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                         }else { m->setAccnosFile(accnosfile); }
 195
 196                         fastafile = validParameter.validFile(parameters, "fasta", true);
 197                         if (fastafile == "not open") { abort = true; }
 198                         else if (fastafile == "not found") {  fastafile = "";  }
 199                         else { m->setFastaFile(fastafile); }
 200
 201                         namefile = validParameter.validFile(parameters, "name", true);
 202                         if (namefile == "not open") { abort = true; }
 203                         else if (namefile == "not found") {  namefile = "";  }
 204                         else { m->setNameFile(namefile); }
 205
 206                         groupfile = validParameter.validFile(parameters, "group", true);
 207                         if (groupfile == "not open") { abort = true; }
 208                         else if (groupfile == "not found") {  groupfile = "";  }
 209                         else { m->setGroupFile(groupfile); }
 210
 211                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 212                         if (alignfile == "not open") { abort = true; }
 213                         else if (alignfile == "not found") {  alignfile = "";  }
 214
 215                         listfile = validParameter.validFile(parameters, "list", true);
 216                         if (listfile == "not open") { abort = true; }
 217                         else if (listfile == "not found") {  listfile = "";  }
 218                         else { m->setListFile(listfile); }
 219
 220                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 221                         if (taxfile == "not open") { abort = true; }
 222                         else if (taxfile == "not found") {  taxfile = "";  }
 223                         else { m->setTaxonomyFile(taxfile); }
 224
 225                         qualfile = validParameter.validFile(parameters, "qfile", true);
 226                         if (qualfile == "not open") { abort = true; }
 227                         else if (qualfile == "not found") {  qualfile = "";  }
 228                         else { m->setQualFile(qualfile); }
 229
 230                         string usedDups = "true";
 231                         string temp = validParameter.validFile(parameters, "dups", false);
 232                         if (temp == "not found") {
 233                                 if (namefile != "") {  temp = "true";                                   }
 234                                 else                            {  temp = "false"; usedDups = "";       }
 235                         }
 236                         dups = m->isTrue(temp);
 237
 238                         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; }
 239
 240                 }
 241
 242         }
 243         catch(exception& e) {
 244                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 245                 exit(1);
 246         }
 247 }
 248 //**********************************************************************************************************************
 249
 250 int RemoveSeqsCommand::execute(){
 251         try {
 252
 253                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 254
 255                 //get names you want to keep
 256                 readAccnos();
 257
 258                 if (m->control_pressed) { return 0; }
 259
 260                 //read through the correct file and output lines you want to keep
 261                 if (namefile != "")                     {               readName();             }
 262                 if (fastafile != "")            {               readFasta();    }
 263                 if (groupfile != "")            {               readGroup();    }
 264                 if (alignfile != "")            {               readAlign();    }
 265                 if (listfile != "")                     {               readList();             }
 266                 if (taxfile != "")                      {               readTax();              }
 267                 if (qualfile != "")                     {               readQual();             }
 268
 269                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 270
 271                 m->mothurOut("Removed " + toString(names.size()) + " sequences."); m->mothurOutEndLine();
 272
 273                 if (outputNames.size() != 0) {
 274                         m->mothurOutEndLine();
 275                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 276                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 277                         m->mothurOutEndLine();
 278
 279                         //set fasta file as new current fastafile
 280                         string current = "";
 281                         itTypes = outputTypes.find("fasta");
 282                         if (itTypes != outputTypes.end()) {
 283                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 284                         }
 285
 286                         itTypes = outputTypes.find("name");
 287                         if (itTypes != outputTypes.end()) {
 288                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 289                         }
 290
 291                         itTypes = outputTypes.find("group");
 292                         if (itTypes != outputTypes.end()) {
 293                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 294                         }
 295
 296                         itTypes = outputTypes.find("list");
 297                         if (itTypes != outputTypes.end()) {
 298                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 299                         }
 300
 301                         itTypes = outputTypes.find("taxonomy");
 302                         if (itTypes != outputTypes.end()) {
 303                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 304                         }
 305
 306                         itTypes = outputTypes.find("qfile");
 307                         if (itTypes != outputTypes.end()) {
 308                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 309                         }
 310                 }
 311
 312                 return 0;
 313         }
 314
 315         catch(exception& e) {
 316                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 317                 exit(1);
 318         }
 319 }
 320
 321 //**********************************************************************************************************************
 322 int RemoveSeqsCommand::readFasta(){
 323         try {
 324                 string thisOutputDir = outputDir;
 325                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 326                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 327
 328                 ofstream out;
 329                 m->openOutputFile(outputFileName, out);
 330
 331                 ifstream in;
 332                 m->openInputFile(fastafile, in);
 333                 string name;
 334
 335                 bool wroteSomething = false;
 336
 337                 while(!in.eof()){
 338                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 339
 340                         Sequence currSeq(in);
 341                         name = currSeq.getName();
 342
 343                         if (name != "") {
 344                                 //if this name is in the accnos file
 345                                 if (names.count(name) == 0) {
 346                                         wroteSomething = true;
 347
 348                                         currSeq.printSequence(out);
 349                                 }
 350                         }
 351                         m->gobble(in);
 352                 }
 353                 in.close();
 354                 out.close();
 355
 356                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 357                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 358
 359                 return 0;
 360
 361         }
 362         catch(exception& e) {
 363                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 364                 exit(1);
 365         }
 366 }
 367 //**********************************************************************************************************************
 368 int RemoveSeqsCommand::readQual(){
 369         try {
 370                 string thisOutputDir = outputDir;
 371                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 372                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 373                 ofstream out;
 374                 m->openOutputFile(outputFileName, out);
 375
 376
 377                 ifstream in;
 378                 m->openInputFile(qualfile, in);
 379                 string name;
 380
 381                 bool wroteSomething = false;
 382
 383
 384                 while(!in.eof()){
 385                         string saveName = "";
 386                         string name = "";
 387                         string scores = "";
 388
 389                         in >> name;
 390
 391                         if (name.length() != 0) {
 392                                 saveName = name.substr(1);
 393                                 while (!in.eof())       {
 394                                         char c = in.get();
 395                                         if (c == 10 || c == 13){        break;  }
 396                                         else { name += c; }
 397                                 }
 398                                 m->gobble(in);
 399                         }
 400
 401                         while(in){
 402                                 char letter= in.get();
 403                                 if(letter == '>'){      in.putback(letter);     break;  }
 404                                 else{ scores += letter; }
 405                         }
 406
 407                         m->gobble(in);
 408
 409                         if (names.count(saveName) == 0) {
 410                                 wroteSomething = true;
 411
 412                                 out << name << endl << scores;
 413                         }
 414
 415                         m->gobble(in);
 416                 }
 417                 in.close();
 418                 out.close();
 419
 420
 421                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 422                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 423
 424                 return 0;
 425
 426         }
 427         catch(exception& e) {
 428                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 429                 exit(1);
 430         }
 431 }
 432 //**********************************************************************************************************************
 433 int RemoveSeqsCommand::readList(){
 434         try {
 435                 string thisOutputDir = outputDir;
 436                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 437                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 438
 439                 ofstream out;
 440                 m->openOutputFile(outputFileName, out);
 441
 442                 ifstream in;
 443                 m->openInputFile(listfile, in);
 444
 445                 bool wroteSomething = false;
 446
 447                 while(!in.eof()){
 448                         //read in list vector
 449                         ListVector list(in);
 450
 451                         //make a new list vector
 452                         ListVector newList;
 453                         newList.setLabel(list.getLabel());
 454
 455                         //for each bin
 456                         for (int i = 0; i < list.getNumBins(); i++) {
 457                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 458
 459                                 //parse out names that are in accnos file
 460                                 string binnames = list.get(i);
 461
 462                                 string newNames = "";
 463                                 while (binnames.find_first_of(',') != -1) {
 464                                         string name = binnames.substr(0,binnames.find_first_of(','));
 465                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 466
 467                                         //if that name is in the .accnos file, add it
 468                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 469                                 }
 470
 471                                 //get last name
 472                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 473
 474                                 //if there are names in this bin add to new list
 475                                 if (newNames != "") {
 476                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 477                                         newList.push_back(newNames);
 478                                 }
 479                         }
 480
 481                         //print new listvector
 482                         if (newList.getNumBins() != 0) {
 483                                 wroteSomething = true;
 484                                 newList.print(out);
 485                         }
 486
 487                         m->gobble(in);
 488                 }
 489                 in.close();
 490                 out.close();
 491
 492                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 493                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 494
 495                 return 0;
 496
 497         }
 498         catch(exception& e) {
 499                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 500                 exit(1);
 501         }
 502 }
 503 //**********************************************************************************************************************
 504 int RemoveSeqsCommand::readName(){
 505         try {
 506                 string thisOutputDir = outputDir;
 507                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 508                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 509
 510                 ofstream out;
 511                 m->openOutputFile(outputFileName, out);
 512
 513                 ifstream in;
 514                 m->openInputFile(namefile, in);
 515                 string name, firstCol, secondCol;
 516
 517                 bool wroteSomething = false;
 518
 519
 520                 while(!in.eof()){
 521                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 522
 523                         in >> firstCol;         m->gobble(in);
 524                         in >> secondCol;
 525
 526                         vector<string> parsedNames;
 527                         m->splitAtComma(secondCol, parsedNames);
 528
 529                         vector<string> validSecond;  validSecond.clear();
 530                         for (int i = 0; i < parsedNames.size(); i++) {
 531                                 if (names.count(parsedNames[i]) == 0) {
 532                                         validSecond.push_back(parsedNames[i]);
 533                                 }
 534                         }
 535
 536                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 537                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 538                         }else {
 539                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 540                                 if (names.count(firstCol) == 0) {
 541
 542                                         wroteSomething = true;
 543
 544                                         out << firstCol << '\t';
 545
 546                                         //you know you have at least one valid second since first column is valid
 547                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 548                                         out << validSecond[validSecond.size()-1] << endl;
 549
 550                                         //make first name in set you come to first column and then add the remaining names to second column
 551                                 }else {
 552
 553                                         //you want part of this row
 554                                         if (validSecond.size() != 0) {
 555
 556                                                 wroteSomething = true;
 557
 558                                                 out << validSecond[0] << '\t';
 559
 560                                                 //you know you have at least one valid second since first column is valid
 561                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 562                                                 out << validSecond[validSecond.size()-1] << endl;
 563                                         }
 564                                 }
 565                         }
 566                         m->gobble(in);
 567                 }
 568                 in.close();
 569                 out.close();
 570
 571                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 572                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 573
 574                 return 0;
 575         }
 576         catch(exception& e) {
 577                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 578                 exit(1);
 579         }
 580 }
 581
 582 //**********************************************************************************************************************
 583 int RemoveSeqsCommand::readGroup(){
 584         try {
 585                 string thisOutputDir = outputDir;
 586                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 587                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 588
 589                 ofstream out;
 590                 m->openOutputFile(outputFileName, out);
 591
 592                 ifstream in;
 593                 m->openInputFile(groupfile, in);
 594                 string name, group;
 595
 596                 bool wroteSomething = false;
 597
 598                 while(!in.eof()){
 599                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 600
 601                         in >> name;                             //read from first column
 602                         in >> group;                    //read from second column
 603
 604                         //if this name is in the accnos file
 605                         if (names.count(name) == 0) {
 606                                 wroteSomething = true;
 607                                 out << name << '\t' << group << endl;
 608                         }
 609
 610                         m->gobble(in);
 611                 }
 612                 in.close();
 613                 out.close();
 614
 615                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 616                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 617
 618                 return 0;
 619         }
 620         catch(exception& e) {
 621                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 622                 exit(1);
 623         }
 624 }
 625 //**********************************************************************************************************************
 626 int RemoveSeqsCommand::readTax(){
 627         try {
 628                 string thisOutputDir = outputDir;
 629                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 630                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 631                 ofstream out;
 632                 m->openOutputFile(outputFileName, out);
 633
 634                 ifstream in;
 635                 m->openInputFile(taxfile, in);
 636                 string name, tax;
 637
 638                 bool wroteSomething = false;
 639
 640                 while(!in.eof()){
 641                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 642
 643                         in >> name;                             //read from first column
 644                         in >> tax;                      //read from second column
 645
 646                         //if this name is in the accnos file
 647                         if (names.count(name) == 0) {
 648                                 wroteSomething = true;
 649                                 out << name << '\t' << tax << endl;
 650                         }
 651
 652                         m->gobble(in);
 653                 }
 654                 in.close();
 655                 out.close();
 656
 657                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 658                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 659
 660                 return 0;
 661         }
 662         catch(exception& e) {
 663                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 664                 exit(1);
 665         }
 666 }
 667 //**********************************************************************************************************************
 668 //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
 669 int RemoveSeqsCommand::readAlign(){
 670         try {
 671                 string thisOutputDir = outputDir;
 672                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 673                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 674
 675                 ofstream out;
 676                 m->openOutputFile(outputFileName, out);
 677
 678                 ifstream in;
 679                 m->openInputFile(alignfile, in);
 680                 string name, junk;
 681
 682                 bool wroteSomething = false;
 683
 684                 //read column headers
 685                 for (int i = 0; i < 16; i++) {
 686                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 687                         else                    {       break;                  }
 688                 }
 689                 out << endl;
 690
 691                 while(!in.eof()){
 692                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 693
 694                         in >> name;                             //read from first column
 695
 696                         //if this name is in the accnos file
 697                         if (names.count(name) == 0) {
 698                                 wroteSomething = true;
 699
 700                                 out << name << '\t';
 701
 702                                 //read rest
 703                                 for (int i = 0; i < 15; i++) {
 704                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 705                                         else                    {       break;                  }
 706                                 }
 707                                 out << endl;
 708
 709                         }else {//still read just don't do anything with it
 710
 711                                 //read rest
 712                                 for (int i = 0; i < 15; i++) {
 713                                         if (!in.eof())  {       in >> junk;             }
 714                                         else                    {       break;                  }
 715                                 }
 716                         }
 717
 718                         m->gobble(in);
 719                 }
 720                 in.close();
 721                 out.close();
 722
 723                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 724                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 725
 726                 return 0;
 727
 728         }
 729         catch(exception& e) {
 730                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 731                 exit(1);
 732         }
 733 }
 734 //**********************************************************************************************************************
 735 void RemoveSeqsCommand::readAccnos(){
 736         try {
 737
 738                 ifstream in;
 739                 m->openInputFile(accnosfile, in);
 740                 string name;
 741
 742                 while(!in.eof()){
 743                         in >> name;
 744
 745                         names.insert(name);
 746
 747                         m->gobble(in);
 748                 }
 749                 in.close();
 750
 751         }
 752         catch(exception& e) {
 753                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 754                 exit(1);
 755         }
 756 }
 757
 758 //**********************************************************************************************************************
 759
 760