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::getValidParameters(){
  16         try {
  17                 string Array[] =  {"fasta","name", "group", "alignreport", "accnos", "qfile","list","taxonomy","outputdir","inputdir", "dups" };
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "RemoveSeqsCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 RemoveSeqsCommand::RemoveSeqsCommand(){
  28         try {
  29                 abort = true;
  30                 //initialize outputTypes
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["taxonomy"] = tempOutNames;
  34                 outputTypes["name"] = tempOutNames;
  35                 outputTypes["group"] = tempOutNames;
  36                 outputTypes["alignreport"] = tempOutNames;
  37                 outputTypes["list"] = tempOutNames;
  38                 outputTypes["qfile"] = tempOutNames;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 vector<string> RemoveSeqsCommand::getRequiredParameters(){
  47         try {
  48                 string Array[] =  {"accnos"};
  49                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "RemoveSeqsCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> RemoveSeqsCommand::getRequiredFiles(){
  59         try {
  60                 vector<string> myArray;
  61                 return myArray;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "RemoveSeqsCommand", "getRequiredFiles");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
  70         try {
  71                 abort = false;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; }
  75
  76                 else {
  77                         //valid paramters for this command
  78                         string Array[] =  {"fasta","name", "group", "alignreport", "accnos", "qfile", "list","taxonomy","outputdir","inputdir", "dups" };
  79                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83
  84                         ValidParameters validParameter;
  85                         map<string,string>::iterator it;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95                         outputTypes["taxonomy"] = tempOutNames;
  96                         outputTypes["name"] = tempOutNames;
  97                         outputTypes["group"] = tempOutNames;
  98                         outputTypes["alignreport"] = tempOutNames;
  99                         outputTypes["list"] = tempOutNames;
 100                         outputTypes["qfile"] = tempOutNames;
 101
 102                         //if the user changes the output directory command factory will send this info to us in the output parameter
 103                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("alignreport");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 116                                 }
 117
 118                                 it = parameters.find("fasta");
 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["fasta"] = inputDir + it->second;            }
 124                                 }
 125
 126                                 it = parameters.find("accnos");
 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["accnos"] = inputDir + it->second;           }
 132                                 }
 133
 134                                 it = parameters.find("list");
 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["list"] = inputDir + it->second;             }
 140                                 }
 141
 142                                 it = parameters.find("name");
 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["name"] = inputDir + it->second;             }
 148                                 }
 149
 150                                 it = parameters.find("group");
 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["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("taxonomy");
 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["taxonomy"] = inputDir + it->second;         }
 164                                 }
 165
 166                                 it = parameters.find("qfile");
 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["qfile"] = inputDir + it->second;            }
 172                                 }
 173                         }
 174
 175
 176                         //check for required parameters
 177                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 178                         if (accnosfile == "not open") { abort = true; }
 179                         else if (accnosfile == "not found") {  accnosfile = "";  m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
 180
 181                         fastafile = validParameter.validFile(parameters, "fasta", true);
 182                         if (fastafile == "not open") { abort = true; }
 183                         else if (fastafile == "not found") {  fastafile = "";  }
 184
 185                         namefile = validParameter.validFile(parameters, "name", true);
 186                         if (namefile == "not open") { abort = true; }
 187                         else if (namefile == "not found") {  namefile = "";  }
 188
 189                         groupfile = validParameter.validFile(parameters, "group", true);
 190                         if (groupfile == "not open") { abort = true; }
 191                         else if (groupfile == "not found") {  groupfile = "";  }
 192
 193                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 194                         if (alignfile == "not open") { abort = true; }
 195                         else if (alignfile == "not found") {  alignfile = "";  }
 196
 197                         listfile = validParameter.validFile(parameters, "list", true);
 198                         if (listfile == "not open") { abort = true; }
 199                         else if (listfile == "not found") {  listfile = "";  }
 200
 201                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 202                         if (taxfile == "not open") { abort = true; }
 203                         else if (taxfile == "not found") {  taxfile = "";  }
 204
 205                         qualfile = validParameter.validFile(parameters, "qfile", true);
 206                         if (qualfile == "not open") { abort = true; }
 207                         else if (qualfile == "not found") {  qualfile = "";  }
 208
 209
 210                         string usedDups = "true";
 211                         string temp = validParameter.validFile(parameters, "dups", false);
 212                         if (temp == "not found") {
 213                                 if (namefile != "") {  temp = "true";                                   }
 214                                 else                            {  temp = "false"; usedDups = "";       }
 215                         }
 216                         dups = m->isTrue(temp);
 217
 218                         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; }
 219
 220                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 221                 }
 222
 223         }
 224         catch(exception& e) {
 225                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 226                 exit(1);
 227         }
 228 }
 229 //**********************************************************************************************************************
 230
 231 void RemoveSeqsCommand::help(){
 232         try {
 233                 m->mothurOut("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");
 234                 m->mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
 235                 m->mothurOut("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");
 236                 m->mothurOut("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");
 237                 m->mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 238                 m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 239                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 240         }
 241         catch(exception& e) {
 242                 m->errorOut(e, "RemoveSeqsCommand", "help");
 243                 exit(1);
 244         }
 245 }
 246
 247 //**********************************************************************************************************************
 248
 249 int RemoveSeqsCommand::execute(){
 250         try {
 251
 252                 if (abort == true) { return 0; }
 253
 254                 //get names you want to keep
 255                 readAccnos();
 256
 257                 if (m->control_pressed) { return 0; }
 258
 259                 //read through the correct file and output lines you want to keep
 260                 if (namefile != "")                     {               readName();             }
 261                 if (fastafile != "")            {               readFasta();    }
 262                 if (groupfile != "")            {               readGroup();    }
 263                 if (alignfile != "")            {               readAlign();    }
 264                 if (listfile != "")                     {               readList();             }
 265                 if (taxfile != "")                      {               readTax();              }
 266                 if (qualfile != "")                     {               readQual();             }
 267
 268                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 269
 270                 m->mothurOut("Removed " + toString(names.size()) + " sequences."); m->mothurOutEndLine();
 271
 272                 if (outputNames.size() != 0) {
 273                         m->mothurOutEndLine();
 274                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 275                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 276                         m->mothurOutEndLine();
 277                 }
 278
 279                 return 0;
 280         }
 281
 282         catch(exception& e) {
 283                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 284                 exit(1);
 285         }
 286 }
 287
 288 //**********************************************************************************************************************
 289 int RemoveSeqsCommand::readFasta(){
 290         try {
 291                 string thisOutputDir = outputDir;
 292                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 293                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 294
 295                 ofstream out;
 296                 m->openOutputFile(outputFileName, out);
 297
 298                 ifstream in;
 299                 m->openInputFile(fastafile, in);
 300                 string name;
 301
 302                 bool wroteSomething = false;
 303
 304                 while(!in.eof()){
 305                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 306
 307                         Sequence currSeq(in);
 308                         name = currSeq.getName();
 309
 310                         if (name != "") {
 311                                 //if this name is in the accnos file
 312                                 if (names.count(name) == 0) {
 313                                         wroteSomething = true;
 314
 315                                         currSeq.printSequence(out);
 316                                 }
 317                         }
 318                         m->gobble(in);
 319                 }
 320                 in.close();
 321                 out.close();
 322
 323                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 324                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 325
 326                 return 0;
 327
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 331                 exit(1);
 332         }
 333 }
 334 //**********************************************************************************************************************
 335 int RemoveSeqsCommand::readQual(){
 336         try {
 337                 string thisOutputDir = outputDir;
 338                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 339                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 340                 ofstream out;
 341                 m->openOutputFile(outputFileName, out);
 342
 343
 344                 ifstream in;
 345                 m->openInputFile(qualfile, in);
 346                 string name;
 347
 348                 bool wroteSomething = false;
 349
 350
 351                 while(!in.eof()){
 352                         string saveName = "";
 353                         string name = "";
 354                         string scores = "";
 355
 356                         in >> name;
 357
 358                         if (name.length() != 0) {
 359                                 saveName = name.substr(1);
 360                                 while (!in.eof())       {
 361                                         char c = in.get();
 362                                         if (c == 10 || c == 13){        break;  }
 363                                         else { name += c; }
 364                                 }
 365                                 m->gobble(in);
 366                         }
 367
 368                         while(in){
 369                                 char letter= in.get();
 370                                 if(letter == '>'){      in.putback(letter);     break;  }
 371                                 else{ scores += letter; }
 372                         }
 373
 374                         m->gobble(in);
 375
 376                         if (names.count(saveName) == 0) {
 377                                 wroteSomething = true;
 378
 379                                 out << name << endl << scores;
 380                         }
 381
 382                         m->gobble(in);
 383                 }
 384                 in.close();
 385                 out.close();
 386
 387
 388                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 389                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 390
 391                 return 0;
 392
 393         }
 394         catch(exception& e) {
 395                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 396                 exit(1);
 397         }
 398 }
 399 //**********************************************************************************************************************
 400 int RemoveSeqsCommand::readList(){
 401         try {
 402                 string thisOutputDir = outputDir;
 403                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 404                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 405
 406                 ofstream out;
 407                 m->openOutputFile(outputFileName, out);
 408
 409                 ifstream in;
 410                 m->openInputFile(listfile, in);
 411
 412                 bool wroteSomething = false;
 413
 414                 while(!in.eof()){
 415                         //read in list vector
 416                         ListVector list(in);
 417
 418                         //make a new list vector
 419                         ListVector newList;
 420                         newList.setLabel(list.getLabel());
 421
 422                         //for each bin
 423                         for (int i = 0; i < list.getNumBins(); i++) {
 424                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 425
 426                                 //parse out names that are in accnos file
 427                                 string binnames = list.get(i);
 428
 429                                 string newNames = "";
 430                                 while (binnames.find_first_of(',') != -1) {
 431                                         string name = binnames.substr(0,binnames.find_first_of(','));
 432                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 433
 434                                         //if that name is in the .accnos file, add it
 435                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 436                                 }
 437
 438                                 //get last name
 439                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 440
 441                                 //if there are names in this bin add to new list
 442                                 if (newNames != "") {
 443                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 444                                         newList.push_back(newNames);
 445                                 }
 446                         }
 447
 448                         //print new listvector
 449                         if (newList.getNumBins() != 0) {
 450                                 wroteSomething = true;
 451                                 newList.print(out);
 452                         }
 453
 454                         m->gobble(in);
 455                 }
 456                 in.close();
 457                 out.close();
 458
 459                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 460                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 461
 462                 return 0;
 463
 464         }
 465         catch(exception& e) {
 466                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 467                 exit(1);
 468         }
 469 }
 470 //**********************************************************************************************************************
 471 int RemoveSeqsCommand::readName(){
 472         try {
 473                 string thisOutputDir = outputDir;
 474                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 475                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 476
 477                 ofstream out;
 478                 m->openOutputFile(outputFileName, out);
 479
 480                 ifstream in;
 481                 m->openInputFile(namefile, in);
 482                 string name, firstCol, secondCol;
 483
 484                 bool wroteSomething = false;
 485
 486                 while(!in.eof()){
 487                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 488
 489                         in >> firstCol;         m->gobble(in);
 490                         in >> secondCol;
 491
 492                         vector<string> parsedNames;
 493                         //parse second column saving each name
 494                         while (secondCol.find_first_of(',') != -1) {
 495                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 496                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 497                                 parsedNames.push_back(name);
 498                         }
 499
 500                         //get name after last ,
 501                         parsedNames.push_back(secondCol);
 502
 503                         vector<string> validSecond;  validSecond.clear();
 504                         for (int i = 0; i < parsedNames.size(); i++) {
 505                                 if (names.count(parsedNames[i]) == 0) {
 506                                         validSecond.push_back(parsedNames[i]);
 507                                 }
 508                         }
 509
 510                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 511                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 512                         }else {
 513                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 514                                 if (names.count(firstCol) == 0) {
 515
 516                                         wroteSomething = true;
 517
 518                                         out << firstCol << '\t';
 519
 520                                         //you know you have at least one valid second since first column is valid
 521                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 522                                         out << validSecond[validSecond.size()-1] << endl;
 523
 524                                         //make first name in set you come to first column and then add the remaining names to second column
 525                                 }else {
 526
 527                                         //you want part of this row
 528                                         if (validSecond.size() != 0) {
 529
 530                                                 wroteSomething = true;
 531
 532                                                 out << validSecond[0] << '\t';
 533
 534                                                 //you know you have at least one valid second since first column is valid
 535                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 536                                                 out << validSecond[validSecond.size()-1] << endl;
 537                                         }
 538                                 }
 539                         }
 540                         m->gobble(in);
 541                 }
 542                 in.close();
 543                 out.close();
 544
 545                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 546                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 547
 548                 return 0;
 549         }
 550         catch(exception& e) {
 551                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 552                 exit(1);
 553         }
 554 }
 555
 556 //**********************************************************************************************************************
 557 int RemoveSeqsCommand::readGroup(){
 558         try {
 559                 string thisOutputDir = outputDir;
 560                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 561                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 562
 563                 ofstream out;
 564                 m->openOutputFile(outputFileName, out);
 565
 566                 ifstream in;
 567                 m->openInputFile(groupfile, in);
 568                 string name, group;
 569
 570                 bool wroteSomething = false;
 571
 572                 while(!in.eof()){
 573                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 574
 575                         in >> name;                             //read from first column
 576                         in >> group;                    //read from second column
 577
 578                         //if this name is in the accnos file
 579                         if (names.count(name) == 0) {
 580                                 wroteSomething = true;
 581                                 out << name << '\t' << group << endl;
 582                         }
 583
 584                         m->gobble(in);
 585                 }
 586                 in.close();
 587                 out.close();
 588
 589                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 590                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 591
 592                 return 0;
 593         }
 594         catch(exception& e) {
 595                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 596                 exit(1);
 597         }
 598 }
 599 //**********************************************************************************************************************
 600 int RemoveSeqsCommand::readTax(){
 601         try {
 602                 string thisOutputDir = outputDir;
 603                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 604                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 605                 ofstream out;
 606                 m->openOutputFile(outputFileName, out);
 607
 608                 ifstream in;
 609                 m->openInputFile(taxfile, in);
 610                 string name, tax;
 611
 612                 bool wroteSomething = false;
 613
 614                 while(!in.eof()){
 615                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 616
 617                         in >> name;                             //read from first column
 618                         in >> tax;                      //read from second column
 619
 620                         //if this name is in the accnos file
 621                         if (names.count(name) == 0) {
 622                                 wroteSomething = true;
 623                                 out << name << '\t' << tax << endl;
 624                         }
 625
 626                         m->gobble(in);
 627                 }
 628                 in.close();
 629                 out.close();
 630
 631                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 632                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 633
 634                 return 0;
 635         }
 636         catch(exception& e) {
 637                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 638                 exit(1);
 639         }
 640 }
 641 //**********************************************************************************************************************
 642 //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
 643 int RemoveSeqsCommand::readAlign(){
 644         try {
 645                 string thisOutputDir = outputDir;
 646                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 647                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 648
 649                 ofstream out;
 650                 m->openOutputFile(outputFileName, out);
 651
 652                 ifstream in;
 653                 m->openInputFile(alignfile, in);
 654                 string name, junk;
 655
 656                 bool wroteSomething = false;
 657
 658                 //read column headers
 659                 for (int i = 0; i < 16; i++) {
 660                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 661                         else                    {       break;                  }
 662                 }
 663                 out << endl;
 664
 665                 while(!in.eof()){
 666                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 667
 668                         in >> name;                             //read from first column
 669
 670                         //if this name is in the accnos file
 671                         if (names.count(name) == 0) {
 672                                 wroteSomething = true;
 673
 674                                 out << name << '\t';
 675
 676                                 //read rest
 677                                 for (int i = 0; i < 15; i++) {
 678                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 679                                         else                    {       break;                  }
 680                                 }
 681                                 out << endl;
 682
 683                         }else {//still read just don't do anything with it
 684
 685                                 //read rest
 686                                 for (int i = 0; i < 15; i++) {
 687                                         if (!in.eof())  {       in >> junk;             }
 688                                         else                    {       break;                  }
 689                                 }
 690                         }
 691
 692                         m->gobble(in);
 693                 }
 694                 in.close();
 695                 out.close();
 696
 697                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 698                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 699
 700                 return 0;
 701
 702         }
 703         catch(exception& e) {
 704                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 705                 exit(1);
 706         }
 707 }
 708 //**********************************************************************************************************************
 709 void RemoveSeqsCommand::readAccnos(){
 710         try {
 711
 712                 ifstream in;
 713                 m->openInputFile(accnosfile, in);
 714                 string name;
 715
 716                 while(!in.eof()){
 717                         in >> name;
 718
 719                         names.insert(name);
 720
 721                         m->gobble(in);
 722                 }
 723                 in.close();
 724
 725         }
 726         catch(exception& e) {
 727                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 728                 exit(1);
 729         }
 730 }
 731
 732 //**********************************************************************************************************************
 733
 734