removegroupscommand.cpp

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