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++) {        m->mothurRemove(outputNames[i]); } return 0; }
 270
 271                 if (outputNames.size() != 0) {
 272                         m->mothurOutEndLine();
 273                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 274                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 275                         m->mothurOutEndLine();
 276
 277                         //set fasta file as new current fastafile
 278                         string current = "";
 279                         itTypes = outputTypes.find("fasta");
 280                         if (itTypes != outputTypes.end()) {
 281                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 282                         }
 283
 284                         itTypes = outputTypes.find("name");
 285                         if (itTypes != outputTypes.end()) {
 286                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 287                         }
 288
 289                         itTypes = outputTypes.find("group");
 290                         if (itTypes != outputTypes.end()) {
 291                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 292                         }
 293
 294                         itTypes = outputTypes.find("list");
 295                         if (itTypes != outputTypes.end()) {
 296                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 297                         }
 298
 299                         itTypes = outputTypes.find("taxonomy");
 300                         if (itTypes != outputTypes.end()) {
 301                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 302                         }
 303
 304                         itTypes = outputTypes.find("qfile");
 305                         if (itTypes != outputTypes.end()) {
 306                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 307                         }
 308                 }
 309
 310                 return 0;
 311         }
 312
 313         catch(exception& e) {
 314                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 315                 exit(1);
 316         }
 317 }
 318
 319 //**********************************************************************************************************************
 320 int RemoveSeqsCommand::readFasta(){
 321         try {
 322                 string thisOutputDir = outputDir;
 323                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 324                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 325
 326                 ofstream out;
 327                 m->openOutputFile(outputFileName, out);
 328
 329                 ifstream in;
 330                 m->openInputFile(fastafile, in);
 331                 string name;
 332
 333                 bool wroteSomething = false;
 334                 int removedCount = 0;
 335
 336                 while(!in.eof()){
 337                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 338
 339                         Sequence currSeq(in);
 340                         name = currSeq.getName();
 341
 342                         if (name != "") {
 343                                 //if this name is in the accnos file
 344                                 if (names.count(name) == 0) {
 345                                         wroteSomething = true;
 346
 347                                         currSeq.printSequence(out);
 348                                 }else {  removedCount++;  }
 349                         }
 350                         m->gobble(in);
 351                 }
 352                 in.close();
 353                 out.close();
 354
 355                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 356                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 357
 358                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 359
 360                 return 0;
 361
 362         }
 363         catch(exception& e) {
 364                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 365                 exit(1);
 366         }
 367 }
 368 //**********************************************************************************************************************
 369 int RemoveSeqsCommand::readQual(){
 370         try {
 371                 string thisOutputDir = outputDir;
 372                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 373                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 374                 ofstream out;
 375                 m->openOutputFile(outputFileName, out);
 376
 377
 378                 ifstream in;
 379                 m->openInputFile(qualfile, in);
 380                 string name;
 381
 382                 bool wroteSomething = false;
 383                 int removedCount = 0;
 384
 385
 386                 while(!in.eof()){
 387                         string saveName = "";
 388                         string name = "";
 389                         string scores = "";
 390
 391                         in >> name;
 392
 393                         if (name.length() != 0) {
 394                                 saveName = name.substr(1);
 395                                 while (!in.eof())       {
 396                                         char c = in.get();
 397                                         if (c == 10 || c == 13){        break;  }
 398                                         else { name += c; }
 399                                 }
 400                                 m->gobble(in);
 401                         }
 402
 403                         while(in){
 404                                 char letter= in.get();
 405                                 if(letter == '>'){      in.putback(letter);     break;  }
 406                                 else{ scores += letter; }
 407                         }
 408
 409                         m->gobble(in);
 410
 411                         if (names.count(saveName) == 0) {
 412                                 wroteSomething = true;
 413
 414                                 out << name << endl << scores;
 415                         }else {  removedCount++;  }
 416
 417                         m->gobble(in);
 418                 }
 419                 in.close();
 420                 out.close();
 421
 422
 423                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 424                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 425
 426                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 427
 428                 return 0;
 429
 430         }
 431         catch(exception& e) {
 432                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 433                 exit(1);
 434         }
 435 }
 436 //**********************************************************************************************************************
 437 int RemoveSeqsCommand::readList(){
 438         try {
 439                 string thisOutputDir = outputDir;
 440                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 441                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 442
 443                 ofstream out;
 444                 m->openOutputFile(outputFileName, out);
 445
 446                 ifstream in;
 447                 m->openInputFile(listfile, in);
 448
 449                 bool wroteSomething = false;
 450                 int removedCount = 0;
 451
 452                 while(!in.eof()){
 453
 454                         removedCount = 0;
 455
 456                         //read in list vector
 457                         ListVector list(in);
 458
 459                         //make a new list vector
 460                         ListVector newList;
 461                         newList.setLabel(list.getLabel());
 462
 463                         //for each bin
 464                         for (int i = 0; i < list.getNumBins(); i++) {
 465                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 466
 467                                 //parse out names that are in accnos file
 468                                 string binnames = list.get(i);
 469
 470                                 string newNames = "";
 471                                 while (binnames.find_first_of(',') != -1) {
 472                                         string name = binnames.substr(0,binnames.find_first_of(','));
 473                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 474
 475                                         //if that name is in the .accnos file, add it
 476                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 477                                         else {  removedCount++;  }
 478                                 }
 479
 480                                 //get last name
 481                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 482                                 else {  removedCount++;  }
 483
 484                                 //if there are names in this bin add to new list
 485                                 if (newNames != "") {
 486                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 487                                         newList.push_back(newNames);
 488                                 }
 489                         }
 490
 491                         //print new listvector
 492                         if (newList.getNumBins() != 0) {
 493                                 wroteSomething = true;
 494                                 newList.print(out);
 495                         }
 496
 497                         m->gobble(in);
 498                 }
 499                 in.close();
 500                 out.close();
 501
 502                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 503                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 504
 505                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 506
 507                 return 0;
 508
 509         }
 510         catch(exception& e) {
 511                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 512                 exit(1);
 513         }
 514 }
 515 //**********************************************************************************************************************
 516 int RemoveSeqsCommand::readName(){
 517         try {
 518                 string thisOutputDir = outputDir;
 519                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 520                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 521
 522                 ofstream out;
 523                 m->openOutputFile(outputFileName, out);
 524
 525                 ifstream in;
 526                 m->openInputFile(namefile, in);
 527                 string name, firstCol, secondCol;
 528
 529                 bool wroteSomething = false;
 530                 int removedCount = 0;
 531
 532                 while(!in.eof()){
 533                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 534
 535                         in >> firstCol;         m->gobble(in);
 536                         in >> secondCol;
 537
 538                         vector<string> parsedNames;
 539                         m->splitAtComma(secondCol, parsedNames);
 540
 541                         vector<string> validSecond;  validSecond.clear();
 542                         for (int i = 0; i < parsedNames.size(); i++) {
 543                                 if (names.count(parsedNames[i]) == 0) {
 544                                         validSecond.push_back(parsedNames[i]);
 545                                 }
 546                         }
 547
 548                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 549                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 550                                 removedCount += parsedNames.size();
 551                         }else {
 552                                 removedCount += parsedNames.size()-validSecond.size();
 553                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 554                                 if (names.count(firstCol) == 0) {
 555
 556                                         wroteSomething = true;
 557
 558                                         out << firstCol << '\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                                         //make first name in set you come to first column and then add the remaining names to second column
 565                                 }else {
 566
 567                                         //you want part of this row
 568                                         if (validSecond.size() != 0) {
 569
 570                                                 wroteSomething = true;
 571
 572                                                 out << validSecond[0] << '\t';
 573
 574                                                 //you know you have at least one valid second since first column is valid
 575                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 576                                                 out << validSecond[validSecond.size()-1] << endl;
 577                                         }
 578                                 }
 579                         }
 580                         m->gobble(in);
 581                 }
 582                 in.close();
 583                 out.close();
 584
 585                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 586                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 587
 588                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 589
 590                 return 0;
 591         }
 592         catch(exception& e) {
 593                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 594                 exit(1);
 595         }
 596 }
 597
 598 //**********************************************************************************************************************
 599 int RemoveSeqsCommand::readGroup(){
 600         try {
 601                 string thisOutputDir = outputDir;
 602                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 603                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 604
 605                 ofstream out;
 606                 m->openOutputFile(outputFileName, out);
 607
 608                 ifstream in;
 609                 m->openInputFile(groupfile, in);
 610                 string name, group;
 611
 612                 bool wroteSomething = false;
 613                 int removedCount = 0;
 614
 615                 while(!in.eof()){
 616                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 617
 618                         in >> name;                             //read from first column
 619                         in >> group;                    //read from second column
 620
 621                         //if this name is in the accnos file
 622                         if (names.count(name) == 0) {
 623                                 wroteSomething = true;
 624                                 out << name << '\t' << group << endl;
 625                         }else {  removedCount++;  }
 626
 627                         m->gobble(in);
 628                 }
 629                 in.close();
 630                 out.close();
 631
 632                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 633                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 634
 635                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 636
 637
 638                 return 0;
 639         }
 640         catch(exception& e) {
 641                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 642                 exit(1);
 643         }
 644 }
 645 //**********************************************************************************************************************
 646 int RemoveSeqsCommand::readTax(){
 647         try {
 648                 string thisOutputDir = outputDir;
 649                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 650                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 651                 ofstream out;
 652                 m->openOutputFile(outputFileName, out);
 653
 654                 ifstream in;
 655                 m->openInputFile(taxfile, in);
 656                 string name, tax;
 657
 658                 bool wroteSomething = false;
 659                 int removedCount = 0;
 660
 661                 while(!in.eof()){
 662                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 663
 664                         in >> name;                             //read from first column
 665                         in >> tax;                      //read from second column
 666
 667                         //if this name is in the accnos file
 668                         if (names.count(name) == 0) {
 669                                 wroteSomething = true;
 670                                 out << name << '\t' << tax << endl;
 671                         }else {  removedCount++;  }
 672
 673                         m->gobble(in);
 674                 }
 675                 in.close();
 676                 out.close();
 677
 678                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 679                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 680
 681                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 682
 683                 return 0;
 684         }
 685         catch(exception& e) {
 686                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 687                 exit(1);
 688         }
 689 }
 690 //**********************************************************************************************************************
 691 //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
 692 int RemoveSeqsCommand::readAlign(){
 693         try {
 694                 string thisOutputDir = outputDir;
 695                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 696                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 697
 698                 ofstream out;
 699                 m->openOutputFile(outputFileName, out);
 700
 701                 ifstream in;
 702                 m->openInputFile(alignfile, in);
 703                 string name, junk;
 704
 705                 bool wroteSomething = false;
 706                 int removedCount = 0;
 707
 708                 //read column headers
 709                 for (int i = 0; i < 16; i++) {
 710                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 711                         else                    {       break;                  }
 712                 }
 713                 out << endl;
 714
 715                 while(!in.eof()){
 716                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 717
 718                         in >> name;                             //read from first column
 719
 720                         //if this name is in the accnos file
 721                         if (names.count(name) == 0) {
 722                                 wroteSomething = true;
 723
 724                                 out << name << '\t';
 725
 726                                 //read rest
 727                                 for (int i = 0; i < 15; i++) {
 728                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 729                                         else                    {       break;                  }
 730                                 }
 731                                 out << endl;
 732
 733                         }else {//still read just don't do anything with it
 734                                 removedCount++;
 735
 736                                 //read rest
 737                                 for (int i = 0; i < 15; i++) {
 738                                         if (!in.eof())  {       in >> junk;             }
 739                                         else                    {       break;                  }
 740                                 }
 741                         }
 742
 743                         m->gobble(in);
 744                 }
 745                 in.close();
 746                 out.close();
 747
 748                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 749                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 750
 751                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 752
 753
 754                 return 0;
 755
 756         }
 757         catch(exception& e) {
 758                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 759                 exit(1);
 760         }
 761 }
 762 //**********************************************************************************************************************
 763 void RemoveSeqsCommand::readAccnos(){
 764         try {
 765
 766                 ifstream in;
 767                 m->openInputFile(accnosfile, in);
 768                 string name;
 769
 770                 while(!in.eof()){
 771                         in >> name;
 772
 773                         names.insert(name);
 774
 775                         m->gobble(in);
 776                 }
 777                 in.close();
 778
 779         }
 780         catch(exception& e) {
 781                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 782                 exit(1);
 783         }
 784 }
 785
 786 //**********************************************************************************************************************
 787
 788