removegroupscommand.cpp

   1 /*
   2  *  removegroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removegroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> RemoveGroupsCommand::setParameters(){
  18         try {
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "FNGLT","fasta",false,false,true); parameters.push_back(pfasta);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "sharedGroup", "none","shared",false,false,true); parameters.push_back(pshared);
  21         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "none","name",false,false,true); parameters.push_back(pname);
  22         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "none", "none","count",false,false,true); parameters.push_back(pcount);
  23                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "sharedGroup", "FNGLT","group",false,false,true); parameters.push_back(pgroup);
  24         CommandParameter pdesign("design", "InputTypes", "", "", "none", "sharedGroup", "FNGLT","design",false,false); parameters.push_back(pdesign);
  25                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "FNGLT","list",false,false,true); parameters.push_back(plist);
  26                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "FNGLT","taxonomy",false,false,true); parameters.push_back(ptaxonomy);
  27                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(paccnos);
  28                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
  29                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  30                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  31
  32                 vector<string> myArray;
  33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "RemoveGroupsCommand", "setParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 string RemoveGroupsCommand::getHelpString(){
  43         try {
  44                 string helpString = "";
  45                 helpString += "The remove.groups command removes sequences from a specfic group or set of groups from the following file types: fasta, name, group, count, list, taxonomy, design or sharedfile.\n";
  46                 helpString += "It outputs a file containing the sequences NOT in the those specified groups, or with a sharedfile eliminates the groups you selected.\n";
  47                 helpString += "The remove.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared, design and groups. The group or count parameter is required, unless you have a current group or count file or are using a sharedfile.\n";
  48                 helpString += "You must also provide an accnos containing the list of groups to remove or set the groups parameter to the groups you wish to remove.\n";
  49                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like removed.  You can separate group names with dashes.\n";
  50                 helpString += "The remove.groups command should be in the following format: remove.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
  51                 helpString += "Example remove.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
  52                 helpString += "or remove.groups(groups=pasture, fasta=amazon.fasta, amazon.groups).\n";
  53                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  54                 return helpString;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "RemoveGroupsCommand", "getHelpString");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 string RemoveGroupsCommand::getOutputPattern(string type) {
  63     try {
  64         string pattern = "";
  65
  66         if (type == "fasta")            {   pattern = "[filename],pick,[extension]";    }
  67         else if (type == "taxonomy")    {   pattern = "[filename],pick,[extension]";    }
  68         else if (type == "name")        {   pattern = "[filename],pick,[extension]";    }
  69         else if (type == "group")       {   pattern = "[filename],pick,[extension]";    }
  70         else if (type == "count")       {   pattern = "[filename],pick,[extension]";    }
  71         else if (type == "list")        {   pattern = "[filename],pick,[extension]";    }
  72         else if (type == "shared")      {   pattern = "[filename],[tag],pick,[extension]";    }
  73         else if (type == "design")      {   pattern = "[filename],[tag],pick,[extension]";    }
  74         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  75
  76         return pattern;
  77     }
  78     catch(exception& e) {
  79         m->errorOut(e, "RemoveGroupsCommand", "getOutputPattern");
  80         exit(1);
  81     }
  82 }
  83 //**********************************************************************************************************************
  84 RemoveGroupsCommand::RemoveGroupsCommand(){
  85         try {
  86                 abort = true; calledHelp = true;
  87                 setParameters();
  88                 vector<string> tempOutNames;
  89                 outputTypes["fasta"] = tempOutNames;
  90                 outputTypes["taxonomy"] = tempOutNames;
  91                 outputTypes["name"] = tempOutNames;
  92                 outputTypes["group"] = tempOutNames;
  93                 outputTypes["list"] = tempOutNames;
  94                 outputTypes["shared"] = tempOutNames;
  95         outputTypes["design"] = tempOutNames;
  96         outputTypes["count"] = tempOutNames;
  97         }
  98         catch(exception& e) {
  99                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 100                 exit(1);
 101         }
 102 }
 103 //**********************************************************************************************************************
 104 RemoveGroupsCommand::RemoveGroupsCommand(string option)  {
 105         try {
 106                 abort = false; calledHelp = false;
 107
 108                 //allow user to run help
 109                 if(option == "help") { help(); abort = true; calledHelp = true; }
 110                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 111
 112                 else {
 113                         vector<string> myArray = setParameters();
 114
 115                         OptionParser parser(option);
 116                         map<string,string> parameters = parser.getParameters();
 117
 118                         ValidParameters validParameter;
 119                         map<string,string>::iterator it;
 120
 121                         //check to make sure all parameters are valid for command
 122                         for (it = parameters.begin(); it != parameters.end(); it++) {
 123                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 124                         }
 125
 126                         //initialize outputTypes
 127                         vector<string> tempOutNames;
 128                         outputTypes["fasta"] = tempOutNames;
 129                         outputTypes["taxonomy"] = tempOutNames;
 130                         outputTypes["name"] = tempOutNames;
 131                         outputTypes["group"] = tempOutNames;
 132                         outputTypes["list"] = tempOutNames;
 133                         outputTypes["shared"] = tempOutNames;
 134             outputTypes["design"] = tempOutNames;
 135             outputTypes["count"] = tempOutNames;
 136
 137
 138                         //if the user changes the output directory command factory will send this info to us in the output parameter
 139                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 140
 141                         //if the user changes the input directory command factory will send this info to us in the output parameter
 142                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 143                         if (inputDir == "not found"){   inputDir = "";          }
 144                         else {
 145                                 string path;
 146                                 it = parameters.find("fasta");
 147                                 //user has given a template file
 148                                 if(it != parameters.end()){
 149                                         path = m->hasPath(it->second);
 150                                         //if the user has not given a path then, add inputdir. else leave path alone.
 151                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 152                                 }
 153
 154                                 it = parameters.find("accnos");
 155                                 //user has given a template file
 156                                 if(it != parameters.end()){
 157                                         path = m->hasPath(it->second);
 158                                         //if the user has not given a path then, add inputdir. else leave path alone.
 159                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 160                                 }
 161
 162                                 it = parameters.find("list");
 163                                 //user has given a template file
 164                                 if(it != parameters.end()){
 165                                         path = m->hasPath(it->second);
 166                                         //if the user has not given a path then, add inputdir. else leave path alone.
 167                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 168                                 }
 169
 170                                 it = parameters.find("name");
 171                                 //user has given a template file
 172                                 if(it != parameters.end()){
 173                                         path = m->hasPath(it->second);
 174                                         //if the user has not given a path then, add inputdir. else leave path alone.
 175                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 176                                 }
 177
 178                                 it = parameters.find("group");
 179                                 //user has given a template file
 180                                 if(it != parameters.end()){
 181                                         path = m->hasPath(it->second);
 182                                         //if the user has not given a path then, add inputdir. else leave path alone.
 183                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 184                                 }
 185
 186                                 it = parameters.find("taxonomy");
 187                                 //user has given a template file
 188                                 if(it != parameters.end()){
 189                                         path = m->hasPath(it->second);
 190                                         //if the user has not given a path then, add inputdir. else leave path alone.
 191                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 192                                 }
 193
 194                                 it = parameters.find("shared");
 195                                 //user has given a template file
 196                                 if(it != parameters.end()){
 197                                         path = m->hasPath(it->second);
 198                                         //if the user has not given a path then, add inputdir. else leave path alone.
 199                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 200                                 }
 201
 202                 it = parameters.find("design");
 203                                 //user has given a template file
 204                                 if(it != parameters.end()){
 205                                         path = m->hasPath(it->second);
 206                                         //if the user has not given a path then, add inputdir. else leave path alone.
 207                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 208                                 }
 209
 210                 it = parameters.find("count");
 211                                 //user has given a template file
 212                                 if(it != parameters.end()){
 213                                         path = m->hasPath(it->second);
 214                                         //if the user has not given a path then, add inputdir. else leave path alone.
 215                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 216                                 }
 217                         }
 218
 219
 220                         //check for required parameters
 221                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 222                         if (accnosfile == "not open") { accnosfile = ""; abort = true; }
 223                         else if (accnosfile == "not found") {  accnosfile = ""; }
 224                         else { m->setAccnosFile(accnosfile); }
 225
 226                         fastafile = validParameter.validFile(parameters, "fasta", true);
 227                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 228                         else if (fastafile == "not found") {  fastafile = "";  }
 229                         else { m->setFastaFile(fastafile); }
 230
 231                         namefile = validParameter.validFile(parameters, "name", true);
 232                         if (namefile == "not open") { namefile = ""; abort = true; }
 233                         else if (namefile == "not found") {  namefile = "";  }
 234                         else { m->setNameFile(namefile); }
 235
 236                         groupfile = validParameter.validFile(parameters, "group", true);
 237                         if (groupfile == "not open") { groupfile = "";  abort = true; }
 238                         else if (groupfile == "not found") {    groupfile = "";         }
 239                         else { m->setGroupFile(groupfile); }
 240
 241                         listfile = validParameter.validFile(parameters, "list", true);
 242                         if (listfile == "not open") { listfile = ""; abort = true; }
 243                         else if (listfile == "not found") {  listfile = "";  }
 244                         else { m->setListFile(listfile); }
 245
 246                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 247                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 248                         else if (taxfile == "not found") {  taxfile = "";  }
 249                         else { m->setTaxonomyFile(taxfile); }
 250
 251             designfile = validParameter.validFile(parameters, "design", true);
 252                         if (designfile == "not open") { designfile = ""; abort = true; }
 253                         else if (designfile == "not found") {  designfile = "";  }
 254                         else { m->setDesignFile(designfile); }
 255
 256                         groups = validParameter.validFile(parameters, "groups", false);
 257                         if (groups == "not found") { groups = ""; }
 258                         else {
 259                                 m->splitAtDash(groups, Groups);
 260                                 m->setGroups(Groups);
 261                         }
 262
 263                         sharedfile = validParameter.validFile(parameters, "shared", true);
 264                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 265                         else if (sharedfile == "not found") {  sharedfile = "";  }
 266                         else { m->setSharedFile(sharedfile); }
 267
 268
 269                         countfile = validParameter.validFile(parameters, "count", true);
 270             if (countfile == "not open") { countfile = ""; abort = true; }
 271             else if (countfile == "not found") { countfile = "";  }
 272             else { m->setCountTableFile(countfile); }
 273
 274             if ((namefile != "") && (countfile != "")) {
 275                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 276             }
 277
 278             if ((groupfile != "") && (countfile != "")) {
 279                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 280             }
 281
 282
 283                         if ((sharedfile == "") && (groupfile == "") && (designfile == "") && (countfile == "")) {
 284                                 //is there are current file available for any of these?
 285                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 286                                         //give priority to group, then shared
 287                                         groupfile = m->getGroupFile();
 288                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 289                                         else {
 290                                                 sharedfile = m->getSharedFile();
 291                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 292                                                 else {
 293                                                         countfile = m->getCountTableFile();
 294                             if (countfile != "") { m->mothurOut("Using " + countfile + " as input file for the count parameter."); m->mothurOutEndLine(); }
 295                             else {
 296                                 m->mothurOut("You have no current groupfile, countfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 297                             }
 298                                                 }
 299                                         }
 300                                 }else {
 301                                         //give priority to shared, then group
 302                                         sharedfile = m->getSharedFile();
 303                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 304                                         else {
 305                                                 groupfile = m->getGroupFile();
 306                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 307                                                 else {
 308                                                         designfile = m->getDesignFile();
 309                             if (designfile != "") { m->mothurOut("Using " + designfile + " as input file for the design parameter."); m->mothurOutEndLine(); }
 310                             else {
 311                                 countfile = m->getCountTableFile();
 312                                 if (countfile != "") { m->mothurOut("Using " + countfile + " as input file for the count parameter."); m->mothurOutEndLine(); }
 313                                 else {
 314                                     m->mothurOut("You have no current groupfile, designfile, countfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 315                                 }
 316
 317                             }
 318                                                 }
 319                                         }
 320                                 }
 321                         }
 322
 323                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file containing group names or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 324
 325                         if ((fastafile == "") && (namefile == "") && (countfile == "") && (groupfile == "")  && (designfile == "") && (sharedfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared, design, count or list."); m->mothurOutEndLine(); abort = true; }
 326                         if (((groupfile == "") && (countfile == "")) && ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")))  { m->mothurOut("If using a fasta, name, taxonomy, group or list, then you must provide a group or count file."); m->mothurOutEndLine(); abort = true; }
 327
 328             if (countfile == "") {
 329                 if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 330                     vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 331                     parser.getNameFile(files);
 332                 }
 333             }
 334                 }
 335
 336         }
 337         catch(exception& e) {
 338                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 339                 exit(1);
 340         }
 341 }
 342 //**********************************************************************************************************************
 343
 344 int RemoveGroupsCommand::execute(){
 345         try {
 346
 347                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 348
 349                 //get groups you want to remove
 350                 if (accnosfile != "") { m->readAccnos(accnosfile, Groups); m->setGroups(Groups);  }
 351
 352                 if (groupfile != "") {
 353                         groupMap = new GroupMap(groupfile);
 354                         groupMap->readMap();
 355
 356                         //make sure groups are valid
 357                         //takes care of user setting groupNames that are invalid or setting groups=all
 358                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 359                         vector<string> checkedGroups;
 360             for (int i = 0; i < Groups.size(); i++) {
 361                 if (m->inUsersGroups(Groups[i], namesGroups)) { checkedGroups.push_back(Groups[i]); }
 362                 else {  m->mothurOut("[WARNING]: " + Groups[i] + " is not a valid group in your groupfile, ignoring.\n"); }
 363             }
 364
 365             if (checkedGroups.size() == 0) { m->mothurOut("[ERROR]: no valid groups, aborting.\n"); delete groupMap; return 0; }
 366                         else {
 367                 Groups = checkedGroups;
 368                 m->setGroups(Groups);
 369             }
 370
 371                         //fill names with names of sequences that are from the groups we want to remove
 372                         fillNames();
 373
 374                         delete groupMap;
 375                 }else if (countfile != ""){
 376             if ((fastafile != "") || (listfile != "") || (taxfile != "")) {
 377                 m->mothurOut("\n[NOTE]: The count file should contain only unique names, so mothur assumes your fasta, list and taxonomy files also contain only uniques.\n\n");
 378             }
 379             CountTable ct;
 380             ct.readTable(countfile, true, false);
 381             if (!ct.hasGroupInfo()) { m->mothurOut("[ERROR]: your count file does not contain group info, aborting.\n"); return 0; }
 382
 383             vector<string> gNamesOfGroups = ct.getNamesOfGroups();
 384             SharedUtil util;
 385             util.setGroups(Groups, gNamesOfGroups);
 386             vector<string> namesOfSeqs = ct.getNamesOfSeqs();
 387             sort(Groups.begin(), Groups.end());
 388
 389             for (int i = 0; i < namesOfSeqs.size(); i++) {
 390                 vector<string> thisSeqsGroups = ct.getGroups(namesOfSeqs[i]);
 391                 if (m->isSubset(Groups, thisSeqsGroups)) { //you only have seqs from these groups so remove you
 392                     names.insert(namesOfSeqs[i]);
 393                 }
 394             }
 395         }
 396
 397
 398                 if (m->control_pressed) { return 0; }
 399
 400                 //read through the correct file and output lines you want to keep
 401                 if (namefile != "")                     {               readName();             }
 402                 if (fastafile != "")            {               readFasta();    }
 403                 if (groupfile != "")            {               readGroup();    }
 404         if (countfile != "")            {               readCount();    }
 405                 if (listfile != "")                     {               readList();             }
 406                 if (taxfile != "")                      {               readTax();              }
 407                 if (sharedfile != "")           {               readShared();   }
 408         if (designfile != "")           {               readDesign();   }
 409
 410                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 411
 412                 if (outputNames.size() != 0) {
 413                         m->mothurOutEndLine();
 414                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 415                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 416                         m->mothurOutEndLine();
 417
 418                         //set fasta file as new current fastafile
 419                         string current = "";
 420                         itTypes = outputTypes.find("fasta");
 421                         if (itTypes != outputTypes.end()) {
 422                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 423                         }
 424
 425                         itTypes = outputTypes.find("name");
 426                         if (itTypes != outputTypes.end()) {
 427                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 428                         }
 429
 430                         itTypes = outputTypes.find("group");
 431                         if (itTypes != outputTypes.end()) {
 432                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 433                         }
 434
 435                         itTypes = outputTypes.find("list");
 436                         if (itTypes != outputTypes.end()) {
 437                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 438                         }
 439
 440                         itTypes = outputTypes.find("taxonomy");
 441                         if (itTypes != outputTypes.end()) {
 442                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 443                         }
 444
 445                         itTypes = outputTypes.find("shared");
 446                         if (itTypes != outputTypes.end()) {
 447                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 448                         }
 449
 450             itTypes = outputTypes.find("design");
 451                         if (itTypes != outputTypes.end()) {
 452                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setDesignFile(current); }
 453                         }
 454
 455             itTypes = outputTypes.find("count");
 456                         if (itTypes != outputTypes.end()) {
 457                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 458                         }
 459                 }
 460
 461                 return 0;
 462         }
 463
 464         catch(exception& e) {
 465                 m->errorOut(e, "RemoveGroupsCommand", "execute");
 466                 exit(1);
 467         }
 468 }
 469
 470 //**********************************************************************************************************************
 471 int RemoveGroupsCommand::readFasta(){
 472         try {
 473                 string thisOutputDir = outputDir;
 474                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 475         map<string, string> variables;
 476         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(fastafile));
 477         variables["[extension]"] = m->getExtension(fastafile);
 478                 string outputFileName = getOutputFileName("fasta", variables);
 479
 480                 ofstream out;
 481                 m->openOutputFile(outputFileName, out);
 482
 483                 ifstream in;
 484                 m->openInputFile(fastafile, in);
 485                 string name;
 486
 487                 bool wroteSomething = false;
 488                 int removedCount = 0;
 489
 490                 while(!in.eof()){
 491                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 492
 493                         Sequence currSeq(in);
 494                         name = currSeq.getName();
 495
 496                         if (name != "") {
 497                                 //if this name is in the accnos file
 498                                 if (names.count(name) == 0) {
 499                                         wroteSomething = true;
 500                                         currSeq.printSequence(out);
 501                                 }else {
 502                                         //if you are not in the accnos file check if you are a name that needs to be changed
 503                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 504                                         if (it != uniqueToRedundant.end()) {
 505                                                 wroteSomething = true;
 506                                                 currSeq.setName(it->second);
 507                                                 currSeq.printSequence(out);
 508                                         }else { removedCount++; }
 509                                 }
 510                         }
 511                         m->gobble(in);
 512                 }
 513                 in.close();
 514                 out.close();
 515
 516                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 517                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 518
 519                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 520
 521                 return 0;
 522
 523         }
 524         catch(exception& e) {
 525                 m->errorOut(e, "RemoveGroupsCommand", "readFasta");
 526                 exit(1);
 527         }
 528 }
 529 //**********************************************************************************************************************
 530 int RemoveGroupsCommand::readShared(){
 531         try {
 532                 string thisOutputDir = outputDir;
 533                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 534
 535                 //get group names from sharedfile so we can set Groups to the groupNames we want to keep
 536                 //that way we can take advantage of the reads in inputdata and sharedRabundVector
 537                 InputData* tempInput = new InputData(sharedfile, "sharedfile");
 538                 vector<SharedRAbundVector*> lookup = tempInput->getSharedRAbundVectors();
 539
 540         map<string, string> variables;
 541         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile));
 542         variables["[extension]"] = m->getExtension(sharedfile);
 543
 544                 //save m->Groups
 545                 vector<string> allGroupsNames = m->getAllGroups();
 546                 vector<string> mothurOutGroups = m->getGroups();
 547
 548                 vector<string> groupsToKeep;
 549                 for (int i = 0; i < allGroupsNames.size(); i++) {
 550                         if (!m->inUsersGroups(allGroupsNames[i], m->getGroups())) {
 551                                 groupsToKeep.push_back(allGroupsNames[i]);
 552                         }
 553                 }
 554
 555                 if (allGroupsNames.size() == groupsToKeep.size()) { m->mothurOut("Your file does not contain any groups you wish to remove."); m->mothurOutEndLine(); m->setGroups(mothurOutGroups); delete tempInput; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  return 0; }
 556
 557                 //reset read
 558                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 559                 delete tempInput;
 560                 m->setGroups(groupsToKeep);
 561                 m->clearAllGroups();
 562                 m->saveNextLabel = "";
 563                 m->printedSharedHeaders = false;
 564                 m->currentSharedBinLabels.clear();
 565                 m->sharedBinLabelsInFile.clear();
 566
 567                 InputData input(sharedfile, "sharedfile");
 568                 lookup = input.getSharedRAbundVectors();
 569
 570                 bool wroteSomething = false;
 571
 572                 while(lookup[0] != NULL) {
 573
 574                         variables["[tag]"] = lookup[0]->getLabel();
 575             string outputFileName = getOutputFileName("shared", variables);
 576                         ofstream out;
 577                         m->openOutputFile(outputFileName, out);
 578                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 579
 580                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 581
 582                         lookup[0]->printHeaders(out);
 583
 584                         for (int i = 0; i < lookup.size(); i++) {
 585                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 586                                 lookup[i]->print(out);
 587                                 wroteSomething = true;
 588
 589                         }
 590
 591                         //get next line to process
 592                         //prevent memory leak
 593                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 594                         lookup = input.getSharedRAbundVectors();
 595
 596                         out.close();
 597                 }
 598
 599
 600                 m->setGroups(mothurOutGroups);
 601
 602                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 603
 604                 string groupsString = "";
 605                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 606                 groupsString += Groups[Groups.size()-1];
 607
 608                 m->mothurOut("Removed groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 609
 610                 return 0;
 611
 612         }
 613         catch(exception& e) {
 614                 m->errorOut(e, "RemoveGroupsCommand", "readShared");
 615                 exit(1);
 616         }
 617 }
 618 //**********************************************************************************************************************
 619 int RemoveGroupsCommand::readList(){
 620         try {
 621                 string thisOutputDir = outputDir;
 622                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 623                 map<string, string> variables;
 624         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
 625         variables["[extension]"] = m->getExtension(listfile);
 626
 627                 ifstream in;
 628                 m->openInputFile(listfile, in);
 629
 630                 bool wroteSomething = false;
 631                 int removedCount = 0;
 632
 633                 while(!in.eof()){
 634
 635                         removedCount = 0;
 636
 637                         //read in list vector
 638                         ListVector list(in);
 639
 640             variables["[tag]"] = list.getLabel();
 641             string outputFileName = getOutputFileName("list", variables);
 642
 643                         ofstream out;
 644                         m->openOutputFile(outputFileName, out);
 645                         outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 646
 647             vector<string> binLabels = list.getLabels();
 648             vector<string> newBinLabels;
 649
 650                         //make a new list vector
 651                         ListVector newList;
 652                         newList.setLabel(list.getLabel());
 653
 654                         //for each bin
 655                         for (int i = 0; i < list.getNumBins(); i++) {
 656                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 657
 658                                 //parse out names that are in accnos file
 659                                 string binnames = list.get(i);
 660
 661                                 string newNames = "";
 662                                 while (binnames.find_first_of(',') != -1) {
 663                                         string name = binnames.substr(0,binnames.find_first_of(','));
 664                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 665
 666                                         //if that name is in the .accnos file, add it
 667                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 668                                         else {
 669                                                 //if you are not in the accnos file check if you are a name that needs to be changed
 670                                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 671                                                 if (it != uniqueToRedundant.end()) {
 672                                                         newNames += it->second + ",";
 673                                                 }else { removedCount++; }
 674                                         }
 675                                 }
 676
 677                                 //get last name
 678                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 679                                 else { //if you are not in the accnos file check if you are a name that needs to be changed
 680                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 681                                         if (it != uniqueToRedundant.end()) {
 682                                                 newNames += it->second + ",";
 683                                         }else { removedCount++; }
 684                                 }
 685
 686                                 //if there are names in this bin add to new list
 687                                 if (newNames != "") {
 688                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 689                                         newList.push_back(newNames);
 690                     newBinLabels.push_back(binLabels[i]);
 691                                 }
 692                         }
 693
 694                         //print new listvector
 695                         if (newList.getNumBins() != 0) {
 696                                 wroteSomething = true;
 697                                 newList.setLabels(newBinLabels);
 698                 newList.printHeaders(out);
 699                                 newList.print(out);
 700                         }
 701
 702                         m->gobble(in);
 703             out.close();
 704                 }
 705                 in.close();
 706
 707
 708                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 709                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 710
 711                 return 0;
 712
 713         }
 714         catch(exception& e) {
 715                 m->errorOut(e, "RemoveGroupsCommand", "readList");
 716                 exit(1);
 717         }
 718 }
 719 //**********************************************************************************************************************
 720 int RemoveGroupsCommand::readName(){
 721         try {
 722                 string thisOutputDir = outputDir;
 723                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 724                 map<string, string> variables;
 725                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(namefile));
 726         variables["[extension]"] = m->getExtension(namefile);
 727                 string outputFileName = getOutputFileName("name", variables);
 728                 ofstream out;
 729                 m->openOutputFile(outputFileName, out);
 730
 731                 ifstream in;
 732                 m->openInputFile(namefile, in);
 733                 string name, firstCol, secondCol;
 734
 735                 bool wroteSomething = false;
 736                 int removedCount = 0;
 737
 738                 while(!in.eof()){
 739                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 740
 741                         in >> firstCol;         m->gobble(in);
 742                         in >> secondCol;
 743
 744                         vector<string> parsedNames;
 745                         m->splitAtComma(secondCol, parsedNames);
 746
 747                         vector<string> validSecond;  validSecond.clear();
 748                         for (int i = 0; i < parsedNames.size(); i++) {
 749                                 if (names.count(parsedNames[i]) == 0) {
 750                                         validSecond.push_back(parsedNames[i]);
 751                                 }
 752                         }
 753
 754                         removedCount += parsedNames.size()-validSecond.size();
 755
 756                         //if the name in the first column is in the set then print it and any other names in second column also in set
 757                         if (names.count(firstCol) == 0) {
 758
 759                                 wroteSomething = true;
 760
 761                                 out << firstCol << '\t';
 762
 763                                 //you know you have at least one valid second since first column is valid
 764                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 765                                 out << validSecond[validSecond.size()-1] << endl;
 766
 767                                 //make first name in set you come to first column and then add the remaining names to second column
 768                         }else {
 769
 770                                 //you want part of this row
 771                                 if (validSecond.size() != 0) {
 772
 773                                         wroteSomething = true;
 774
 775                                         out << validSecond[0] << '\t';
 776
 777                                         //you know you have at least one valid second since first column is valid
 778                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 779                                         out << validSecond[validSecond.size()-1] << endl;
 780                                         uniqueToRedundant[firstCol] = validSecond[0];
 781                                 }
 782                         }
 783
 784                         m->gobble(in);
 785                 }
 786                 in.close();
 787                 out.close();
 788
 789                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 790                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 791
 792                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 793
 794                 return 0;
 795         }
 796         catch(exception& e) {
 797                 m->errorOut(e, "RemoveGroupsCommand", "readName");
 798                 exit(1);
 799         }
 800 }
 801
 802 //**********************************************************************************************************************
 803 int RemoveGroupsCommand::readGroup(){
 804         try {
 805                 string thisOutputDir = outputDir;
 806                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 807         map<string, string> variables;
 808                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(groupfile));
 809         variables["[extension]"] = m->getExtension(groupfile);
 810                 string outputFileName = getOutputFileName("group", variables);
 811                 ofstream out;
 812                 m->openOutputFile(outputFileName, out);
 813
 814                 ifstream in;
 815                 m->openInputFile(groupfile, in);
 816                 string name, group;
 817
 818                 bool wroteSomething = false;
 819                 int removedCount = 0;
 820
 821                 while(!in.eof()){
 822                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 823
 824                         in >> name;                             //read from first column
 825                         in >> group;                    //read from second column
 826
 827                         //if this name is in the accnos file
 828                         if (names.count(name) == 0) {
 829                                 wroteSomething = true;
 830                                 out << name << '\t' << group << endl;
 831                         }else {  removedCount++;  }
 832
 833                         m->gobble(in);
 834                 }
 835                 in.close();
 836                 out.close();
 837
 838                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 839                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 840
 841                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 842
 843
 844                 return 0;
 845         }
 846         catch(exception& e) {
 847                 m->errorOut(e, "RemoveGroupsCommand", "readGroup");
 848                 exit(1);
 849         }
 850 }
 851 //**********************************************************************************************************************
 852 int RemoveGroupsCommand::readCount(){
 853         try {
 854                 string thisOutputDir = outputDir;
 855                 if (outputDir == "") {  thisOutputDir += m->hasPath(countfile);  }
 856                 map<string, string> variables;
 857                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
 858         variables["[extension]"] = m->getExtension(countfile);
 859                 string outputFileName = getOutputFileName("count", variables);
 860
 861                 ofstream out;
 862                 m->openOutputFile(outputFileName, out);
 863
 864                 ifstream in;
 865                 m->openInputFile(countfile, in);
 866
 867                 bool wroteSomething = false;
 868                 int removedCount = 0;
 869
 870         string headers = m->getline(in); m->gobble(in);
 871         vector<string> columnHeaders = m->splitWhiteSpace(headers);
 872
 873         vector<string> groups;
 874         map<int, string> originalGroupIndexes;
 875         map<string, int> GroupIndexes;
 876         set<int> indexOfGroupsChosen;
 877         for (int i = 2; i < columnHeaders.size(); i++) {  groups.push_back(columnHeaders[i]);  originalGroupIndexes[i-2] = columnHeaders[i]; }
 878         //sort groups to keep consistent with how we store the groups in groupmap
 879         sort(groups.begin(), groups.end());
 880         for (int i = 0; i < groups.size(); i++) {  GroupIndexes[groups[i]] = i; }
 881
 882                 vector<string> groupsToKeep;
 883                 for (int i = 0; i < groups.size(); i++) {
 884                         if (!m->inUsersGroups(groups[i], Groups)) { groupsToKeep.push_back(groups[i]); }
 885                 }
 886         sort(groupsToKeep.begin(), groupsToKeep.end());
 887         out << "Representative_Sequence\ttotal\t";
 888         for (int i = 0; i < groupsToKeep.size(); i++) { out << groupsToKeep[i] << '\t'; indexOfGroupsChosen.insert(GroupIndexes[groupsToKeep[i]]); }
 889         out << endl;
 890
 891         string name; int oldTotal;
 892         while (!in.eof()) {
 893
 894             if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 895
 896             in >> name; m->gobble(in); in >> oldTotal; m->gobble(in);
 897             if (m->debug) { m->mothurOut("[DEBUG]: " + name + '\t' + toString(oldTotal) + "\n"); }
 898
 899             if (names.count(name) == 0) {
 900                 //if group info, then read it
 901                 vector<int> selectedCounts; int thisTotal = 0; int temp;
 902                 for (int i = 0; i < groups.size(); i++) {
 903                     int thisIndex = GroupIndexes[originalGroupIndexes[i]];
 904                     in >> temp;  m->gobble(in);
 905                     if (indexOfGroupsChosen.count(thisIndex) != 0) { //we want this group
 906                         selectedCounts.push_back(temp); thisTotal += temp;
 907                     }
 908                 }
 909
 910                 out << name << '\t' << thisTotal << '\t';
 911                 for (int i = 0; i < selectedCounts.size(); i++) {  out << selectedCounts[i] << '\t'; }
 912                 out << endl;
 913
 914                 wroteSomething = true;
 915                 removedCount+= (oldTotal - thisTotal);
 916             }else {  m->getline(in); removedCount += oldTotal; }
 917
 918             m->gobble(in);
 919         }
 920         in.close();
 921                 out.close();
 922
 923                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 924                 outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
 925
 926                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your count file."); m->mothurOutEndLine();
 927
 928                 return 0;
 929         }
 930         catch(exception& e) {
 931                 m->errorOut(e, "RemoveGroupsCommand", "readCount");
 932                 exit(1);
 933         }
 934 }
 935 //**********************************************************************************************************************
 936 int RemoveGroupsCommand::readDesign(){
 937         try {
 938                 string thisOutputDir = outputDir;
 939                 if (outputDir == "") {  thisOutputDir += m->hasPath(designfile);  }
 940         map<string, string> variables;
 941                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(designfile));
 942         variables["[extension]"] = m->getExtension(designfile);
 943                 string outputFileName = getOutputFileName("design", variables);
 944
 945                 ofstream out;
 946                 m->openOutputFile(outputFileName, out);
 947
 948                 ifstream in;
 949                 m->openInputFile(designfile, in);
 950                 string name, group;
 951
 952                 bool wroteSomething = false;
 953                 int removedCount = 0;
 954
 955                 while(!in.eof()){
 956                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 957
 958                         in >> name;                             //read from first column
 959                         in >> group;                    //read from second column
 960
 961                         //if this name is in the accnos file
 962                         if (!(m->inUsersGroups(name, Groups))) {
 963                                 wroteSomething = true;
 964                                 out << name << '\t' << group << endl;
 965                         }else {  removedCount++;  }
 966
 967                         m->gobble(in);
 968                 }
 969                 in.close();
 970                 out.close();
 971
 972                 if (wroteSomething == false) {  m->mothurOut("Your file contains only groups from the groups you wish to remove."); m->mothurOutEndLine();  }
 973                 outputTypes["design"].push_back(outputFileName); outputNames.push_back(outputFileName);
 974
 975                 m->mothurOut("Removed " + toString(removedCount) + " groups from your design file."); m->mothurOutEndLine();
 976
 977
 978                 return 0;
 979         }
 980         catch(exception& e) {
 981                 m->errorOut(e, "RemoveGroupsCommand", "readDesign");
 982                 exit(1);
 983         }
 984 }
 985
 986 //**********************************************************************************************************************
 987 int RemoveGroupsCommand::readTax(){
 988         try {
 989                 string thisOutputDir = outputDir;
 990                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 991                 map<string, string> variables;
 992                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(taxfile));
 993         variables["[extension]"] = m->getExtension(taxfile);
 994                 string outputFileName = getOutputFileName("taxonomy", variables);
 995                 ofstream out;
 996                 m->openOutputFile(outputFileName, out);
 997
 998                 ifstream in;
 999                 m->openInputFile(taxfile, in);
1000                 string name, tax;
1001
1002                 bool wroteSomething = false;
1003                 int removedCount = 0;
1004
1005                 while(!in.eof()){
1006                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
1007
1008                         in >> name;                             //read from first column
1009                         in >> tax;                      //read from second column
1010
1011                         //if this name is in the accnos file
1012                         if (names.count(name) == 0) {
1013                                 wroteSomething = true;
1014                                 out << name << '\t' << tax << endl;
1015                         }else {  //if you are not in the accnos file check if you are a name that needs to be changed
1016                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
1017                                 if (it != uniqueToRedundant.end()) {
1018                                         wroteSomething = true;
1019                                         out << it->second << '\t' << tax << endl;
1020                                 }else { removedCount++; }  }
1021
1022                         m->gobble(in);
1023                 }
1024                 in.close();
1025                 out.close();
1026
1027                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
1028                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
1029
1030                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
1031
1032                 return 0;
1033         }
1034         catch(exception& e) {
1035                 m->errorOut(e, "RemoveGroupsCommand", "readTax");
1036                 exit(1);
1037         }
1038 }
1039 //**********************************************************************************************************************
1040 int RemoveGroupsCommand::fillNames(){
1041         try {
1042                 vector<string> seqs = groupMap->getNamesSeqs();
1043
1044                 for (int i = 0; i < seqs.size(); i++) {
1045
1046                         if (m->control_pressed) { return 0; }
1047
1048                         string group = groupMap->getGroup(seqs[i]);
1049
1050                         if (m->inUsersGroups(group, Groups)) {
1051                                 names.insert(seqs[i]);
1052                         }
1053                 }
1054
1055                 return 0;
1056         }
1057         catch(exception& e) {
1058                 m->errorOut(e, "RemoveGroupsCommand", "fillNames");
1059                 exit(1);
1060         }
1061 }
1062
1063 //**********************************************************************************************************************
1064
1065
1066