openOutputFile(outputFileName, out);
 703
 704                 ifstream in;
 705                 m->openInputFile(groupfile, in);
 706                 string name, group;
 707
 708                 bool wroteSomething = false;
 709                 int removedCount = 0;
 710
 711                 while(!in.eof()){
 712                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 713
 714                         in >> name;                             //read from first column
 715                         in >> group;                    //read from second column
 716
 717                         //if this name is in the accnos file
 718                         if (names.count(name) == 0) {
 719                                 wroteSomething = true;
 720                                 out << name << '\t' << group << endl;
 721                         }else {  removedCount++;  }
 722
 723                         m->gobble(in);
 724                 }
 725                 in.close();
 726                 out.close();
 727
 728                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 729                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 730
 731                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 732
 733
 734                 return 0;
 735         }
 736         catch(exception& e) {
 737                 m->errorOut(e, "RemoveGroupsCommand", "readGroup");
 738                 exit(1);
 739         }
 740 }
 741 //**********************************************************************************************************************
 742 int RemoveGroupsCommand::readDesign(){
 743         try {
 744                 string thisOutputDir = outputDir;
 745                 if (outputDir == "") {  thisOutputDir += m->hasPath(designfile);  }
 746                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(designfile)) + "pick" + m->getExtension(designfile);
 747
 748                 ofstream out;
 749                 m->openOutputFile(outputFileName, out);
 750
 751                 ifstream in;
 752                 m->openInputFile(designfile, in);
 753                 string name, group;
 754
 755                 bool wroteSomething = false;
 756                 int removedCount = 0;
 757
 758                 while(!in.eof()){
 759                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 760
 761                         in >> name;                             //read from first column
 762                         in >> group;                    //read from second column
 763
 764                         //if this name is in the accnos file
 765                         if (!(m->inUsersGroups(name, Groups))) {
 766                                 wroteSomething = true;
 767                                 out << name << '\t' << group << endl;
 768                         }else {  removedCount++;  }
 769
 770                         m->gobble(in);
 771                 }
 772                 in.close();
 773                 out.close();
 774
 775                 if (wroteSomething == false) {  m->mothurOut("Your file contains only groups from the groups you wish to remove."); m->mothurOutEndLine();  }
 776                 outputTypes["design"].push_back(outputFileName); outputNames.push_back(outputFileName);
 777
 778                 m->mothurOut("Removed " + toString(removedCount) + " groups from your design file."); m->mothurOutEndLine();
 779
 780
 781                 return 0;
 782         }
 783         catch(exception& e) {
 784                 m->errorOut(e, "RemoveGroupsCommand", "readDesign");
 785                 exit(1);
 786         }
 787 }
 788
 789 //**********************************************************************************************************************
 790 int RemoveGroupsCommand::readTax(){
 791         try {
 792                 string thisOutputDir = outputDir;
 793                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 794                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 795                 ofstream out;
 796                 m->openOutputFile(outputFileName, out);
 797
 798                 ifstream in;
 799                 m->openInputFile(taxfile, in);
 800                 string name, tax;
 801
 802                 bool wroteSomething = false;
 803                 int removedCount = 0;
 804
 805                 while(!in.eof()){
 806                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 807
 808                         in >> name;                             //read from first column
 809                         in >> tax;                      //read from second column
 810
 811                         //if this name is in the accnos file
 812                         if (names.count(name) == 0) {
 813                                 wroteSomething = true;
 814                                 out << name << '\t' << tax << endl;
 815                         }else {  //if you are not in the accnos file check if you are a name that needs to be changed
 816                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 817                                 if (it != uniqueToRedundant.end()) {
 818                                         wroteSomething = true;
 819                                         out << it->second << '\t' << tax << endl;
 820                                 }else { removedCount++; }  }
 821
 822                         m->gobble(in);
 823                 }
 824                 in.close();
 825                 out.close();
 826
 827                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 828                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 829
 830                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 831
 832                 return 0;
 833         }
 834         catch(exception& e) {
 835                 m->errorOut(e, "RemoveGroupsCommand", "readTax");
 836                 exit(1);
 837         }
 838 }
 839 //**********************************************************************************************************************
 840 void RemoveGroupsCommand::readAccnos(){
 841         try {
 842                 Groups.clear();
 843
 844                 ifstream in;
 845                 m->openInputFile(accnosfile, in);
 846                 string name;
 847
 848                 while(!in.eof()){
 849                         in >> name;
 850
 851                         Groups.push_back(name);
 852
 853                         m->gobble(in);
 854                 }
 855                 in.close();
 856
 857                 m->setGroups(Groups);
 858
 859         }
 860         catch(exception& e) {
 861                 m->errorOut(e, "RemoveGroupsCommand", "readAccnos");
 862                 exit(1);
 863         }
 864 }
 865 //**********************************************************************************************************************
 866 int RemoveGroupsCommand::fillNames(){
 867         try {
 868                 vector<string> seqs = groupMap->getNamesSeqs();
 869
 870                 for (int i = 0; i < seqs.size(); i++) {
 871
 872                         if (m->control_pressed) { return 0; }
 873
 874                         string group = groupMap->getGroup(seqs[i]);
 875
 876                         if (m->inUsersGroups(group, Groups)) {
 877                                 names.insert(seqs[i]);
 878                         }
 879                 }
 880
 881                 return 0;
 882         }
 883         catch(exception& e) {
 884                 m->errorOut(e, "RemoveGroupsCommand", "fillNames");
 885                 exit(1);
 886         }
 887 }
 888
 889 //**********************************************************************************************************************
 890
 891
 892