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                         groupfile = validParameter.validFile(parameters, "group", true);
 194                         if (groupfile == "not open") { abort = true; }
 195                         else if (groupfile == "not found") {
 196                                 //if there is a current group file, use it
 197                                 groupfile = m->getGroupFile();
 198                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 199                                 else {  m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 200                         }else { m->setGroupFile(groupfile); }
 201
 202                         listfile = validParameter.validFile(parameters, "list", true);
 203                         if (listfile == "not open") { listfile = ""; abort = true; }
 204                         else if (listfile == "not found") {  listfile = "";  }
 205                         else { m->setListFile(listfile); }
 206
 207                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 208                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 209                         else if (taxfile == "not found") {  taxfile = "";  }
 210                         else { m->setTaxonomyFile(taxfile); }
 211
 212                         groups = validParameter.validFile(parameters, "groups", false);
 213                         if (groups == "not found") { groups = ""; }
 214                         else {
 215                                 m->splitAtDash(groups, Groups);
 216                                 m->setGroups(Groups);
 217                         }
 218
 219                         sharedfile = validParameter.validFile(parameters, "shared", true);
 220                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 221                         else if (sharedfile == "not found") {  sharedfile = "";  }
 222                         else { m->setSharedFile(sharedfile); }
 223
 224                         groupfile = validParameter.validFile(parameters, "group", true);
 225                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 226                         else if (groupfile == "not found") {    groupfile = ""; }
 227                         else { m->setGroupFile(groupfile); }
 228
 229                         if ((sharedfile == "") && (groupfile == "")) {
 230                                 //is there are current file available for any of these?
 231                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 232                                         //give priority to group, then shared
 233                                         groupfile = m->getGroupFile();
 234                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 235                                         else {
 236                                                 sharedfile = m->getSharedFile();
 237                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 238                                                 else {
 239                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 240                                                 }
 241                                         }
 242                                 }else {
 243                                         //give priority to shared, then group
 244                                         sharedfile = m->getSharedFile();
 245                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 246                                         else {
 247                                                 groupfile = m->getGroupFile();
 248                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 249                                                 else {
 250                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 251                                                 }
 252                                         }
 253                                 }
 254                         }
 255
 256                         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; }
 257
 258                         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; }
 259                         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; }
 260                 }
 261
 262         }
 263         catch(exception& e) {
 264                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 265                 exit(1);
 266         }
 267 }
 268 //**********************************************************************************************************************
 269
 270 int RemoveGroupsCommand::execute(){
 271         try {
 272
 273                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 274
 275                 //get groups you want to remove
 276                 if (accnosfile != "") { readAccnos(); }
 277
 278                 if (groupfile != "") {
 279                         groupMap = new GroupMap(groupfile);
 280                         groupMap->readMap();
 281
 282                         //make sure groups are valid
 283                         //takes care of user setting groupNames that are invalid or setting groups=all
 284                         SharedUtil* util = new SharedUtil();
 285                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 286                         util->setGroups(Groups, namesGroups);
 287                         delete util;
 288
 289                         //fill names with names of sequences that are from the groups we want to remove
 290                         fillNames();
 291
 292                         delete groupMap;
 293                 }
 294
 295                 if (m->control_pressed) { return 0; }
 296
 297                 //read through the correct file and output lines you want to keep
 298                 if (namefile != "")                     {               readName();             }
 299                 if (fastafile != "")            {               readFasta();    }
 300                 if (groupfile != "")            {               readGroup();    }
 301                 if (listfile != "")                     {               readList();             }
 302                 if (taxfile != "")                      {               readTax();              }
 303                 if (sharedfile != "")           {               readShared();   }
 304
 305                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 306
 307                 if (outputNames.size() != 0) {
 308                         m->mothurOutEndLine();
 309                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 310                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 311                         m->mothurOutEndLine();
 312
 313                         //set fasta file as new current fastafile
 314                         string current = "";
 315                         itTypes = outputTypes.find("fasta");
 316                         if (itTypes != outputTypes.end()) {
 317                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 318                         }
 319
 320                         itTypes = outputTypes.find("name");
 321                         if (itTypes != outputTypes.end()) {
 322                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 323                         }
 324
 325                         itTypes = outputTypes.find("group");
 326                         if (itTypes != outputTypes.end()) {
 327                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 328                         }
 329
 330                         itTypes = outputTypes.find("list");
 331                         if (itTypes != outputTypes.end()) {
 332                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 333                         }
 334
 335                         itTypes = outputTypes.find("taxonomy");
 336                         if (itTypes != outputTypes.end()) {
 337                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 338                         }
 339
 340                         itTypes = outputTypes.find("shared");
 341                         if (itTypes != outputTypes.end()) {
 342                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 343                         }
 344                 }
 345
 346                 return 0;
 347         }
 348
 349         catch(exception& e) {
 350                 m->errorOut(e, "RemoveGroupsCommand", "execute");
 351                 exit(1);
 352         }
 353 }
 354
 355 //**********************************************************************************************************************
 356 int RemoveGroupsCommand::readFasta(){
 357         try {
 358                 string thisOutputDir = outputDir;
 359                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 360                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 361
 362                 ofstream out;
 363                 m->openOutputFile(outputFileName, out);
 364
 365                 ifstream in;
 366                 m->openInputFile(fastafile, in);
 367                 string name;
 368
 369                 bool wroteSomething = false;
 370                 int removedCount = 0;
 371
 372                 while(!in.eof()){
 373                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 374
 375                         Sequence currSeq(in);
 376                         name = currSeq.getName();
 377
 378                         if (name != "") {
 379                                 //if this name is in the accnos file
 380                                 if (names.count(name) == 0) {
 381                                         wroteSomething = true;
 382                                         currSeq.printSequence(out);
 383                                 }else {
 384                                         //if you are not in the accnos file check if you are a name that needs to be changed
 385                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 386                                         if (it != uniqueToRedundant.end()) {
 387                                                 wroteSomething = true;
 388                                                 currSeq.setName(it->second);
 389                                                 currSeq.printSequence(out);
 390                                         }else { removedCount++; }
 391                                 }
 392                         }
 393                         m->gobble(in);
 394                 }
 395                 in.close();
 396                 out.close();
 397
 398                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 399                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 400
 401                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 402
 403                 return 0;
 404
 405         }
 406         catch(exception& e) {
 407                 m->errorOut(e, "RemoveGroupsCommand", "readFasta");
 408                 exit(1);
 409         }
 410 }
 411 //**********************************************************************************************************************
 412 int RemoveGroupsCommand::readShared(){
 413         try {
 414                 string thisOutputDir = outputDir;
 415                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 416
 417                 //get group names from sharedfile so we can set Groups to the groupNames we want to keep
 418                 //that way we can take advantage of the reads in inputdata and sharedRabundVector
 419                 InputData* tempInput = new InputData(sharedfile, "sharedfile");
 420                 vector<SharedRAbundVector*> lookup = tempInput->getSharedRAbundVectors();
 421
 422                 //save m->Groups
 423                 vector<string> allGroupsNames = m->getAllGroups();
 424                 vector<string> mothurOutGroups = m->getGroups();
 425
 426                 vector<string> groupsToKeep;
 427                 for (int i = 0; i < allGroupsNames.size(); i++) {
 428                         if (!m->inUsersGroups(allGroupsNames[i], m->getGroups())) {
 429                                 groupsToKeep.push_back(allGroupsNames[i]);
 430                         }
 431                 }
 432
 433                 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; }
 434
 435                 //reset read
 436                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 437                 delete tempInput;
 438                 m->setGroups(groupsToKeep);
 439                 m->clearAllGroups();
 440                 m->names.clear();
 441                 m->saveNextLabel = "";
 442                 m->printedHeaders = false;
 443                 m->currentBinLabels.clear();
 444                 m->binLabelsInFile.clear();
 445
 446                 InputData input(sharedfile, "sharedfile");
 447                 lookup = input.getSharedRAbundVectors();
 448
 449                 bool wroteSomething = false;
 450
 451                 while(lookup[0] != NULL) {
 452
 453                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 454                         ofstream out;
 455                         m->openOutputFile(outputFileName, out);
 456                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 457
 458                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 459
 460                         lookup[0]->printHeaders(out);
 461
 462                         for (int i = 0; i < lookup.size(); i++) {
 463                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 464                                 lookup[i]->print(out);
 465                                 wroteSomething = true;
 466
 467                         }
 468
 469                         //get next line to process
 470                         //prevent memory leak
 471                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 472                         lookup = input.getSharedRAbundVectors();
 473
 474                         out.close();
 475                 }
 476
 477
 478                 m->setGroups(mothurOutGroups);
 479
 480                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 481
 482                 string groupsString = "";
 483                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 484                 groupsString += Groups[Groups.size()-1];
 485
 486                 m->mothurOut("Removed groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 487
 488                 return 0;
 489
 490         }
 491         catch(exception& e) {
 492                 m->errorOut(e, "RemoveGroupsCommand", "readShared");
 493                 exit(1);
 494         }
 495 }
 496 //**********************************************************************************************************************
 497 int RemoveGroupsCommand::readList(){
 498         try {
 499                 string thisOutputDir = outputDir;
 500                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 501                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 502
 503                 ofstream out;
 504                 m->openOutputFile(outputFileName, out);
 505
 506                 ifstream in;
 507                 m->openInputFile(listfile, in);
 508
 509                 bool wroteSomething = false;
 510                 int removedCount = 0;
 511
 512                 while(!in.eof()){
 513
 514                         removedCount = 0;
 515
 516                         //read in list vector
 517                         ListVector list(in);
 518
 519                         //make a new list vector
 520                         ListVector newList;
 521                         newList.setLabel(list.getLabel());
 522
 523                         //for each bin
 524                         for (int i = 0; i < list.getNumBins(); i++) {
 525                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 526
 527                                 //parse out names that are in accnos file
 528                                 string binnames = list.get(i);
 529
 530                                 string newNames = "";
 531                                 while (binnames.find_first_of(',') != -1) {
 532                                         string name = binnames.substr(0,binnames.find_first_of(','));
 533                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 534
 535                                         //if that name is in the .accnos file, add it
 536                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 537                                         else {
 538                                                 //if you are not in the accnos file check if you are a name that needs to be changed
 539                                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 540                                                 if (it != uniqueToRedundant.end()) {
 541                                                         newNames += it->second + ",";
 542                                                 }else { removedCount++; }
 543                                         }
 544                                 }
 545
 546                                 //get last name
 547                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 548                                 else { //if you are not in the accnos file check if you are a name that needs to be changed
 549                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 550                                         if (it != uniqueToRedundant.end()) {
 551                                                 newNames += it->second + ",";
 552                                         }else { removedCount++; }
 553                                 }
 554
 555                                 //if there are names in this bin add to new list
 556                                 if (newNames != "") {
 557                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 558                                         newList.push_back(newNames);
 559                                 }
 560                         }
 561
 562                         //print new listvector
 563                         if (newList.getNumBins() != 0) {
 564                                 wroteSomething = true;
 565                                 newList.print(out);
 566                         }
 567
 568                         m->gobble(in);
 569                 }
 570                 in.close();
 571                 out.close();
 572
 573                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 574                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 575
 576                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 577
 578                 return 0;
 579
 580         }
 581         catch(exception& e) {
 582                 m->errorOut(e, "RemoveGroupsCommand", "readList");
 583                 exit(1);
 584         }
 585 }
 586 //**********************************************************************************************************************
 587 int RemoveGroupsCommand::readName(){
 588         try {
 589                 string thisOutputDir = outputDir;
 590                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 591                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 592
 593                 ofstream out;
 594                 m->openOutputFile(outputFileName, out);
 595
 596                 ifstream in;
 597                 m->openInputFile(namefile, in);
 598                 string name, firstCol, secondCol;
 599
 600                 bool wroteSomething = false;
 601                 int removedCount = 0;
 602
 603                 while(!in.eof()){
 604                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 605
 606                         in >> firstCol;         m->gobble(in);
 607                         in >> secondCol;
 608
 609                         vector<string> parsedNames;
 610                         m->splitAtComma(secondCol, parsedNames);
 611
 612                         vector<string> validSecond;  validSecond.clear();
 613                         for (int i = 0; i < parsedNames.size(); i++) {
 614                                 if (names.count(parsedNames[i]) == 0) {
 615                                         validSecond.push_back(parsedNames[i]);
 616                                 }
 617                         }
 618
 619                         removedCount += parsedNames.size()-validSecond.size();
 620
 621                         //if the name in the first column is in the set then print it and any other names in second column also in set
 622                         if (names.count(firstCol) == 0) {
 623
 624                                 wroteSomething = true;
 625
 626                                 out << firstCol << '\t';
 627
 628                                 //you know you have at least one valid second since first column is valid
 629                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 630                                 out << validSecond[validSecond.size()-1] << endl;
 631
 632                                 //make first name in set you come to first column and then add the remaining names to second column
 633                         }else {
 634
 635                                 //you want part of this row
 636                                 if (validSecond.size() != 0) {
 637
 638                                         wroteSomething = true;
 639
 640                                         out << validSecond[0] << '\t';
 641
 642                                         //you know you have at least one valid second since first column is valid
 643                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 644                                         out << validSecond[validSecond.size()-1] << endl;
 645                                         uniqueToRedundant[firstCol] = validSecond[0];
 646                                 }
 647                         }
 648
 649                         m->gobble(in);
 650                 }
 651                 in.close();
 652                 out.close();
 653
 654                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 655                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 656
 657                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 658
 659                 return 0;
 660         }
 661         catch(exception& e) {
 662                 m->errorOut(e, "RemoveGroupsCommand", "readName");
 663                 exit(1);
 664         }
 665 }
 666
 667 //**********************************************************************************************************************
 668 int RemoveGroupsCommand::readGroup(){
 669         try {
 670                 string thisOutputDir = outputDir;
 671                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 672                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 673
 674                 ofstream out;
 675                 m->openOutputFile(outputFileName, out);
 676
 677                 ifstream in;
 678                 m->openInputFile(groupfile, in);
 679                 string name, group;
 680
 681                 bool wroteSomething = false;
 682                 int removedCount = 0;
 683
 684                 while(!in.eof()){
 685                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 686
 687                         in >> name;                             //read from first column
 688                         in >> group;                    //read from second column
 689
 690                         //if this name is in the accnos file
 691                         if (names.count(name) == 0) {
 692                                 wroteSomething = true;
 693                                 out << name << '\t' << group << endl;
 694                         }else {  removedCount++;  }
 695
 696                         m->gobble(in);
 697                 }
 698                 in.close();
 699                 out.close();
 700
 701                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 702                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 703
 704                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 705
 706
 707                 return 0;
 708         }
 709         catch(exception& e) {
 710                 m->errorOut(e, "RemoveGroupsCommand", "readGroup");
 711                 exit(1);
 712         }
 713 }
 714 //**********************************************************************************************************************
 715 int RemoveGroupsCommand::readTax(){
 716         try {
 717                 string thisOutputDir = outputDir;
 718                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 719                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 720                 ofstream out;
 721                 m->openOutputFile(outputFileName, out);
 722
 723                 ifstream in;
 724                 m->openInputFile(taxfile, in);
 725                 string name, tax;
 726
 727                 bool wroteSomething = false;
 728                 int removedCount = 0;
 729
 730                 while(!in.eof()){
 731                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 732
 733                         in >> name;                             //read from first column
 734                         in >> tax;                      //read from second column
 735
 736                         //if this name is in the accnos file
 737                         if (names.count(name) == 0) {
 738                                 wroteSomething = true;
 739                                 out << name << '\t' << tax << endl;
 740                         }else {  //if you are not in the accnos file check if you are a name that needs to be changed
 741                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 742                                 if (it != uniqueToRedundant.end()) {
 743                                         wroteSomething = true;
 744                                         out << it->second << '\t' << tax << endl;
 745                                 }else { removedCount++; }  }
 746
 747                         m->gobble(in);
 748                 }
 749                 in.close();
 750                 out.close();
 751
 752                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 753                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 754
 755                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 756
 757                 return 0;
 758         }
 759         catch(exception& e) {
 760                 m->errorOut(e, "RemoveGroupsCommand", "readTax");
 761                 exit(1);
 762         }
 763 }
 764 //**********************************************************************************************************************
 765 void RemoveGroupsCommand::readAccnos(){
 766         try {
 767                 Groups.clear();
 768
 769                 ifstream in;
 770                 m->openInputFile(accnosfile, in);
 771                 string name;
 772
 773                 while(!in.eof()){
 774                         in >> name;
 775
 776                         Groups.push_back(name);
 777
 778                         m->gobble(in);
 779                 }
 780                 in.close();
 781
 782                 m->setGroups(Groups);
 783
 784         }
 785         catch(exception& e) {
 786                 m->errorOut(e, "RemoveGroupsCommand", "readAccnos");
 787                 exit(1);
 788         }
 789 }
 790 //**********************************************************************************************************************
 791 int RemoveGroupsCommand::fillNames(){
 792         try {
 793                 vector<string> seqs = groupMap->getNamesSeqs();
 794
 795                 for (int i = 0; i < seqs.size(); i++) {
 796
 797                         if (m->control_pressed) { return 0; }
 798
 799                         string group = groupMap->getGroup(seqs[i]);
 800
 801                         if (m->inUsersGroups(group, Groups)) {
 802                                 names.insert(seqs[i]);
 803                         }
 804                 }
 805
 806                 return 0;
 807         }
 808         catch(exception& e) {
 809                 m->errorOut(e, "RemoveGroupsCommand", "fillNames");
 810                 exit(1);
 811         }
 812 }
 813
 814 //**********************************************************************************************************************
 815
 816
 817