removeseqscommand.cpp

   1 /*
   2  *  removeseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "removeseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15 vector<string> RemoveSeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
  22                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
  23                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pqfile);
  24                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(paccnos);
  25                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pdups);
  26                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  27                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  28
  29                 vector<string> myArray;
  30                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  31                 return myArray;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "RemoveSeqsCommand", "setParameters");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 string RemoveSeqsCommand::getHelpString(){
  40         try {
  41                 string helpString = "";
  42                 helpString += "The remove.seqs command reads an .accnos file and at least one of the following file types: fasta, name, group, list, taxonomy, quality or alignreport file.\n";
  43                 helpString += "It outputs a file containing the sequences NOT in the .accnos file.\n";
  44                 helpString += "The remove.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos and at least one of the file parameters.\n";
  45                 helpString += "The dups parameter allows you to remove the entire line from a name file if you remove any name from the line. default=true. \n";
  46                 helpString += "The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
  47                 helpString += "Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
  48                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  49                 return helpString;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "RemoveSeqsCommand", "getHelpString");
  53                 exit(1);
  54         }
  55 }
  56 //**********************************************************************************************************************
  57 string RemoveSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  58         try {
  59         string outputFileName = "";
  60                 map<string, vector<string> >::iterator it;
  61
  62         //is this a type this command creates
  63         it = outputTypes.find(type);
  64         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  65         else {
  66             if (type == "fasta")            {   outputFileName =  "pick" + m->getExtension(inputName);   }
  67             else if (type == "taxonomy")    {   outputFileName =  "pick" + m->getExtension(inputName);   }
  68             else if (type == "name")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  69             else if (type == "group")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  70             else if (type == "list")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  71             else if (type == "qfile")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  72             else if (type == "alignreport") {   outputFileName =  "pick.align.report";                   }
  73             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  74         }
  75         return outputFileName;
  76         }
  77         catch(exception& e) {
  78                 m->errorOut(e, "RemoveSeqsCommand", "getOutputFileNameTag");
  79                 exit(1);
  80         }
  81 }
  82
  83 //**********************************************************************************************************************
  84 RemoveSeqsCommand::RemoveSeqsCommand(){
  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["alignreport"] = tempOutNames;
  94                 outputTypes["list"] = tempOutNames;
  95                 outputTypes["qfile"] = tempOutNames;
  96         }
  97         catch(exception& e) {
  98                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
  99                 exit(1);
 100         }
 101 }
 102 //**********************************************************************************************************************
 103 RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
 104         try {
 105                 abort = false; calledHelp = false;
 106
 107                 //allow user to run help
 108                 if(option == "help") { help(); abort = true; calledHelp = true; }
 109                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 110
 111                 else {
 112                         vector<string> myArray = setParameters();
 113
 114                         OptionParser parser(option);
 115                         map<string,string> parameters = parser.getParameters();
 116
 117                         ValidParameters validParameter;
 118                         map<string,string>::iterator it;
 119
 120                         //check to make sure all parameters are valid for command
 121                         for (it = parameters.begin(); it != parameters.end(); it++) {
 122                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 123                         }
 124
 125                         //initialize outputTypes
 126                         vector<string> tempOutNames;
 127                         outputTypes["fasta"] = tempOutNames;
 128                         outputTypes["taxonomy"] = tempOutNames;
 129                         outputTypes["name"] = tempOutNames;
 130                         outputTypes["group"] = tempOutNames;
 131                         outputTypes["alignreport"] = tempOutNames;
 132                         outputTypes["list"] = tempOutNames;
 133                         outputTypes["qfile"] = tempOutNames;
 134
 135                         //if the user changes the output directory command factory will send this info to us in the output parameter
 136                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 137
 138                         //if the user changes the input directory command factory will send this info to us in the output parameter
 139                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 140                         if (inputDir == "not found"){   inputDir = "";          }
 141                         else {
 142                                 string path;
 143                                 it = parameters.find("alignreport");
 144                                 //user has given a template file
 145                                 if(it != parameters.end()){
 146                                         path = m->hasPath(it->second);
 147                                         //if the user has not given a path then, add inputdir. else leave path alone.
 148                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 149                                 }
 150
 151                                 it = parameters.find("fasta");
 152                                 //user has given a template file
 153                                 if(it != parameters.end()){
 154                                         path = m->hasPath(it->second);
 155                                         //if the user has not given a path then, add inputdir. else leave path alone.
 156                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 157                                 }
 158
 159                                 it = parameters.find("accnos");
 160                                 //user has given a template file
 161                                 if(it != parameters.end()){
 162                                         path = m->hasPath(it->second);
 163                                         //if the user has not given a path then, add inputdir. else leave path alone.
 164                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 165                                 }
 166
 167                                 it = parameters.find("list");
 168                                 //user has given a template file
 169                                 if(it != parameters.end()){
 170                                         path = m->hasPath(it->second);
 171                                         //if the user has not given a path then, add inputdir. else leave path alone.
 172                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 173                                 }
 174
 175                                 it = parameters.find("name");
 176                                 //user has given a template file
 177                                 if(it != parameters.end()){
 178                                         path = m->hasPath(it->second);
 179                                         //if the user has not given a path then, add inputdir. else leave path alone.
 180                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 181                                 }
 182
 183                                 it = parameters.find("group");
 184                                 //user has given a template file
 185                                 if(it != parameters.end()){
 186                                         path = m->hasPath(it->second);
 187                                         //if the user has not given a path then, add inputdir. else leave path alone.
 188                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 189                                 }
 190
 191                                 it = parameters.find("taxonomy");
 192                                 //user has given a template file
 193                                 if(it != parameters.end()){
 194                                         path = m->hasPath(it->second);
 195                                         //if the user has not given a path then, add inputdir. else leave path alone.
 196                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 197                                 }
 198
 199                                 it = parameters.find("qfile");
 200                                 //user has given a template file
 201                                 if(it != parameters.end()){
 202                                         path = m->hasPath(it->second);
 203                                         //if the user has not given a path then, add inputdir. else leave path alone.
 204                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 205                                 }
 206                         }
 207
 208
 209                         //check for required parameters
 210                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 211                         if (accnosfile == "not open") { abort = true; }
 212                         else if (accnosfile == "not found") {
 213                                 accnosfile = m->getAccnosFile();
 214                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
 215                                 else {
 216                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
 217                                         abort = true;
 218                                 }
 219                         }else { m->setAccnosFile(accnosfile); }
 220
 221                         fastafile = validParameter.validFile(parameters, "fasta", true);
 222                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 223                         else if (fastafile == "not found") {  fastafile = "";  }
 224                         else { m->setFastaFile(fastafile); }
 225
 226                         namefile = validParameter.validFile(parameters, "name", true);
 227                         if (namefile == "not open") { namefile = ""; abort = true; }
 228                         else if (namefile == "not found") {  namefile = "";  }
 229                         else { m->setNameFile(namefile); }
 230
 231                         groupfile = validParameter.validFile(parameters, "group", true);
 232                         if (groupfile == "not open") { abort = true; }
 233                         else if (groupfile == "not found") {  groupfile = "";  }
 234                         else { m->setGroupFile(groupfile); }
 235
 236                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 237                         if (alignfile == "not open") { abort = true; }
 238                         else if (alignfile == "not found") {  alignfile = "";  }
 239
 240                         listfile = validParameter.validFile(parameters, "list", true);
 241                         if (listfile == "not open") { abort = true; }
 242                         else if (listfile == "not found") {  listfile = "";  }
 243                         else { m->setListFile(listfile); }
 244
 245                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 246                         if (taxfile == "not open") { abort = true; }
 247                         else if (taxfile == "not found") {  taxfile = "";  }
 248                         else { m->setTaxonomyFile(taxfile); }
 249
 250                         qualfile = validParameter.validFile(parameters, "qfile", true);
 251                         if (qualfile == "not open") { abort = true; }
 252                         else if (qualfile == "not found") {  qualfile = "";  }
 253                         else { m->setQualFile(qualfile); }
 254
 255                         string usedDups = "true";
 256                         string temp = validParameter.validFile(parameters, "dups", false);
 257                         if (temp == "not found") {
 258                                 if (namefile != "") {  temp = "true";                                   }
 259                                 else                            {  temp = "false"; usedDups = "";       }
 260                         }
 261                         dups = m->isTrue(temp);
 262
 263                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, group, taxonomy, quality, alignreport or list."); m->mothurOutEndLine(); abort = true; }
 264
 265                         if ((fastafile != "") && (namefile == "")) {
 266                                 vector<string> files; files.push_back(fastafile);
 267                                 parser.getNameFile(files);
 268                         }
 269                 }
 270
 271         }
 272         catch(exception& e) {
 273                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 274                 exit(1);
 275         }
 276 }
 277 //**********************************************************************************************************************
 278
 279 int RemoveSeqsCommand::execute(){
 280         try {
 281
 282                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 283
 284                 //get names you want to keep
 285                 names = m->readAccnos(accnosfile);
 286
 287                 if (m->control_pressed) { return 0; }
 288
 289                 //read through the correct file and output lines you want to keep
 290                 if (namefile != "")                     {               readName();             }
 291                 if (fastafile != "")            {               readFasta();    }
 292                 if (groupfile != "")            {               readGroup();    }
 293                 if (alignfile != "")            {               readAlign();    }
 294                 if (listfile != "")                     {               readList();             }
 295                 if (taxfile != "")                      {               readTax();              }
 296                 if (qualfile != "")                     {               readQual();             }
 297
 298                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 299
 300                 if (outputNames.size() != 0) {
 301                         m->mothurOutEndLine();
 302                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 303                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 304                         m->mothurOutEndLine();
 305
 306                         //set fasta file as new current fastafile
 307                         string current = "";
 308                         itTypes = outputTypes.find("fasta");
 309                         if (itTypes != outputTypes.end()) {
 310                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 311                         }
 312
 313                         itTypes = outputTypes.find("name");
 314                         if (itTypes != outputTypes.end()) {
 315                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 316                         }
 317
 318                         itTypes = outputTypes.find("group");
 319                         if (itTypes != outputTypes.end()) {
 320                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 321                         }
 322
 323                         itTypes = outputTypes.find("list");
 324                         if (itTypes != outputTypes.end()) {
 325                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 326                         }
 327
 328                         itTypes = outputTypes.find("taxonomy");
 329                         if (itTypes != outputTypes.end()) {
 330                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 331                         }
 332
 333                         itTypes = outputTypes.find("qfile");
 334                         if (itTypes != outputTypes.end()) {
 335                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 336                         }
 337                 }
 338
 339                 return 0;
 340         }
 341
 342         catch(exception& e) {
 343                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 344                 exit(1);
 345         }
 346 }
 347
 348 //**********************************************************************************************************************
 349 int RemoveSeqsCommand::readFasta(){
 350         try {
 351                 string thisOutputDir = outputDir;
 352                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 353                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta", fastafile);
 354
 355                 ofstream out;
 356                 m->openOutputFile(outputFileName, out);
 357
 358                 ifstream in;
 359                 m->openInputFile(fastafile, in);
 360                 string name;
 361
 362                 bool wroteSomething = false;
 363                 int removedCount = 0;
 364
 365                 while(!in.eof()){
 366                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 367
 368                         Sequence currSeq(in);
 369                         name = currSeq.getName();
 370
 371                         if (name != "") {
 372                                 //if this name is in the accnos file
 373                                 if (names.count(name) == 0) {
 374                                         wroteSomething = true;
 375
 376                                         currSeq.printSequence(out);
 377                                 }else {  removedCount++;  }
 378                         }
 379                         m->gobble(in);
 380                 }
 381                 in.close();
 382                 out.close();
 383
 384                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 385                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 386
 387                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 388
 389                 return 0;
 390
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 394                 exit(1);
 395         }
 396 }
 397 //**********************************************************************************************************************
 398 int RemoveSeqsCommand::readQual(){
 399         try {
 400                 string thisOutputDir = outputDir;
 401                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 402                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + getOutputFileNameTag("qfile", qualfile);
 403                 ofstream out;
 404                 m->openOutputFile(outputFileName, out);
 405
 406
 407                 ifstream in;
 408                 m->openInputFile(qualfile, in);
 409                 string name;
 410
 411                 bool wroteSomething = false;
 412                 int removedCount = 0;
 413
 414
 415                 while(!in.eof()){
 416                         string saveName = "";
 417                         string name = "";
 418                         string scores = "";
 419
 420                         in >> name;
 421
 422                         if (name.length() != 0) {
 423                                 saveName = name.substr(1);
 424                                 while (!in.eof())       {
 425                                         char c = in.get();
 426                                         if (c == 10 || c == 13){        break;  }
 427                                         else { name += c; }
 428                                 }
 429                                 m->gobble(in);
 430                         }
 431
 432                         while(in){
 433                                 char letter= in.get();
 434                                 if(letter == '>'){      in.putback(letter);     break;  }
 435                                 else{ scores += letter; }
 436                         }
 437
 438                         m->gobble(in);
 439
 440                         if (names.count(saveName) == 0) {
 441                                 wroteSomething = true;
 442
 443                                 out << name << endl << scores;
 444                         }else {  removedCount++;  }
 445
 446                         m->gobble(in);
 447                 }
 448                 in.close();
 449                 out.close();
 450
 451
 452                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 453                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 454
 455                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 456
 457                 return 0;
 458
 459         }
 460         catch(exception& e) {
 461                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 462                 exit(1);
 463         }
 464 }
 465 //**********************************************************************************************************************
 466 int RemoveSeqsCommand::readList(){
 467         try {
 468                 string thisOutputDir = outputDir;
 469                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 470                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + getOutputFileNameTag("list", listfile);
 471                 ofstream out;
 472                 m->openOutputFile(outputFileName, out);
 473
 474                 ifstream in;
 475                 m->openInputFile(listfile, in);
 476
 477                 bool wroteSomething = false;
 478                 int removedCount = 0;
 479
 480                 while(!in.eof()){
 481
 482                         removedCount = 0;
 483
 484                         //read in list vector
 485                         ListVector list(in);
 486
 487                         //make a new list vector
 488                         ListVector newList;
 489                         newList.setLabel(list.getLabel());
 490
 491                         //for each bin
 492                         for (int i = 0; i < list.getNumBins(); i++) {
 493                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 494
 495                                 //parse out names that are in accnos file
 496                                 string binnames = list.get(i);
 497
 498                                 string newNames = "";
 499                                 while (binnames.find_first_of(',') != -1) {
 500                                         string name = binnames.substr(0,binnames.find_first_of(','));
 501                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 502
 503                                         //if that name is in the .accnos file, add it
 504                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 505                                         else {  removedCount++;  }
 506                                 }
 507
 508                                 //get last name
 509                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 510                                 else {  removedCount++;  }
 511
 512                                 //if there are names in this bin add to new list
 513                                 if (newNames != "") {
 514                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 515                                         newList.push_back(newNames);
 516                                 }
 517                         }
 518
 519                         //print new listvector
 520                         if (newList.getNumBins() != 0) {
 521                                 wroteSomething = true;
 522                                 newList.print(out);
 523                         }
 524
 525                         m->gobble(in);
 526                 }
 527                 in.close();
 528                 out.close();
 529
 530                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 531                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 532
 533                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 534
 535                 return 0;
 536
 537         }
 538         catch(exception& e) {
 539                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 540                 exit(1);
 541         }
 542 }
 543 //**********************************************************************************************************************
 544 int RemoveSeqsCommand::readName(){
 545         try {
 546                 string thisOutputDir = outputDir;
 547                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 548                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + getOutputFileNameTag("name", namefile);
 549                 ofstream out;
 550                 m->openOutputFile(outputFileName, out);
 551
 552                 ifstream in;
 553                 m->openInputFile(namefile, in);
 554                 string name, firstCol, secondCol;
 555
 556                 bool wroteSomething = false;
 557                 int removedCount = 0;
 558
 559                 while(!in.eof()){
 560                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 561
 562                         in >> firstCol;         m->gobble(in);
 563                         in >> secondCol;
 564
 565                         vector<string> parsedNames;
 566                         m->splitAtComma(secondCol, parsedNames);
 567
 568                         vector<string> validSecond;  validSecond.clear();
 569                         for (int i = 0; i < parsedNames.size(); i++) {
 570                                 if (names.count(parsedNames[i]) == 0) {
 571                                         validSecond.push_back(parsedNames[i]);
 572                                 }
 573                         }
 574
 575                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 576                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 577                                 removedCount += parsedNames.size();
 578                         }else {
 579                                 removedCount += parsedNames.size()-validSecond.size();
 580                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 581                                 if (names.count(firstCol) == 0) {
 582
 583                                         wroteSomething = true;
 584
 585                                         out << firstCol << '\t';
 586
 587                                         //you know you have at least one valid second since first column is valid
 588                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 589                                         out << validSecond[validSecond.size()-1] << endl;
 590
 591                                         //make first name in set you come to first column and then add the remaining names to second column
 592                                 }else {
 593
 594                                         //you want part of this row
 595                                         if (validSecond.size() != 0) {
 596
 597                                                 wroteSomething = true;
 598
 599                                                 out << validSecond[0] << '\t';
 600
 601                                                 //you know you have at least one valid second since first column is valid
 602                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 603                                                 out << validSecond[validSecond.size()-1] << endl;
 604                                         }
 605                                 }
 606                         }
 607                         m->gobble(in);
 608                 }
 609                 in.close();
 610                 out.close();
 611
 612                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 613                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 614
 615                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 616
 617                 return 0;
 618         }
 619         catch(exception& e) {
 620                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 621                 exit(1);
 622         }
 623 }
 624
 625 //**********************************************************************************************************************
 626 int RemoveSeqsCommand::readGroup(){
 627         try {
 628                 string thisOutputDir = outputDir;
 629                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 630                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
 631                 ofstream out;
 632                 m->openOutputFile(outputFileName, out);
 633
 634                 ifstream in;
 635                 m->openInputFile(groupfile, in);
 636                 string name, group;
 637
 638                 bool wroteSomething = false;
 639                 int removedCount = 0;
 640
 641                 while(!in.eof()){
 642                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 643
 644                         in >> name;                             //read from first column
 645                         in >> group;                    //read from second column
 646
 647                         //if this name is in the accnos file
 648                         if (names.count(name) == 0) {
 649                                 wroteSomething = true;
 650                                 out << name << '\t' << group << endl;
 651                         }else {  removedCount++;  }
 652
 653                         m->gobble(in);
 654                 }
 655                 in.close();
 656                 out.close();
 657
 658                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 659                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 660
 661                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 662
 663
 664                 return 0;
 665         }
 666         catch(exception& e) {
 667                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 668                 exit(1);
 669         }
 670 }
 671 //**********************************************************************************************************************
 672 int RemoveSeqsCommand::readTax(){
 673         try {
 674                 string thisOutputDir = outputDir;
 675                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 676                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + getOutputFileNameTag("taxonomy", taxfile);
 677                 ofstream out;
 678                 m->openOutputFile(outputFileName, out);
 679
 680                 ifstream in;
 681                 m->openInputFile(taxfile, in);
 682                 string name, tax;
 683
 684                 bool wroteSomething = false;
 685                 int removedCount = 0;
 686
 687                 while(!in.eof()){
 688                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 689
 690                         in >> name;                             //read from first column
 691                         in >> tax;                      //read from second column
 692
 693                         //if this name is in the accnos file
 694                         if (names.count(name) == 0) {
 695                                 wroteSomething = true;
 696                                 out << name << '\t' << tax << endl;
 697                         }else {  removedCount++;  }
 698
 699                         m->gobble(in);
 700                 }
 701                 in.close();
 702                 out.close();
 703
 704                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 705                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 706
 707                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 708
 709                 return 0;
 710         }
 711         catch(exception& e) {
 712                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 713                 exit(1);
 714         }
 715 }
 716 //**********************************************************************************************************************
 717 //alignreport file has a column header line then all other lines contain 16 columns.  we just want the first column since that contains the name
 718 int RemoveSeqsCommand::readAlign(){
 719         try {
 720                 string thisOutputDir = outputDir;
 721                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 722                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + getOutputFileNameTag("alignreport");
 723
 724                 ofstream out;
 725                 m->openOutputFile(outputFileName, out);
 726
 727                 ifstream in;
 728                 m->openInputFile(alignfile, in);
 729                 string name, junk;
 730
 731                 bool wroteSomething = false;
 732                 int removedCount = 0;
 733
 734                 //read column headers
 735                 for (int i = 0; i < 16; i++) {
 736                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 737                         else                    {       break;                  }
 738                 }
 739                 out << endl;
 740
 741                 while(!in.eof()){
 742                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 743
 744                         in >> name;                             //read from first column
 745
 746                         //if this name is in the accnos file
 747                         if (names.count(name) == 0) {
 748                                 wroteSomething = true;
 749
 750                                 out << name << '\t';
 751
 752                                 //read rest
 753                                 for (int i = 0; i < 15; i++) {
 754                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 755                                         else                    {       break;                  }
 756                                 }
 757                                 out << endl;
 758
 759                         }else {//still read just don't do anything with it
 760                                 removedCount++;
 761
 762                                 //read rest
 763                                 for (int i = 0; i < 15; i++) {
 764                                         if (!in.eof())  {       in >> junk;             }
 765                                         else                    {       break;                  }
 766                                 }
 767                         }
 768
 769                         m->gobble(in);
 770                 }
 771                 in.close();
 772                 out.close();
 773
 774                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 775                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 776
 777                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 778
 779
 780                 return 0;
 781
 782         }
 783         catch(exception& e) {
 784                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 785                 exit(1);
 786         }
 787 }
 788 //**********************************************************************************************************************
 789
 790