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") { fastafile = ""; 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") { namefile = ""; 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                         if ((fastafile != "") && (namefile == "")) {
 241                                 vector<string> files; files.push_back(fastafile);
 242                                 parser.getNameFile(files);
 243                         }
 244                 }
 245
 246         }
 247         catch(exception& e) {
 248                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 249                 exit(1);
 250         }
 251 }
 252 //**********************************************************************************************************************
 253
 254 int RemoveSeqsCommand::execute(){
 255         try {
 256
 257                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 258
 259                 //get names you want to keep
 260                 readAccnos();
 261
 262                 if (m->control_pressed) { return 0; }
 263
 264                 //read through the correct file and output lines you want to keep
 265                 if (namefile != "")                     {               readName();             }
 266                 if (fastafile != "")            {               readFasta();    }
 267                 if (groupfile != "")            {               readGroup();    }
 268                 if (alignfile != "")            {               readAlign();    }
 269                 if (listfile != "")                     {               readList();             }
 270                 if (taxfile != "")                      {               readTax();              }
 271                 if (qualfile != "")                     {               readQual();             }
 272
 273                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 274
 275                 if (outputNames.size() != 0) {
 276                         m->mothurOutEndLine();
 277                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 278                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 279                         m->mothurOutEndLine();
 280
 281                         //set fasta file as new current fastafile
 282                         string current = "";
 283                         itTypes = outputTypes.find("fasta");
 284                         if (itTypes != outputTypes.end()) {
 285                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 286                         }
 287
 288                         itTypes = outputTypes.find("name");
 289                         if (itTypes != outputTypes.end()) {
 290                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 291                         }
 292
 293                         itTypes = outputTypes.find("group");
 294                         if (itTypes != outputTypes.end()) {
 295                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 296                         }
 297
 298                         itTypes = outputTypes.find("list");
 299                         if (itTypes != outputTypes.end()) {
 300                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 301                         }
 302
 303                         itTypes = outputTypes.find("taxonomy");
 304                         if (itTypes != outputTypes.end()) {
 305                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 306                         }
 307
 308                         itTypes = outputTypes.find("qfile");
 309                         if (itTypes != outputTypes.end()) {
 310                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 311                         }
 312                 }
 313
 314                 return 0;
 315         }
 316
 317         catch(exception& e) {
 318                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 319                 exit(1);
 320         }
 321 }
 322
 323 //**********************************************************************************************************************
 324 int RemoveSeqsCommand::readFasta(){
 325         try {
 326                 string thisOutputDir = outputDir;
 327                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 328                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 329
 330                 ofstream out;
 331                 m->openOutputFile(outputFileName, out);
 332
 333                 ifstream in;
 334                 m->openInputFile(fastafile, in);
 335                 string name;
 336
 337                 bool wroteSomething = false;
 338                 int removedCount = 0;
 339
 340                 while(!in.eof()){
 341                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 342
 343                         Sequence currSeq(in);
 344                         name = currSeq.getName();
 345
 346                         if (name != "") {
 347                                 //if this name is in the accnos file
 348                                 if (names.count(name) == 0) {
 349                                         wroteSomething = true;
 350
 351                                         currSeq.printSequence(out);
 352                                 }else {  removedCount++;  }
 353                         }
 354                         m->gobble(in);
 355                 }
 356                 in.close();
 357                 out.close();
 358
 359                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 360                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 361
 362                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 363
 364                 return 0;
 365
 366         }
 367         catch(exception& e) {
 368                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 369                 exit(1);
 370         }
 371 }
 372 //**********************************************************************************************************************
 373 int RemoveSeqsCommand::readQual(){
 374         try {
 375                 string thisOutputDir = outputDir;
 376                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 377                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 378                 ofstream out;
 379                 m->openOutputFile(outputFileName, out);
 380
 381
 382                 ifstream in;
 383                 m->openInputFile(qualfile, in);
 384                 string name;
 385
 386                 bool wroteSomething = false;
 387                 int removedCount = 0;
 388
 389
 390                 while(!in.eof()){
 391                         string saveName = "";
 392                         string name = "";
 393                         string scores = "";
 394
 395                         in >> name;
 396
 397                         if (name.length() != 0) {
 398                                 saveName = name.substr(1);
 399                                 while (!in.eof())       {
 400                                         char c = in.get();
 401                                         if (c == 10 || c == 13){        break;  }
 402                                         else { name += c; }
 403                                 }
 404                                 m->gobble(in);
 405                         }
 406
 407                         while(in){
 408                                 char letter= in.get();
 409                                 if(letter == '>'){      in.putback(letter);     break;  }
 410                                 else{ scores += letter; }
 411                         }
 412
 413                         m->gobble(in);
 414
 415                         if (names.count(saveName) == 0) {
 416                                 wroteSomething = true;
 417
 418                                 out << name << endl << scores;
 419                         }else {  removedCount++;  }
 420
 421                         m->gobble(in);
 422                 }
 423                 in.close();
 424                 out.close();
 425
 426
 427                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 428                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 429
 430                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 431
 432                 return 0;
 433
 434         }
 435         catch(exception& e) {
 436                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 437                 exit(1);
 438         }
 439 }
 440 //**********************************************************************************************************************
 441 int RemoveSeqsCommand::readList(){
 442         try {
 443                 string thisOutputDir = outputDir;
 444                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 445                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 446
 447                 ofstream out;
 448                 m->openOutputFile(outputFileName, out);
 449
 450                 ifstream in;
 451                 m->openInputFile(listfile, in);
 452
 453                 bool wroteSomething = false;
 454                 int removedCount = 0;
 455
 456                 while(!in.eof()){
 457
 458                         removedCount = 0;
 459
 460                         //read in list vector
 461                         ListVector list(in);
 462
 463                         //make a new list vector
 464                         ListVector newList;
 465                         newList.setLabel(list.getLabel());
 466
 467                         //for each bin
 468                         for (int i = 0; i < list.getNumBins(); i++) {
 469                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 470
 471                                 //parse out names that are in accnos file
 472                                 string binnames = list.get(i);
 473
 474                                 string newNames = "";
 475                                 while (binnames.find_first_of(',') != -1) {
 476                                         string name = binnames.substr(0,binnames.find_first_of(','));
 477                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 478
 479                                         //if that name is in the .accnos file, add it
 480                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 481                                         else {  removedCount++;  }
 482                                 }
 483
 484                                 //get last name
 485                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 486                                 else {  removedCount++;  }
 487
 488                                 //if there are names in this bin add to new list
 489                                 if (newNames != "") {
 490                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 491                                         newList.push_back(newNames);
 492                                 }
 493                         }
 494
 495                         //print new listvector
 496                         if (newList.getNumBins() != 0) {
 497                                 wroteSomething = true;
 498                                 newList.print(out);
 499                         }
 500
 501                         m->gobble(in);
 502                 }
 503                 in.close();
 504                 out.close();
 505
 506                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 507                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 508
 509                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 510
 511                 return 0;
 512
 513         }
 514         catch(exception& e) {
 515                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 516                 exit(1);
 517         }
 518 }
 519 //**********************************************************************************************************************
 520 int RemoveSeqsCommand::readName(){
 521         try {
 522                 string thisOutputDir = outputDir;
 523                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 524                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 525
 526                 ofstream out;
 527                 m->openOutputFile(outputFileName, out);
 528
 529                 ifstream in;
 530                 m->openInputFile(namefile, in);
 531                 string name, firstCol, secondCol;
 532
 533                 bool wroteSomething = false;
 534                 int removedCount = 0;
 535
 536                 while(!in.eof()){
 537                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 538
 539                         in >> firstCol;         m->gobble(in);
 540                         in >> secondCol;
 541
 542                         vector<string> parsedNames;
 543                         m->splitAtComma(secondCol, parsedNames);
 544
 545                         vector<string> validSecond;  validSecond.clear();
 546                         for (int i = 0; i < parsedNames.size(); i++) {
 547                                 if (names.count(parsedNames[i]) == 0) {
 548                                         validSecond.push_back(parsedNames[i]);
 549                                 }
 550                         }
 551
 552                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 553                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 554                                 removedCount += parsedNames.size();
 555                         }else {
 556                                 removedCount += parsedNames.size()-validSecond.size();
 557                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 558                                 if (names.count(firstCol) == 0) {
 559
 560                                         wroteSomething = true;
 561
 562                                         out << firstCol << '\t';
 563
 564                                         //you know you have at least one valid second since first column is valid
 565                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 566                                         out << validSecond[validSecond.size()-1] << endl;
 567
 568                                         //make first name in set you come to first column and then add the remaining names to second column
 569                                 }else {
 570
 571                                         //you want part of this row
 572                                         if (validSecond.size() != 0) {
 573
 574                                                 wroteSomething = true;
 575
 576                                                 out << validSecond[0] << '\t';
 577
 578                                                 //you know you have at least one valid second since first column is valid
 579                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 580                                                 out << validSecond[validSecond.size()-1] << endl;
 581                                         }
 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["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 591
 592                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 593
 594                 return 0;
 595         }
 596         catch(exception& e) {
 597                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 598                 exit(1);
 599         }
 600 }
 601
 602 //**********************************************************************************************************************
 603 int RemoveSeqsCommand::readGroup(){
 604         try {
 605                 string thisOutputDir = outputDir;
 606                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 607                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 608
 609                 ofstream out;
 610                 m->openOutputFile(outputFileName, out);
 611
 612                 ifstream in;
 613                 m->openInputFile(groupfile, in);
 614                 string name, group;
 615
 616                 bool wroteSomething = false;
 617                 int removedCount = 0;
 618
 619                 while(!in.eof()){
 620                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 621
 622                         in >> name;                             //read from first column
 623                         in >> group;                    //read from second column
 624
 625                         //if this name is in the accnos file
 626                         if (names.count(name) == 0) {
 627                                 wroteSomething = true;
 628                                 out << name << '\t' << group << endl;
 629                         }else {  removedCount++;  }
 630
 631                         m->gobble(in);
 632                 }
 633                 in.close();
 634                 out.close();
 635
 636                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 637                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 638
 639                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 640
 641
 642                 return 0;
 643         }
 644         catch(exception& e) {
 645                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 646                 exit(1);
 647         }
 648 }
 649 //**********************************************************************************************************************
 650 int RemoveSeqsCommand::readTax(){
 651         try {
 652                 string thisOutputDir = outputDir;
 653                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 654                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 655                 ofstream out;
 656                 m->openOutputFile(outputFileName, out);
 657
 658                 ifstream in;
 659                 m->openInputFile(taxfile, in);
 660                 string name, tax;
 661
 662                 bool wroteSomething = false;
 663                 int removedCount = 0;
 664
 665                 while(!in.eof()){
 666                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 667
 668                         in >> name;                             //read from first column
 669                         in >> tax;                      //read from second column
 670
 671                         //if this name is in the accnos file
 672                         if (names.count(name) == 0) {
 673                                 wroteSomething = true;
 674                                 out << name << '\t' << tax << endl;
 675                         }else {  removedCount++;  }
 676
 677                         m->gobble(in);
 678                 }
 679                 in.close();
 680                 out.close();
 681
 682                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 683                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 684
 685                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 686
 687                 return 0;
 688         }
 689         catch(exception& e) {
 690                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 691                 exit(1);
 692         }
 693 }
 694 //**********************************************************************************************************************
 695 //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
 696 int RemoveSeqsCommand::readAlign(){
 697         try {
 698                 string thisOutputDir = outputDir;
 699                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 700                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 701
 702                 ofstream out;
 703                 m->openOutputFile(outputFileName, out);
 704
 705                 ifstream in;
 706                 m->openInputFile(alignfile, in);
 707                 string name, junk;
 708
 709                 bool wroteSomething = false;
 710                 int removedCount = 0;
 711
 712                 //read column headers
 713                 for (int i = 0; i < 16; i++) {
 714                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 715                         else                    {       break;                  }
 716                 }
 717                 out << endl;
 718
 719                 while(!in.eof()){
 720                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 721
 722                         in >> name;                             //read from first column
 723
 724                         //if this name is in the accnos file
 725                         if (names.count(name) == 0) {
 726                                 wroteSomething = true;
 727
 728                                 out << name << '\t';
 729
 730                                 //read rest
 731                                 for (int i = 0; i < 15; i++) {
 732                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 733                                         else                    {       break;                  }
 734                                 }
 735                                 out << endl;
 736
 737                         }else {//still read just don't do anything with it
 738                                 removedCount++;
 739
 740                                 //read rest
 741                                 for (int i = 0; i < 15; i++) {
 742                                         if (!in.eof())  {       in >> junk;             }
 743                                         else                    {       break;                  }
 744                                 }
 745                         }
 746
 747                         m->gobble(in);
 748                 }
 749                 in.close();
 750                 out.close();
 751
 752                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 753                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 754
 755                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 756
 757
 758                 return 0;
 759
 760         }
 761         catch(exception& e) {
 762                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 763                 exit(1);
 764         }
 765 }
 766 //**********************************************************************************************************************
 767 void RemoveSeqsCommand::readAccnos(){
 768         try {
 769
 770                 ifstream in;
 771                 m->openInputFile(accnosfile, in);
 772                 string name;
 773
 774                 while(!in.eof()){
 775                         in >> name;
 776
 777                         names.insert(name);
 778
 779                         m->gobble(in);
 780                 }
 781                 in.close();
 782
 783         }
 784         catch(exception& e) {
 785                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 786                 exit(1);
 787         }
 788 }
 789
 790 //**********************************************************************************************************************
 791
 792