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