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
  58 //**********************************************************************************************************************
  59 RemoveSeqsCommand::RemoveSeqsCommand(){
  60         try {
  61                 abort = true; calledHelp = true;
  62                 setParameters();
  63                 vector<string> tempOutNames;
  64                 outputTypes["fasta"] = tempOutNames;
  65                 outputTypes["taxonomy"] = tempOutNames;
  66                 outputTypes["name"] = tempOutNames;
  67                 outputTypes["group"] = tempOutNames;
  68                 outputTypes["alignreport"] = tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70                 outputTypes["qfile"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["fasta"] = tempOutNames;
 103                         outputTypes["taxonomy"] = tempOutNames;
 104                         outputTypes["name"] = tempOutNames;
 105                         outputTypes["group"] = tempOutNames;
 106                         outputTypes["alignreport"] = tempOutNames;
 107                         outputTypes["list"] = tempOutNames;
 108                         outputTypes["qfile"] = tempOutNames;
 109
 110                         //if the user changes the output directory command factory will send this info to us in the output parameter
 111                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 112
 113                         //if the user changes the input directory command factory will send this info to us in the output parameter
 114                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 115                         if (inputDir == "not found"){   inputDir = "";          }
 116                         else {
 117                                 string path;
 118                                 it = parameters.find("alignreport");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 124                                 }
 125
 126                                 it = parameters.find("fasta");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 132                                 }
 133
 134                                 it = parameters.find("accnos");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 140                                 }
 141
 142                                 it = parameters.find("list");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 148                                 }
 149
 150                                 it = parameters.find("name");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 156                                 }
 157
 158                                 it = parameters.find("group");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 164                                 }
 165
 166                                 it = parameters.find("taxonomy");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 172                                 }
 173
 174                                 it = parameters.find("qfile");
 175                                 //user has given a template file
 176                                 if(it != parameters.end()){
 177                                         path = m->hasPath(it->second);
 178                                         //if the user has not given a path then, add inputdir. else leave path alone.
 179                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 180                                 }
 181                         }
 182
 183
 184                         //check for required parameters
 185                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 186                         if (accnosfile == "not open") { abort = true; }
 187                         else if (accnosfile == "not found") {
 188                                 accnosfile = m->getAccnosFile();
 189                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
 190                                 else {
 191                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
 192                                         abort = true;
 193                                 }
 194                         }
 195
 196                         fastafile = validParameter.validFile(parameters, "fasta", true);
 197                         if (fastafile == "not open") { abort = true; }
 198                         else if (fastafile == "not found") {  fastafile = "";  }
 199
 200                         namefile = validParameter.validFile(parameters, "name", true);
 201                         if (namefile == "not open") { abort = true; }
 202                         else if (namefile == "not found") {  namefile = "";  }
 203
 204                         groupfile = validParameter.validFile(parameters, "group", true);
 205                         if (groupfile == "not open") { abort = true; }
 206                         else if (groupfile == "not found") {  groupfile = "";  }
 207
 208                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 209                         if (alignfile == "not open") { abort = true; }
 210                         else if (alignfile == "not found") {  alignfile = "";  }
 211
 212                         listfile = validParameter.validFile(parameters, "list", true);
 213                         if (listfile == "not open") { abort = true; }
 214                         else if (listfile == "not found") {  listfile = "";  }
 215
 216                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 217                         if (taxfile == "not open") { abort = true; }
 218                         else if (taxfile == "not found") {  taxfile = "";  }
 219
 220                         qualfile = validParameter.validFile(parameters, "qfile", true);
 221                         if (qualfile == "not open") { abort = true; }
 222                         else if (qualfile == "not found") {  qualfile = "";  }
 223
 224
 225                         string usedDups = "true";
 226                         string temp = validParameter.validFile(parameters, "dups", false);
 227                         if (temp == "not found") {
 228                                 if (namefile != "") {  temp = "true";                                   }
 229                                 else                            {  temp = "false"; usedDups = "";       }
 230                         }
 231                         dups = m->isTrue(temp);
 232
 233                         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; }
 234
 235                 }
 236
 237         }
 238         catch(exception& e) {
 239                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 240                 exit(1);
 241         }
 242 }
 243 //**********************************************************************************************************************
 244
 245 int RemoveSeqsCommand::execute(){
 246         try {
 247
 248                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 249
 250                 //get names you want to keep
 251                 readAccnos();
 252
 253                 if (m->control_pressed) { return 0; }
 254
 255                 //read through the correct file and output lines you want to keep
 256                 if (namefile != "")                     {               readName();             }
 257                 if (fastafile != "")            {               readFasta();    }
 258                 if (groupfile != "")            {               readGroup();    }
 259                 if (alignfile != "")            {               readAlign();    }
 260                 if (listfile != "")                     {               readList();             }
 261                 if (taxfile != "")                      {               readTax();              }
 262                 if (qualfile != "")                     {               readQual();             }
 263
 264                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 265
 266                 m->mothurOut("Removed " + toString(names.size()) + " sequences."); m->mothurOutEndLine();
 267
 268                 if (outputNames.size() != 0) {
 269                         m->mothurOutEndLine();
 270                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 271                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 272                         m->mothurOutEndLine();
 273
 274                         //set fasta file as new current fastafile
 275                         string current = "";
 276                         itTypes = outputTypes.find("fasta");
 277                         if (itTypes != outputTypes.end()) {
 278                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 279                         }
 280
 281                         itTypes = outputTypes.find("name");
 282                         if (itTypes != outputTypes.end()) {
 283                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 284                         }
 285
 286                         itTypes = outputTypes.find("group");
 287                         if (itTypes != outputTypes.end()) {
 288                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 289                         }
 290
 291                         itTypes = outputTypes.find("list");
 292                         if (itTypes != outputTypes.end()) {
 293                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 294                         }
 295
 296                         itTypes = outputTypes.find("taxonomy");
 297                         if (itTypes != outputTypes.end()) {
 298                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 299                         }
 300
 301                         itTypes = outputTypes.find("qfile");
 302                         if (itTypes != outputTypes.end()) {
 303                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 304                         }
 305                 }
 306
 307                 return 0;
 308         }
 309
 310         catch(exception& e) {
 311                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 312                 exit(1);
 313         }
 314 }
 315
 316 //**********************************************************************************************************************
 317 int RemoveSeqsCommand::readFasta(){
 318         try {
 319                 string thisOutputDir = outputDir;
 320                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 321                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 322
 323                 ofstream out;
 324                 m->openOutputFile(outputFileName, out);
 325
 326                 ifstream in;
 327                 m->openInputFile(fastafile, in);
 328                 string name;
 329
 330                 bool wroteSomething = false;
 331
 332                 while(!in.eof()){
 333                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 334
 335                         Sequence currSeq(in);
 336                         name = currSeq.getName();
 337
 338                         if (name != "") {
 339                                 //if this name is in the accnos file
 340                                 if (names.count(name) == 0) {
 341                                         wroteSomething = true;
 342
 343                                         currSeq.printSequence(out);
 344                                 }
 345                         }
 346                         m->gobble(in);
 347                 }
 348                 in.close();
 349                 out.close();
 350
 351                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 352                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 353
 354                 return 0;
 355
 356         }
 357         catch(exception& e) {
 358                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 359                 exit(1);
 360         }
 361 }
 362 //**********************************************************************************************************************
 363 int RemoveSeqsCommand::readQual(){
 364         try {
 365                 string thisOutputDir = outputDir;
 366                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 367                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 368                 ofstream out;
 369                 m->openOutputFile(outputFileName, out);
 370
 371
 372                 ifstream in;
 373                 m->openInputFile(qualfile, in);
 374                 string name;
 375
 376                 bool wroteSomething = false;
 377
 378
 379                 while(!in.eof()){
 380                         string saveName = "";
 381                         string name = "";
 382                         string scores = "";
 383
 384                         in >> name;
 385
 386                         if (name.length() != 0) {
 387                                 saveName = name.substr(1);
 388                                 while (!in.eof())       {
 389                                         char c = in.get();
 390                                         if (c == 10 || c == 13){        break;  }
 391                                         else { name += c; }
 392                                 }
 393                                 m->gobble(in);
 394                         }
 395
 396                         while(in){
 397                                 char letter= in.get();
 398                                 if(letter == '>'){      in.putback(letter);     break;  }
 399                                 else{ scores += letter; }
 400                         }
 401
 402                         m->gobble(in);
 403
 404                         if (names.count(saveName) == 0) {
 405                                 wroteSomething = true;
 406
 407                                 out << name << endl << scores;
 408                         }
 409
 410                         m->gobble(in);
 411                 }
 412                 in.close();
 413                 out.close();
 414
 415
 416                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 417                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 418
 419                 return 0;
 420
 421         }
 422         catch(exception& e) {
 423                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 424                 exit(1);
 425         }
 426 }
 427 //**********************************************************************************************************************
 428 int RemoveSeqsCommand::readList(){
 429         try {
 430                 string thisOutputDir = outputDir;
 431                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 432                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 433
 434                 ofstream out;
 435                 m->openOutputFile(outputFileName, out);
 436
 437                 ifstream in;
 438                 m->openInputFile(listfile, in);
 439
 440                 bool wroteSomething = false;
 441
 442                 while(!in.eof()){
 443                         //read in list vector
 444                         ListVector list(in);
 445
 446                         //make a new list vector
 447                         ListVector newList;
 448                         newList.setLabel(list.getLabel());
 449
 450                         //for each bin
 451                         for (int i = 0; i < list.getNumBins(); i++) {
 452                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 453
 454                                 //parse out names that are in accnos file
 455                                 string binnames = list.get(i);
 456
 457                                 string newNames = "";
 458                                 while (binnames.find_first_of(',') != -1) {
 459                                         string name = binnames.substr(0,binnames.find_first_of(','));
 460                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 461
 462                                         //if that name is in the .accnos file, add it
 463                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 464                                 }
 465
 466                                 //get last name
 467                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 468
 469                                 //if there are names in this bin add to new list
 470                                 if (newNames != "") {
 471                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 472                                         newList.push_back(newNames);
 473                                 }
 474                         }
 475
 476                         //print new listvector
 477                         if (newList.getNumBins() != 0) {
 478                                 wroteSomething = true;
 479                                 newList.print(out);
 480                         }
 481
 482                         m->gobble(in);
 483                 }
 484                 in.close();
 485                 out.close();
 486
 487                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 488                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 489
 490                 return 0;
 491
 492         }
 493         catch(exception& e) {
 494                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 495                 exit(1);
 496         }
 497 }
 498 //**********************************************************************************************************************
 499 int RemoveSeqsCommand::readName(){
 500         try {
 501                 string thisOutputDir = outputDir;
 502                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 503                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 504
 505                 ofstream out;
 506                 m->openOutputFile(outputFileName, out);
 507
 508                 ifstream in;
 509                 m->openInputFile(namefile, in);
 510                 string name, firstCol, secondCol;
 511
 512                 bool wroteSomething = false;
 513
 514
 515                 while(!in.eof()){
 516                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 517
 518                         in >> firstCol;         m->gobble(in);
 519                         in >> secondCol;
 520
 521                         vector<string> parsedNames;
 522                         m->splitAtComma(secondCol, parsedNames);
 523
 524                         vector<string> validSecond;  validSecond.clear();
 525                         for (int i = 0; i < parsedNames.size(); i++) {
 526                                 if (names.count(parsedNames[i]) == 0) {
 527                                         validSecond.push_back(parsedNames[i]);
 528                                 }
 529                         }
 530
 531                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 532                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 533                         }else {
 534                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 535                                 if (names.count(firstCol) == 0) {
 536
 537                                         wroteSomething = true;
 538
 539                                         out << firstCol << '\t';
 540
 541                                         //you know you have at least one valid second since first column is valid
 542                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 543                                         out << validSecond[validSecond.size()-1] << endl;
 544
 545                                         //make first name in set you come to first column and then add the remaining names to second column
 546                                 }else {
 547
 548                                         //you want part of this row
 549                                         if (validSecond.size() != 0) {
 550
 551                                                 wroteSomething = true;
 552
 553                                                 out << validSecond[0] << '\t';
 554
 555                                                 //you know you have at least one valid second since first column is valid
 556                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 557                                                 out << validSecond[validSecond.size()-1] << endl;
 558                                         }
 559                                 }
 560                         }
 561                         m->gobble(in);
 562                 }
 563                 in.close();
 564                 out.close();
 565
 566                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 567                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 568
 569                 return 0;
 570         }
 571         catch(exception& e) {
 572                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 573                 exit(1);
 574         }
 575 }
 576
 577 //**********************************************************************************************************************
 578 int RemoveSeqsCommand::readGroup(){
 579         try {
 580                 string thisOutputDir = outputDir;
 581                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 582                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 583
 584                 ofstream out;
 585                 m->openOutputFile(outputFileName, out);
 586
 587                 ifstream in;
 588                 m->openInputFile(groupfile, in);
 589                 string name, group;
 590
 591                 bool wroteSomething = false;
 592
 593                 while(!in.eof()){
 594                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 595
 596                         in >> name;                             //read from first column
 597                         in >> group;                    //read from second column
 598
 599                         //if this name is in the accnos file
 600                         if (names.count(name) == 0) {
 601                                 wroteSomething = true;
 602                                 out << name << '\t' << group << endl;
 603                         }
 604
 605                         m->gobble(in);
 606                 }
 607                 in.close();
 608                 out.close();
 609
 610                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 611                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 612
 613                 return 0;
 614         }
 615         catch(exception& e) {
 616                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 617                 exit(1);
 618         }
 619 }
 620 //**********************************************************************************************************************
 621 int RemoveSeqsCommand::readTax(){
 622         try {
 623                 string thisOutputDir = outputDir;
 624                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 625                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 626                 ofstream out;
 627                 m->openOutputFile(outputFileName, out);
 628
 629                 ifstream in;
 630                 m->openInputFile(taxfile, in);
 631                 string name, tax;
 632
 633                 bool wroteSomething = false;
 634
 635                 while(!in.eof()){
 636                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 637
 638                         in >> name;                             //read from first column
 639                         in >> tax;                      //read from second column
 640
 641                         //if this name is in the accnos file
 642                         if (names.count(name) == 0) {
 643                                 wroteSomething = true;
 644                                 out << name << '\t' << tax << endl;
 645                         }
 646
 647                         m->gobble(in);
 648                 }
 649                 in.close();
 650                 out.close();
 651
 652                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 653                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 654
 655                 return 0;
 656         }
 657         catch(exception& e) {
 658                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 659                 exit(1);
 660         }
 661 }
 662 //**********************************************************************************************************************
 663 //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
 664 int RemoveSeqsCommand::readAlign(){
 665         try {
 666                 string thisOutputDir = outputDir;
 667                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 668                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 669
 670                 ofstream out;
 671                 m->openOutputFile(outputFileName, out);
 672
 673                 ifstream in;
 674                 m->openInputFile(alignfile, in);
 675                 string name, junk;
 676
 677                 bool wroteSomething = false;
 678
 679                 //read column headers
 680                 for (int i = 0; i < 16; i++) {
 681                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 682                         else                    {       break;                  }
 683                 }
 684                 out << endl;
 685
 686                 while(!in.eof()){
 687                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 688
 689                         in >> name;                             //read from first column
 690
 691                         //if this name is in the accnos file
 692                         if (names.count(name) == 0) {
 693                                 wroteSomething = true;
 694
 695                                 out << name << '\t';
 696
 697                                 //read rest
 698                                 for (int i = 0; i < 15; i++) {
 699                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 700                                         else                    {       break;                  }
 701                                 }
 702                                 out << endl;
 703
 704                         }else {//still read just don't do anything with it
 705
 706                                 //read rest
 707                                 for (int i = 0; i < 15; i++) {
 708                                         if (!in.eof())  {       in >> junk;             }
 709                                         else                    {       break;                  }
 710                                 }
 711                         }
 712
 713                         m->gobble(in);
 714                 }
 715                 in.close();
 716                 out.close();
 717
 718                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 719                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 720
 721                 return 0;
 722
 723         }
 724         catch(exception& e) {
 725                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 726                 exit(1);
 727         }
 728 }
 729 //**********************************************************************************************************************
 730 void RemoveSeqsCommand::readAccnos(){
 731         try {
 732
 733                 ifstream in;
 734                 m->openInputFile(accnosfile, in);
 735                 string name;
 736
 737                 while(!in.eof()){
 738                         in >> name;
 739
 740                         names.insert(name);
 741
 742                         m->gobble(in);
 743                 }
 744                 in.close();
 745
 746         }
 747         catch(exception& e) {
 748                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 749                 exit(1);
 750         }
 751 }
 752
 753 //**********************************************************************************************************************
 754
 755