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