getgroupscommand.cpp

   1 /*
   2  *  getgroupscommand.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 "getgroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> GetGroupsCommand::setParameters(){
  18         try {
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(pfasta);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "sharedGroup", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  22                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "sharedGroup", "FNGLT",false,false); parameters.push_back(pgroup);
  23         CommandParameter pdesign("design", "InputTypes", "", "", "none", "sharedGroup", "FNGLT",false,false); parameters.push_back(pdesign);
  24                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(plist);
  25                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(ptaxonomy);
  26                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  27                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "GetGroupsCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string GetGroupsCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 helpString += "The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy, design or shared file.\n";
  45                 helpString += "It outputs a file containing the sequences in the those specified groups, or a sharedfile containing only those groups.\n";
  46                 helpString += "The get.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared, design and groups. The group parameter is required, unless you have a current group file, or are using a shared file.\n";
  47                 helpString += "You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n";
  48                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like.  You can separate group names with dashes.\n";
  49                 helpString += "The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
  50                 helpString += "Example get.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
  51                 helpString += "or get.groups(groups=pasture, fasta=amazon.fasta, group=amazon.groups).\n";
  52                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  53                 return helpString;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "GetGroupsCommand", "getHelpString");
  57                 exit(1);
  58         }
  59 }
  60 //**********************************************************************************************************************
  61 GetGroupsCommand::GetGroupsCommand(){
  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         outputTypes["design"] = tempOutNames;
  73         }
  74         catch(exception& e) {
  75                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
  76                 exit(1);
  77         }
  78 }
  79 //**********************************************************************************************************************
  80 GetGroupsCommand::GetGroupsCommand(string option)  {
  81         try {
  82                 abort = false; calledHelp = false;
  83
  84                 //allow user to run help
  85                 if(option == "help") { help(); abort = true; calledHelp = true; }
  86                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  87
  88                 else {
  89                         vector<string> myArray = setParameters();
  90
  91                         OptionParser parser(option);
  92                         map<string,string> parameters = parser.getParameters();
  93
  94                         ValidParameters validParameter;
  95                         map<string,string>::iterator it;
  96
  97                         //check to make sure all parameters are valid for command
  98                         for (it = parameters.begin(); it != parameters.end(); it++) {
  99                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 100                         }
 101
 102                         //initialize outputTypes
 103                         vector<string> tempOutNames;
 104                         outputTypes["fasta"] = tempOutNames;
 105                         outputTypes["taxonomy"] = tempOutNames;
 106                         outputTypes["name"] = tempOutNames;
 107                         outputTypes["group"] = tempOutNames;
 108                         outputTypes["list"] = tempOutNames;
 109                         outputTypes["shared"] = tempOutNames;
 110             outputTypes["design"] = tempOutNames;
 111
 112
 113                         //if the user changes the output directory command factory will send this info to us in the output parameter
 114                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 115
 116                         //if the user changes the input directory command factory will send this info to us in the output parameter
 117                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 118                         if (inputDir == "not found"){   inputDir = "";          }
 119                         else {
 120                                 string path;
 121                                 it = parameters.find("fasta");
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 127                                 }
 128
 129                                 it = parameters.find("accnos");
 130                                 //user has given a template file
 131                                 if(it != parameters.end()){
 132                                         path = m->hasPath(it->second);
 133                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 135                                 }
 136
 137                                 it = parameters.find("list");
 138                                 //user has given a template file
 139                                 if(it != parameters.end()){
 140                                         path = m->hasPath(it->second);
 141                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 143                                 }
 144
 145                                 it = parameters.find("name");
 146                                 //user has given a template file
 147                                 if(it != parameters.end()){
 148                                         path = m->hasPath(it->second);
 149                                         //if the user has not given a path then, add inputdir. else leave path alone.
 150                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 151                                 }
 152
 153                                 it = parameters.find("group");
 154                                 //user has given a template file
 155                                 if(it != parameters.end()){
 156                                         path = m->hasPath(it->second);
 157                                         //if the user has not given a path then, add inputdir. else leave path alone.
 158                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 159                                 }
 160
 161                                 it = parameters.find("taxonomy");
 162                                 //user has given a template file
 163                                 if(it != parameters.end()){
 164                                         path = m->hasPath(it->second);
 165                                         //if the user has not given a path then, add inputdir. else leave path alone.
 166                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 167                                 }
 168
 169                                 it = parameters.find("shared");
 170                                 //user has given a template file
 171                                 if(it != parameters.end()){
 172                                         path = m->hasPath(it->second);
 173                                         //if the user has not given a path then, add inputdir. else leave path alone.
 174                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 175                                 }
 176
 177                 it = parameters.find("design");
 178                                 //user has given a template file
 179                                 if(it != parameters.end()){
 180                                         path = m->hasPath(it->second);
 181                                         //if the user has not given a path then, add inputdir. else leave path alone.
 182                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 183                                 }
 184                         }
 185
 186
 187                         //check for required parameters
 188                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 189                         if (accnosfile == "not open") { abort = true; }
 190                         else if (accnosfile == "not found") {  accnosfile = ""; }
 191                         else { m->setAccnosFile(accnosfile); }
 192
 193                         fastafile = validParameter.validFile(parameters, "fasta", true);
 194                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 195                         else if (fastafile == "not found") {  fastafile = "";  }
 196                         else { m->setFastaFile(fastafile); }
 197
 198                         namefile = validParameter.validFile(parameters, "name", true);
 199                         if (namefile == "not open") { namefile = ""; abort = true; }
 200                         else if (namefile == "not found") {  namefile = "";  }
 201                         else { m->setNameFile(namefile); }
 202
 203                         groupfile = validParameter.validFile(parameters, "group", true);
 204                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 205                         else if (groupfile == "not found") {  groupfile = "";                   }
 206                         else { m->setGroupFile(groupfile); }
 207
 208                         listfile = validParameter.validFile(parameters, "list", true);
 209                         if (listfile == "not open") { abort = true; }
 210                         else if (listfile == "not found") {  listfile = "";  }
 211                         else { m->setListFile(listfile); }
 212
 213                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 214                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 215                         else if (taxfile == "not found") {  taxfile = "";  }
 216                         else { m->setTaxonomyFile(taxfile); }
 217
 218                         groups = validParameter.validFile(parameters, "groups", false);
 219                         if (groups == "not found") { groups = ""; }
 220                         else {
 221                                 m->splitAtDash(groups, Groups);
 222                                 m->setGroups(Groups);
 223                         }
 224
 225                         sharedfile = validParameter.validFile(parameters, "shared", true);
 226                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 227                         else if (sharedfile == "not found") {  sharedfile = "";  }
 228                         else { m->setSharedFile(sharedfile); }
 229
 230                         groupfile = validParameter.validFile(parameters, "group", true);
 231                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 232                         else if (groupfile == "not found") {    groupfile = ""; }
 233                         else { m->setGroupFile(groupfile); }
 234
 235             designfile = validParameter.validFile(parameters, "design", true);
 236                         if (designfile == "not open") { designfile = ""; abort = true; }
 237                         else if (designfile == "not found") {   designfile = "";        }
 238                         else { m->setDesignFile(designfile); }
 239
 240                         if ((sharedfile == "") && (groupfile == "") && (designfile == "")) {
 241                                 //is there are current file available for any of these?
 242                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 243                                         //give priority to group, then shared
 244                                         groupfile = m->getGroupFile();
 245                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 246                                         else {
 247                                                 sharedfile = m->getSharedFile();
 248                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared 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                                 }else {
 254                                         //give priority to shared, then group
 255                                         sharedfile = m->getSharedFile();
 256                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 257                                         else {
 258                                                 groupfile = m->getGroupFile();
 259                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 260                                                 else {
 261                                                         designfile = m->getDesignFile();
 262                             if (designfile != "") { m->mothurOut("Using " + designfile + " as input file for the design parameter."); m->mothurOutEndLine(); }
 263                             else {
 264                                 m->mothurOut("You have no current groupfile or sharedfile or designfile and one is required."); m->mothurOutEndLine(); abort = true;
 265                             }
 266                                                 }
 267                                         }
 268                                 }
 269                         }
 270
 271                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 272
 273                         if ((fastafile == "") && (namefile == "") && (groupfile == "")  && (designfile == "") && (sharedfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared, design or list."); m->mothurOutEndLine(); abort = true; }
 274                         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; }
 275
 276                         if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 277                                 vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 278                                 parser.getNameFile(files);
 279                         }
 280                 }
 281
 282         }
 283         catch(exception& e) {
 284                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
 285                 exit(1);
 286         }
 287 }
 288 //**********************************************************************************************************************
 289
 290 int GetGroupsCommand::execute(){
 291         try {
 292
 293                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 294
 295                 //get groups you want to remove
 296                 if (accnosfile != "") { readAccnos(); }
 297
 298                 if (groupfile != "") {
 299                         groupMap = new GroupMap(groupfile);
 300                         groupMap->readMap();
 301
 302                         //make sure groups are valid
 303                         //takes care of user setting groupNames that are invalid or setting groups=all
 304                         SharedUtil* util = new SharedUtil();
 305                         vector<string> gNamesOfGroups = groupMap->getNamesOfGroups();
 306                         util->setGroups(Groups, gNamesOfGroups);
 307                         groupMap->setNamesOfGroups(gNamesOfGroups);
 308                         delete util;
 309
 310                         //fill names with names of sequences that are from the groups we want to remove
 311                         fillNames();
 312
 313                         delete groupMap;
 314                 }
 315
 316                 if (m->control_pressed) { return 0; }
 317
 318                 //read through the correct file and output lines you want to keep
 319                 if (namefile != "")                     {               readName();             }
 320                 if (fastafile != "")            {               readFasta();    }
 321                 if (groupfile != "")            {               readGroup();    }
 322                 if (listfile != "")                     {               readList();             }
 323                 if (taxfile != "")                      {               readTax();              }
 324                 if (sharedfile != "")           {               readShared();   }
 325         if (designfile != "")           {               readDesign();   }
 326
 327                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 328
 329
 330                 if (outputNames.size() != 0) {
 331                         m->mothurOutEndLine();
 332                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 333                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 334                         m->mothurOutEndLine();
 335
 336                         //set fasta file as new current fastafile
 337                         string current = "";
 338                         itTypes = outputTypes.find("fasta");
 339                         if (itTypes != outputTypes.end()) {
 340                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 341                         }
 342
 343                         itTypes = outputTypes.find("name");
 344                         if (itTypes != outputTypes.end()) {
 345                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 346                         }
 347
 348                         itTypes = outputTypes.find("group");
 349                         if (itTypes != outputTypes.end()) {
 350                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 351                         }
 352
 353                         itTypes = outputTypes.find("list");
 354                         if (itTypes != outputTypes.end()) {
 355                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 356                         }
 357
 358                         itTypes = outputTypes.find("taxonomy");
 359                         if (itTypes != outputTypes.end()) {
 360                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 361                         }
 362
 363                         itTypes = outputTypes.find("shared");
 364                         if (itTypes != outputTypes.end()) {
 365                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 366                         }
 367
 368             itTypes = outputTypes.find("design");
 369                         if (itTypes != outputTypes.end()) {
 370                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setDesignFile(current); }
 371                         }
 372                 }
 373
 374                 return 0;
 375         }
 376
 377         catch(exception& e) {
 378                 m->errorOut(e, "GetGroupsCommand", "execute");
 379                 exit(1);
 380         }
 381 }
 382
 383 //**********************************************************************************************************************
 384 int GetGroupsCommand::readFasta(){
 385         try {
 386                 string thisOutputDir = outputDir;
 387                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 388                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 389
 390                 ofstream out;
 391                 m->openOutputFile(outputFileName, out);
 392
 393                 ifstream in;
 394                 m->openInputFile(fastafile, in);
 395                 string name;
 396
 397                 bool wroteSomething = false;
 398                 int selectedCount = 0;
 399
 400                 while(!in.eof()){
 401                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 402
 403                         Sequence currSeq(in);
 404                         name = currSeq.getName();
 405
 406                         if (name != "") {
 407                                 //if this name is in the accnos file
 408                                 if (names.count(name) != 0) {
 409                                         wroteSomething = true;
 410
 411                                         currSeq.printSequence(out);
 412                                         selectedCount++;
 413                                 }else{
 414                                         //if you are not in the accnos file check if you are a name that needs to be changed
 415                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 416                                         if (it != uniqueToRedundant.end()) {
 417                                                 wroteSomething = true;
 418                                                 currSeq.setName(it->second);
 419                                                 currSeq.printSequence(out);
 420                                                 selectedCount++;
 421                                         }
 422                                 }
 423                         }
 424                         m->gobble(in);
 425                 }
 426                 in.close();
 427                 out.close();
 428
 429                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 430                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 431
 432                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 433
 434
 435                 return 0;
 436
 437         }
 438         catch(exception& e) {
 439                 m->errorOut(e, "GetGroupsCommand", "readFasta");
 440                 exit(1);
 441         }
 442 }
 443 //**********************************************************************************************************************
 444 int GetGroupsCommand::readShared(){
 445         try {
 446                 string thisOutputDir = outputDir;
 447                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 448
 449                 InputData input(sharedfile, "sharedfile");
 450                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 451
 452                 bool wroteSomething = false;
 453
 454                 while(lookup[0] != NULL) {
 455
 456                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 457                         ofstream out;
 458                         m->openOutputFile(outputFileName, out);
 459                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 460
 461                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 462
 463                         lookup[0]->printHeaders(out);
 464
 465                         for (int i = 0; i < lookup.size(); i++) {
 466                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 467                                 lookup[i]->print(out);
 468                                 wroteSomething = true;
 469
 470                         }
 471
 472                         //get next line to process
 473                         //prevent memory leak
 474                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 475                         lookup = input.getSharedRAbundVectors();
 476
 477                         out.close();
 478                 }
 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("Selected groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 487
 488                 return 0;
 489
 490         }
 491         catch(exception& e) {
 492                 m->errorOut(e, "GetGroupsCommand", "readShared");
 493                 exit(1);
 494         }
 495 }
 496 //**********************************************************************************************************************
 497 int GetGroupsCommand::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 selectedCount = 0;
 511
 512                 while(!in.eof()){
 513
 514                         selectedCount = 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 + ",";  selectedCount++;  }
 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                                                         selectedCount++;
 543                                                 }
 544                                         }
 545                                 }
 546
 547                                 //get last name
 548                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  selectedCount++;  }
 549                                 else{
 550                                         //if you are not in the accnos file check if you are a name that needs to be changed
 551                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 552                                         if (it != uniqueToRedundant.end()) {
 553                                                 newNames += it->second + ",";
 554                                                 selectedCount++;
 555                                         }
 556                                 }
 557
 558                                 //if there are names in this bin add to new list
 559                                 if (newNames != "") {
 560                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 561                                         newList.push_back(newNames);
 562                                 }
 563                         }
 564
 565                         //print new listvector
 566                         if (newList.getNumBins() != 0) {
 567                                 wroteSomething = true;
 568                                 newList.print(out);
 569                         }
 570
 571                         m->gobble(in);
 572                 }
 573                 in.close();
 574                 out.close();
 575
 576                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 577                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 578
 579                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 580
 581                 return 0;
 582
 583         }
 584         catch(exception& e) {
 585                 m->errorOut(e, "GetGroupsCommand", "readList");
 586                 exit(1);
 587         }
 588 }
 589 //**********************************************************************************************************************
 590 int GetGroupsCommand::readName(){
 591         try {
 592                 string thisOutputDir = outputDir;
 593                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 594                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 595
 596                 ofstream out;
 597                 m->openOutputFile(outputFileName, out);
 598
 599                 ifstream in;
 600                 m->openInputFile(namefile, in);
 601                 string name, firstCol, secondCol;
 602
 603                 bool wroteSomething = false;
 604                 int selectedCount = 0;
 605
 606                 while(!in.eof()){
 607                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 608
 609                         in >> firstCol;         m->gobble(in);
 610                         in >> secondCol;
 611
 612                         vector<string> parsedNames;
 613                         m->splitAtComma(secondCol, parsedNames);
 614
 615                         vector<string> validSecond;  validSecond.clear();
 616                         for (int i = 0; i < parsedNames.size(); i++) {
 617                                 if (names.count(parsedNames[i]) != 0) {
 618                                         validSecond.push_back(parsedNames[i]);
 619                                 }
 620                         }
 621
 622                         selectedCount += validSecond.size();
 623
 624                         //if the name in the first column is in the set then print it and any other names in second column also in set
 625                         if (names.count(firstCol) != 0) {
 626
 627                                 wroteSomething = true;
 628
 629                                 out << firstCol << '\t';
 630
 631                                 //you know you have at least one valid second since first column is valid
 632                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 633                                 out << validSecond[validSecond.size()-1] << endl;
 634
 635                                 //make first name in set you come to first column and then add the remaining names to second column
 636                         }else {
 637
 638                                 //you want part of this row
 639                                 if (validSecond.size() != 0) {
 640
 641                                         wroteSomething = true;
 642
 643                                         out << validSecond[0] << '\t';
 644
 645                                         //you know you have at least one valid second since first column is valid
 646                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 647                                         out << validSecond[validSecond.size()-1] << endl;
 648                                         uniqueToRedundant[firstCol] = validSecond[0];
 649                                 }
 650                         }
 651
 652                         m->gobble(in);
 653                 }
 654                 in.close();
 655                 out.close();
 656
 657                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 658                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 659
 660                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 661
 662                 return 0;
 663         }
 664         catch(exception& e) {
 665                 m->errorOut(e, "GetGroupsCommand", "readName");
 666                 exit(1);
 667         }
 668 }
 669
 670 //**********************************************************************************************************************
 671 int GetGroupsCommand::readGroup(){
 672         try {
 673                 string thisOutputDir = outputDir;
 674                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 675                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 676
 677                 ofstream out;
 678                 m->openOutputFile(outputFileName, out);
 679
 680                 ifstream in;
 681                 m->openInputFile(groupfile, in);
 682                 string name, group;
 683
 684                 bool wroteSomething = false;
 685                 int selectedCount = 0;
 686
 687                 while(!in.eof()){
 688                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 689
 690                         in >> name;                             //read from first column
 691                         in >> group;                    //read from second column
 692
 693                         //if this name is in the accnos file
 694                         if (names.count(name) != 0) {
 695                                 wroteSomething = true;
 696                                 out << name << '\t' << group << endl;
 697                                 selectedCount++;
 698                         }
 699
 700                         m->gobble(in);
 701                 }
 702                 in.close();
 703                 out.close();
 704
 705                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 706                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 707
 708                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 709
 710                 return 0;
 711         }
 712         catch(exception& e) {
 713                 m->errorOut(e, "GetGroupsCommand", "readGroup");
 714                 exit(1);
 715         }
 716 }
 717 //**********************************************************************************************************************
 718 int GetGroupsCommand::readDesign(){
 719         try {
 720                 string thisOutputDir = outputDir;
 721                 if (outputDir == "") {  thisOutputDir += m->hasPath(designfile);  }
 722                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(designfile)) + "pick" + m->getExtension(designfile);
 723
 724                 ofstream out;
 725                 m->openOutputFile(outputFileName, out);
 726
 727                 ifstream in;
 728                 m->openInputFile(designfile, in);
 729                 string name, group;
 730
 731                 bool wroteSomething = false;
 732                 int selectedCount = 0;
 733
 734                 while(!in.eof()){
 735                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 736
 737                         in >> name;                             //read from first column
 738                         in >> group;                    //read from second column
 739
 740                         //if this name is in the accnos file
 741                         if (m->inUsersGroups(name, Groups)) {
 742                                 wroteSomething = true;
 743                                 out << name << '\t' << group << endl;
 744                 selectedCount++;
 745                         }
 746
 747                         m->gobble(in);
 748                 }
 749                 in.close();
 750                 out.close();
 751
 752                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain groups from the groups you wish to get."); m->mothurOutEndLine();  }
 753                 outputTypes["design"].push_back(outputFileName); outputNames.push_back(outputFileName);
 754
 755                 m->mothurOut("Selected " + toString(selectedCount) + " groups from your design file."); m->mothurOutEndLine();
 756
 757
 758                 return 0;
 759         }
 760         catch(exception& e) {
 761                 m->errorOut(e, "GetGroupsCommand", "readDesign");
 762                 exit(1);
 763         }
 764 }
 765
 766
 767 //**********************************************************************************************************************
 768 int GetGroupsCommand::readTax(){
 769         try {
 770                 string thisOutputDir = outputDir;
 771                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 772                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 773                 ofstream out;
 774                 m->openOutputFile(outputFileName, out);
 775
 776                 ifstream in;
 777                 m->openInputFile(taxfile, in);
 778                 string name, tax;
 779
 780                 bool wroteSomething = false;
 781
 782                 while(!in.eof()){
 783                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 784
 785                         in >> name;                             //read from first column
 786                         in >> tax;                      //read from second column
 787
 788                         //if this name is in the accnos file
 789                         if (names.count(name) != 0) {
 790                                 wroteSomething = true;
 791                                 out << name << '\t' << tax << endl;
 792                         }else{
 793                                 //if you are not in the accnos file check if you are a name that needs to be changed
 794                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 795                                 if (it != uniqueToRedundant.end()) {
 796                                         wroteSomething = true;
 797                                         out << it->second << '\t' << tax << endl;
 798                                 }
 799                         }
 800
 801                         m->gobble(in);
 802                 }
 803                 in.close();
 804                 out.close();
 805
 806                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 807                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 808
 809                 return 0;
 810         }
 811         catch(exception& e) {
 812                 m->errorOut(e, "GetGroupsCommand", "readTax");
 813                 exit(1);
 814         }
 815 }
 816 //**********************************************************************************************************************
 817 void GetGroupsCommand::readAccnos(){
 818         try {
 819                 Groups.clear();
 820
 821                 ifstream in;
 822                 m->openInputFile(accnosfile, in);
 823                 string name;
 824
 825                 while(!in.eof()){
 826                         in >> name;
 827
 828                         Groups.push_back(name);
 829
 830                         m->gobble(in);
 831                 }
 832                 in.close();
 833
 834                 m->setGroups(Groups);
 835
 836         }
 837         catch(exception& e) {
 838                 m->errorOut(e, "GetGroupsCommand", "readAccnos");
 839                 exit(1);
 840         }
 841 }
 842 //**********************************************************************************************************************
 843 int GetGroupsCommand::fillNames(){
 844         try {
 845                 vector<string> seqs = groupMap->getNamesSeqs();
 846
 847                 for (int i = 0; i < seqs.size(); i++) {
 848
 849                         if (m->control_pressed) { return 0; }
 850
 851                         string group = groupMap->getGroup(seqs[i]);
 852
 853                         if (m->inUsersGroups(group, Groups)) {
 854                                 names.insert(seqs[i]);
 855                         }
 856                 }
 857
 858                 return 0;
 859         }
 860         catch(exception& e) {
 861                 m->errorOut(e, "GetGroupsCommand", "fillNames");
 862                 exit(1);
 863         }
 864 }
 865
 866 //**********************************************************************************************************************
 867
 868