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", "", "", "NameCount", "FNGLT", "none",false,false); parameters.push_back(pname);
  19         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "FNGLT", "none",false,false); parameters.push_back(pcount);
  20                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "FNGLT", "none",false,false); parameters.push_back(pgroup);
  21                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  22                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
  23                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
  24                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pqfile);
  25                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(paccnos);
  26                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pdups);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "RemoveSeqsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string RemoveSeqsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The remove.seqs command reads an .accnos file and at least one of the following file types: fasta, name, group, count, list, taxonomy, quality or alignreport file.\n";
  44                 helpString += "It outputs a file containing the sequences NOT in the .accnos file.\n";
  45                 helpString += "The remove.seqs command parameters are accnos, fasta, name, group, count, list, taxonomy, qfile, alignreport and dups.  You must provide accnos and at least one of the file parameters.\n";
  46                 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";
  47                 helpString += "The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
  48                 helpString += "Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
  49                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  50                 return helpString;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "RemoveSeqsCommand", "getHelpString");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 string RemoveSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  59         try {
  60         string outputFileName = "";
  61                 map<string, vector<string> >::iterator it;
  62
  63         //is this a type this command creates
  64         it = outputTypes.find(type);
  65         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  66         else {
  67             if (type == "fasta")            {   outputFileName =  "pick" + m->getExtension(inputName);   }
  68             else if (type == "taxonomy")    {   outputFileName =  "pick" + m->getExtension(inputName);   }
  69             else if (type == "name")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  70             else if (type == "group")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  71             else if (type == "list")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  72             else if (type == "qfile")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  73             else if (type == "alignreport") {   outputFileName =  "pick.align.report";                   }
  74             else if (type == "count")       {   outputFileName =  "pick.count.table";   }
  75             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  76         }
  77         return outputFileName;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "RemoveSeqsCommand", "getOutputFileNameTag");
  81                 exit(1);
  82         }
  83 }
  84
  85 //**********************************************************************************************************************
  86 RemoveSeqsCommand::RemoveSeqsCommand(){
  87         try {
  88                 abort = true; calledHelp = true;
  89                 setParameters();
  90                 vector<string> tempOutNames;
  91                 outputTypes["fasta"] = tempOutNames;
  92                 outputTypes["taxonomy"] = tempOutNames;
  93                 outputTypes["name"] = tempOutNames;
  94                 outputTypes["group"] = tempOutNames;
  95                 outputTypes["alignreport"] = tempOutNames;
  96                 outputTypes["list"] = tempOutNames;
  97                 outputTypes["qfile"] = tempOutNames;
  98         outputTypes["count"] = tempOutNames;
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106 RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
 107         try {
 108                 abort = false; calledHelp = false;
 109
 110                 //allow user to run help
 111                 if(option == "help") { help(); abort = true; calledHelp = true; }
 112                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 113
 114                 else {
 115                         vector<string> myArray = setParameters();
 116
 117                         OptionParser parser(option);
 118                         map<string,string> parameters = parser.getParameters();
 119
 120                         ValidParameters validParameter;
 121                         map<string,string>::iterator it;
 122
 123                         //check to make sure all parameters are valid for command
 124                         for (it = parameters.begin(); it != parameters.end(); it++) {
 125                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 126                         }
 127
 128                         //initialize outputTypes
 129                         vector<string> tempOutNames;
 130                         outputTypes["fasta"] = tempOutNames;
 131                         outputTypes["taxonomy"] = tempOutNames;
 132                         outputTypes["name"] = tempOutNames;
 133                         outputTypes["group"] = tempOutNames;
 134                         outputTypes["alignreport"] = tempOutNames;
 135                         outputTypes["list"] = tempOutNames;
 136                         outputTypes["qfile"] = tempOutNames;
 137             outputTypes["count"] = tempOutNames;
 138
 139                         //if the user changes the output directory command factory will send this info to us in the output parameter
 140                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 141
 142                         //if the user changes the input directory command factory will send this info to us in the output parameter
 143                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 144                         if (inputDir == "not found"){   inputDir = "";          }
 145                         else {
 146                                 string path;
 147                                 it = parameters.find("alignreport");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 153                                 }
 154
 155                                 it = parameters.find("fasta");
 156                                 //user has given a template file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 161                                 }
 162
 163                                 it = parameters.find("accnos");
 164                                 //user has given a template file
 165                                 if(it != parameters.end()){
 166                                         path = m->hasPath(it->second);
 167                                         //if the user has not given a path then, add inputdir. else leave path alone.
 168                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 169                                 }
 170
 171                                 it = parameters.find("list");
 172                                 //user has given a template file
 173                                 if(it != parameters.end()){
 174                                         path = m->hasPath(it->second);
 175                                         //if the user has not given a path then, add inputdir. else leave path alone.
 176                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 177                                 }
 178
 179                                 it = parameters.find("name");
 180                                 //user has given a template file
 181                                 if(it != parameters.end()){
 182                                         path = m->hasPath(it->second);
 183                                         //if the user has not given a path then, add inputdir. else leave path alone.
 184                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 185                                 }
 186
 187                                 it = parameters.find("group");
 188                                 //user has given a template file
 189                                 if(it != parameters.end()){
 190                                         path = m->hasPath(it->second);
 191                                         //if the user has not given a path then, add inputdir. else leave path alone.
 192                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 193                                 }
 194
 195                                 it = parameters.find("taxonomy");
 196                                 //user has given a template file
 197                                 if(it != parameters.end()){
 198                                         path = m->hasPath(it->second);
 199                                         //if the user has not given a path then, add inputdir. else leave path alone.
 200                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 201                                 }
 202
 203                                 it = parameters.find("qfile");
 204                                 //user has given a template file
 205                                 if(it != parameters.end()){
 206                                         path = m->hasPath(it->second);
 207                                         //if the user has not given a path then, add inputdir. else leave path alone.
 208                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 209                                 }
 210
 211                 it = parameters.find("count");
 212                                 //user has given a template file
 213                                 if(it != parameters.end()){
 214                                         path = m->hasPath(it->second);
 215                                         //if the user has not given a path then, add inputdir. else leave path alone.
 216                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 217                                 }
 218                         }
 219
 220
 221                         //check for required parameters
 222                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 223                         if (accnosfile == "not open") { abort = true; }
 224                         else if (accnosfile == "not found") {
 225                                 accnosfile = m->getAccnosFile();
 226                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
 227                                 else {
 228                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
 229                                         abort = true;
 230                                 }
 231                         }else { m->setAccnosFile(accnosfile); }
 232
 233                         fastafile = validParameter.validFile(parameters, "fasta", true);
 234                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 235                         else if (fastafile == "not found") {  fastafile = "";  }
 236                         else { m->setFastaFile(fastafile); }
 237
 238                         namefile = validParameter.validFile(parameters, "name", true);
 239                         if (namefile == "not open") { namefile = ""; abort = true; }
 240                         else if (namefile == "not found") {  namefile = "";  }
 241                         else { m->setNameFile(namefile); }
 242
 243                         groupfile = validParameter.validFile(parameters, "group", true);
 244                         if (groupfile == "not open") { abort = true; }
 245                         else if (groupfile == "not found") {  groupfile = "";  }
 246                         else { m->setGroupFile(groupfile); }
 247
 248                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 249                         if (alignfile == "not open") { abort = true; }
 250                         else if (alignfile == "not found") {  alignfile = "";  }
 251
 252                         listfile = validParameter.validFile(parameters, "list", true);
 253                         if (listfile == "not open") { abort = true; }
 254                         else if (listfile == "not found") {  listfile = "";  }
 255                         else { m->setListFile(listfile); }
 256
 257                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 258                         if (taxfile == "not open") { abort = true; }
 259                         else if (taxfile == "not found") {  taxfile = "";  }
 260                         else { m->setTaxonomyFile(taxfile); }
 261
 262                         qualfile = validParameter.validFile(parameters, "qfile", true);
 263                         if (qualfile == "not open") { abort = true; }
 264                         else if (qualfile == "not found") {  qualfile = "";  }
 265                         else { m->setQualFile(qualfile); }
 266
 267                         string usedDups = "true";
 268                         string temp = validParameter.validFile(parameters, "dups", false);
 269                         if (temp == "not found") {
 270                                 if (namefile != "") {  temp = "true";                                   }
 271                                 else                            {  temp = "false"; usedDups = "";       }
 272                         }
 273                         dups = m->isTrue(temp);
 274
 275             countfile = validParameter.validFile(parameters, "count", true);
 276             if (countfile == "not open") { countfile = ""; abort = true; }
 277             else if (countfile == "not found") { countfile = "";  }
 278             else { m->setCountTableFile(countfile); }
 279
 280             if ((namefile != "") && (countfile != "")) {
 281                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 282             }
 283
 284             if ((groupfile != "") && (countfile != "")) {
 285                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 286             }
 287
 288                         if ((countfile == "") && (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; }
 289
 290             if (countfile == "") {
 291                 if ((fastafile != "") && (namefile == "")) {
 292                     vector<string> files; files.push_back(fastafile);
 293                     parser.getNameFile(files);
 294                 }
 295             }
 296                 }
 297
 298         }
 299         catch(exception& e) {
 300                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 301                 exit(1);
 302         }
 303 }
 304 //**********************************************************************************************************************
 305
 306 int RemoveSeqsCommand::execute(){
 307         try {
 308
 309                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 310
 311                 //get names you want to keep
 312                 names = m->readAccnos(accnosfile);
 313
 314                 if (m->control_pressed) { return 0; }
 315
 316         if (countfile != "") {
 317             if ((fastafile != "") || (listfile != "") || (taxfile != "")) {
 318                 m->mothurOut("\n[NOTE]: The count file should contain only unique names, so mothur assumes your fasta, list and taxonomy files also contain only uniques.\n\n");
 319             }
 320         }
 321
 322                 //read through the correct file and output lines you want to keep
 323                 if (namefile != "")                     {               readName();             }
 324                 if (fastafile != "")            {               readFasta();    }
 325                 if (groupfile != "")            {               readGroup();    }
 326                 if (alignfile != "")            {               readAlign();    }
 327                 if (listfile != "")                     {               readList();             }
 328                 if (taxfile != "")                      {               readTax();              }
 329                 if (qualfile != "")                     {               readQual();             }
 330         if (countfile != "")            {               readCount();            }
 331
 332                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 333
 334                 if (outputNames.size() != 0) {
 335                         m->mothurOutEndLine();
 336                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 337                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 338                         m->mothurOutEndLine();
 339
 340                         //set fasta file as new current fastafile
 341                         string current = "";
 342                         itTypes = outputTypes.find("fasta");
 343                         if (itTypes != outputTypes.end()) {
 344                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 345                         }
 346
 347                         itTypes = outputTypes.find("name");
 348                         if (itTypes != outputTypes.end()) {
 349                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 350                         }
 351
 352                         itTypes = outputTypes.find("group");
 353                         if (itTypes != outputTypes.end()) {
 354                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 355                         }
 356
 357                         itTypes = outputTypes.find("list");
 358                         if (itTypes != outputTypes.end()) {
 359                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 360                         }
 361
 362                         itTypes = outputTypes.find("taxonomy");
 363                         if (itTypes != outputTypes.end()) {
 364                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 365                         }
 366
 367                         itTypes = outputTypes.find("qfile");
 368                         if (itTypes != outputTypes.end()) {
 369                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 370                         }
 371
 372             itTypes = outputTypes.find("count");
 373                         if (itTypes != outputTypes.end()) {
 374                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 375                         }
 376                 }
 377
 378                 return 0;
 379         }
 380
 381         catch(exception& e) {
 382                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 383                 exit(1);
 384         }
 385 }
 386
 387 //**********************************************************************************************************************
 388 int RemoveSeqsCommand::readFasta(){
 389         try {
 390                 string thisOutputDir = outputDir;
 391                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 392                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta", fastafile);
 393
 394                 ofstream out;
 395                 m->openOutputFile(outputFileName, out);
 396
 397                 ifstream in;
 398                 m->openInputFile(fastafile, in);
 399                 string name;
 400
 401                 bool wroteSomething = false;
 402                 int removedCount = 0;
 403
 404                 while(!in.eof()){
 405                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 406
 407                         Sequence currSeq(in);
 408                         name = currSeq.getName();
 409
 410                         if (name != "") {
 411                                 //if this name is in the accnos file
 412                                 if (names.count(name) == 0) {
 413                                         wroteSomething = true;
 414
 415                                         currSeq.printSequence(out);
 416                                 }else {  removedCount++;  }
 417                         }
 418                         m->gobble(in);
 419                 }
 420                 in.close();
 421                 out.close();
 422
 423                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 424                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 425
 426                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 427
 428                 return 0;
 429
 430         }
 431         catch(exception& e) {
 432                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 433                 exit(1);
 434         }
 435 }
 436 //**********************************************************************************************************************
 437 int RemoveSeqsCommand::readQual(){
 438         try {
 439                 string thisOutputDir = outputDir;
 440                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 441                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + getOutputFileNameTag("qfile", qualfile);
 442                 ofstream out;
 443                 m->openOutputFile(outputFileName, out);
 444
 445
 446                 ifstream in;
 447                 m->openInputFile(qualfile, in);
 448                 string name;
 449
 450                 bool wroteSomething = false;
 451                 int removedCount = 0;
 452
 453
 454                 while(!in.eof()){
 455                         string saveName = "";
 456                         string name = "";
 457                         string scores = "";
 458
 459                         in >> name;
 460
 461                         if (name.length() != 0) {
 462                                 saveName = name.substr(1);
 463                                 while (!in.eof())       {
 464                                         char c = in.get();
 465                                         if (c == 10 || c == 13){        break;  }
 466                                         else { name += c; }
 467                                 }
 468                                 m->gobble(in);
 469                         }
 470
 471                         while(in){
 472                                 char letter= in.get();
 473                                 if(letter == '>'){      in.putback(letter);     break;  }
 474                                 else{ scores += letter; }
 475                         }
 476
 477                         m->gobble(in);
 478
 479                         if (names.count(saveName) == 0) {
 480                                 wroteSomething = true;
 481
 482                                 out << name << endl << scores;
 483                         }else {  removedCount++;  }
 484
 485                         m->gobble(in);
 486                 }
 487                 in.close();
 488                 out.close();
 489
 490
 491                 if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 492                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 493
 494                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 495
 496                 return 0;
 497
 498         }
 499         catch(exception& e) {
 500                 m->errorOut(e, "RemoveSeqsCommand", "readQual");
 501                 exit(1);
 502         }
 503 }
 504 //**********************************************************************************************************************
 505 int RemoveSeqsCommand::readCount(){
 506         try {
 507
 508                 string thisOutputDir = outputDir;
 509                 if (outputDir == "") {  thisOutputDir += m->hasPath(countfile);  }
 510                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(countfile)) + getOutputFileNameTag("count", countfile);
 511
 512                 ofstream out;
 513                 m->openOutputFile(outputFileName, out);
 514
 515                 ifstream in;
 516                 m->openInputFile(countfile, in);
 517
 518                 bool wroteSomething = false;
 519                 int removedCount = 0;
 520
 521         string headers = m->getline(in); m->gobble(in);
 522         out << headers << endl;
 523
 524         string name, rest; int thisTotal;
 525         while (!in.eof()) {
 526
 527             if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 528
 529             in >> name; m->gobble(in);
 530             in >> thisTotal; m->gobble(in);
 531             rest = m->getline(in); m->gobble(in);
 532             if (m->debug) { m->mothurOut("[DEBUG]: " + name + '\t' + rest + "\n"); }
 533
 534             if (names.count(name) == 0) {
 535                 out << name << '\t' << thisTotal << '\t' << rest << endl;
 536                 wroteSomething = true;
 537             }else { removedCount += thisTotal; }
 538         }
 539         in.close();
 540                 out.close();
 541
 542                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 543                 outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
 544
 545                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your count file."); m->mothurOutEndLine();
 546
 547                 return 0;
 548         }
 549         catch(exception& e) {
 550                 m->errorOut(e, "RemoveSeqsCommand", "readCount");
 551                 exit(1);
 552         }
 553 }
 554 //**********************************************************************************************************************
 555 int RemoveSeqsCommand::readList(){
 556         try {
 557                 string thisOutputDir = outputDir;
 558                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 559                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + getOutputFileNameTag("list", listfile);
 560                 ofstream out;
 561                 m->openOutputFile(outputFileName, out);
 562
 563                 ifstream in;
 564                 m->openInputFile(listfile, in);
 565
 566                 bool wroteSomething = false;
 567                 int removedCount = 0;
 568
 569                 while(!in.eof()){
 570
 571                         removedCount = 0;
 572
 573                         //read in list vector
 574                         ListVector list(in);
 575
 576                         //make a new list vector
 577                         ListVector newList;
 578                         newList.setLabel(list.getLabel());
 579
 580                         //for each bin
 581                         for (int i = 0; i < list.getNumBins(); i++) {
 582                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 583
 584                                 //parse out names that are in accnos file
 585                                 string binnames = list.get(i);
 586
 587                                 string newNames = "";
 588                                 while (binnames.find_first_of(',') != -1) {
 589                                         string name = binnames.substr(0,binnames.find_first_of(','));
 590                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 591
 592                                         //if that name is in the .accnos file, add it
 593                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 594                                         else {  removedCount++;  }
 595                                 }
 596
 597                                 //get last name
 598                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 599                                 else {  removedCount++;  }
 600
 601                                 //if there are names in this bin add to new list
 602                                 if (newNames != "") {
 603                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 604                                         newList.push_back(newNames);
 605                                 }
 606                         }
 607
 608                         //print new listvector
 609                         if (newList.getNumBins() != 0) {
 610                                 wroteSomething = true;
 611                                 newList.print(out);
 612                         }
 613
 614                         m->gobble(in);
 615                 }
 616                 in.close();
 617                 out.close();
 618
 619                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 620                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 621
 622                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
 623
 624                 return 0;
 625
 626         }
 627         catch(exception& e) {
 628                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 629                 exit(1);
 630         }
 631 }
 632 //**********************************************************************************************************************
 633 int RemoveSeqsCommand::readName(){
 634         try {
 635                 string thisOutputDir = outputDir;
 636                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 637                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + getOutputFileNameTag("name", namefile);
 638                 ofstream out;
 639                 m->openOutputFile(outputFileName, out);
 640
 641                 ifstream in;
 642                 m->openInputFile(namefile, in);
 643                 string name, firstCol, secondCol;
 644
 645                 bool wroteSomething = false;
 646                 int removedCount = 0;
 647
 648                 while(!in.eof()){
 649                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 650
 651                         in >> firstCol;         m->gobble(in);
 652                         in >> secondCol;
 653
 654                         vector<string> parsedNames;
 655                         m->splitAtComma(secondCol, parsedNames);
 656
 657                         vector<string> validSecond;  validSecond.clear();
 658                         for (int i = 0; i < parsedNames.size(); i++) {
 659                                 if (names.count(parsedNames[i]) == 0) {
 660                                         validSecond.push_back(parsedNames[i]);
 661                                 }
 662                         }
 663
 664                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 665                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 666                                 removedCount += parsedNames.size();
 667                         }else {
 668                                 removedCount += parsedNames.size()-validSecond.size();
 669                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 670                                 if (names.count(firstCol) == 0) {
 671
 672                                         wroteSomething = true;
 673
 674                                         out << firstCol << '\t';
 675
 676                                         //you know you have at least one valid second since first column is valid
 677                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 678                                         out << validSecond[validSecond.size()-1] << endl;
 679
 680                                         //make first name in set you come to first column and then add the remaining names to second column
 681                                 }else {
 682
 683                                         //you want part of this row
 684                                         if (validSecond.size() != 0) {
 685
 686                                                 wroteSomething = true;
 687
 688                                                 out << validSecond[0] << '\t';
 689
 690                                                 //you know you have at least one valid second since first column is valid
 691                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 692                                                 out << validSecond[validSecond.size()-1] << endl;
 693                                         }
 694                                 }
 695                         }
 696                         m->gobble(in);
 697                 }
 698                 in.close();
 699                 out.close();
 700
 701                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 702                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 703
 704                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
 705
 706                 return 0;
 707         }
 708         catch(exception& e) {
 709                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 710                 exit(1);
 711         }
 712 }
 713
 714 //**********************************************************************************************************************
 715 int RemoveSeqsCommand::readGroup(){
 716         try {
 717                 string thisOutputDir = outputDir;
 718                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 719                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
 720                 ofstream out;
 721                 m->openOutputFile(outputFileName, out);
 722
 723                 ifstream in;
 724                 m->openInputFile(groupfile, in);
 725                 string name, group;
 726
 727                 bool wroteSomething = false;
 728                 int removedCount = 0;
 729
 730                 while(!in.eof()){
 731                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 732
 733                         in >> name;                             //read from first column
 734                         in >> group;                    //read from second column
 735
 736                         //if this name is in the accnos file
 737                         if (names.count(name) == 0) {
 738                                 wroteSomething = true;
 739                                 out << name << '\t' << group << endl;
 740                         }else {  removedCount++;  }
 741
 742                         m->gobble(in);
 743                 }
 744                 in.close();
 745                 out.close();
 746
 747                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 748                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 749
 750                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
 751
 752
 753                 return 0;
 754         }
 755         catch(exception& e) {
 756                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 757                 exit(1);
 758         }
 759 }
 760 //**********************************************************************************************************************
 761 int RemoveSeqsCommand::readTax(){
 762         try {
 763                 string thisOutputDir = outputDir;
 764                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 765                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + getOutputFileNameTag("taxonomy", taxfile);
 766                 ofstream out;
 767                 m->openOutputFile(outputFileName, out);
 768
 769                 ifstream in;
 770                 m->openInputFile(taxfile, in);
 771                 string name, tax;
 772
 773                 bool wroteSomething = false;
 774                 int removedCount = 0;
 775
 776                 while(!in.eof()){
 777                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 778
 779                         in >> name;                             //read from first column
 780                         in >> tax;                      //read from second column
 781
 782                         //if this name is in the accnos file
 783                         if (names.count(name) == 0) {
 784                                 wroteSomething = true;
 785                                 out << name << '\t' << tax << endl;
 786                         }else {  removedCount++;  }
 787
 788                         m->gobble(in);
 789                 }
 790                 in.close();
 791                 out.close();
 792
 793                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 794                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 795
 796                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 797
 798                 return 0;
 799         }
 800         catch(exception& e) {
 801                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 802                 exit(1);
 803         }
 804 }
 805 //**********************************************************************************************************************
 806 //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
 807 int RemoveSeqsCommand::readAlign(){
 808         try {
 809                 string thisOutputDir = outputDir;
 810                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 811                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + getOutputFileNameTag("alignreport");
 812
 813                 ofstream out;
 814                 m->openOutputFile(outputFileName, out);
 815
 816                 ifstream in;
 817                 m->openInputFile(alignfile, in);
 818                 string name, junk;
 819
 820                 bool wroteSomething = false;
 821                 int removedCount = 0;
 822
 823                 //read column headers
 824                 for (int i = 0; i < 16; i++) {
 825                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 826                         else                    {       break;                  }
 827                 }
 828                 out << endl;
 829
 830                 while(!in.eof()){
 831                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 832
 833                         in >> name;                             //read from first column
 834
 835                         //if this name is in the accnos file
 836                         if (names.count(name) == 0) {
 837                                 wroteSomething = true;
 838
 839                                 out << name << '\t';
 840
 841                                 //read rest
 842                                 for (int i = 0; i < 15; i++) {
 843                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 844                                         else                    {       break;                  }
 845                                 }
 846                                 out << endl;
 847
 848                         }else {//still read just don't do anything with it
 849                                 removedCount++;
 850
 851                                 //read rest
 852                                 for (int i = 0; i < 15; i++) {
 853                                         if (!in.eof())  {       in >> junk;             }
 854                                         else                    {       break;                  }
 855                                 }
 856                         }
 857
 858                         m->gobble(in);
 859                 }
 860                 in.close();
 861                 out.close();
 862
 863                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 864                 outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
 865
 866                 m->mothurOut("Removed " + toString(removedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 867
 868
 869                 return 0;
 870
 871         }
 872         catch(exception& e) {
 873                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 874                 exit(1);
 875         }
 876 }
 877 //**********************************************************************************************************************
 878
 879