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