removelineagecommand.cpp

   1 /*
   2  *  removelineagecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/24/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removelineagecommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15 vector<string> RemoveLineageCommand::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,true); parameters.push_back(ptaxonomy);
  22                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
  23                 CommandParameter ptaxon("taxon", "String", "", "", "", "", "",false,true); parameters.push_back(ptaxon);
  24                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pdups);
  25                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  26                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  27
  28                 vector<string> myArray;
  29                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  30                 return myArray;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "RemoveLineageCommand", "setParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 string RemoveLineageCommand::getHelpString(){
  39         try {
  40                 string helpString = "";
  41                 helpString += "The remove.lineage command reads a taxonomy file and any of the following file types: fasta, name, group, list or alignreport file.\n";
  42                 helpString += "It outputs a file containing only the sequences from the taxonomy file that are not from the taxon you requested to be removed.\n";
  43                 helpString += "The remove.lineage command parameters are taxon, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide taxonomy unless you have a valid current taxonomy file.\n";
  44                 helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=false. \n";
  45                 helpString += "The taxon parameter allows you to select the taxons you would like to remove, and is required.\n";
  46                 helpString += "You may enter your taxons with confidence scores, doing so will remove only those sequences that belong to the taxonomy and whose cofidence scores fall below the scores you give.\n";
  47                 helpString += "If they belong to the taxonomy and have confidences above those you provide the sequence will not be removed.\n";
  48                 helpString += "The remove.lineage command should be in the following format: remove.lineage(taxonomy=yourTaxonomyFile, taxon=yourTaxons).\n";
  49                 helpString += "Example remove.lineage(taxonomy=amazon.silva.taxonomy, taxon=Bacteria;Firmicutes;Bacilli;Lactobacillales;).\n";
  50                 helpString += "Note: If you are running mothur in script mode you must wrap the taxon in ' characters so mothur will ignore the ; in the taxon.\n";
  51                 helpString += "Example remove.lineage(taxonomy=amazon.silva.taxonomy, taxon='Bacteria;Firmicutes;Bacilli;Lactobacillales;').\n";
  52                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  53                 return helpString;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "RemoveLineageCommand", "getHelpString");
  57                 exit(1);
  58         }
  59 }
  60
  61
  62 //**********************************************************************************************************************
  63 RemoveLineageCommand::RemoveLineageCommand(){
  64         try {
  65                 abort = true; calledHelp = true;
  66                 setParameters();
  67                 vector<string> tempOutNames;
  68                 outputTypes["fasta"] = tempOutNames;
  69                 outputTypes["taxonomy"] = tempOutNames;
  70                 outputTypes["name"] = tempOutNames;
  71                 outputTypes["group"] = tempOutNames;
  72                 outputTypes["alignreport"] = tempOutNames;
  73                 outputTypes["list"] = tempOutNames;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "RemoveLineageCommand", "RemoveLineageCommand");
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81 RemoveLineageCommand::RemoveLineageCommand(string option)  {
  82         try {
  83                 abort = false; calledHelp = false;
  84
  85                 //allow user to run help
  86                 if(option == "help") { help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string,string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string,string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["fasta"] = tempOutNames;
 106                         outputTypes["taxonomy"] = tempOutNames;
 107                         outputTypes["name"] = tempOutNames;
 108                         outputTypes["group"] = tempOutNames;
 109                         outputTypes["alignreport"] = tempOutNames;
 110                         outputTypes["list"] = tempOutNames;
 111
 112                         //if the user changes the output directory command factory will send this info to us in the output parameter
 113                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 114
 115                         //if the user changes the input directory command factory will send this info to us in the output parameter
 116                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 117                         if (inputDir == "not found"){   inputDir = "";          }
 118                         else {
 119                                 string path;
 120                                 it = parameters.find("alignreport");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 126                                 }
 127
 128                                 it = parameters.find("fasta");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 134                                 }
 135
 136                                 it = parameters.find("list");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 142                                 }
 143
 144                                 it = parameters.find("name");
 145                                 //user has given a template file
 146                                 if(it != parameters.end()){
 147                                         path = m->hasPath(it->second);
 148                                         //if the user has not given a path then, add inputdir. else leave path alone.
 149                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 150                                 }
 151
 152                                 it = parameters.find("group");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 158                                 }
 159
 160                                 it = parameters.find("taxonomy");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 166                                 }
 167                         }
 168
 169
 170                         //check for required parameters
 171                         fastafile = validParameter.validFile(parameters, "fasta", true);
 172                         if (fastafile == "not open") { abort = true; }
 173                         else if (fastafile == "not found") {  fastafile = "";  }
 174
 175                         namefile = validParameter.validFile(parameters, "name", true);
 176                         if (namefile == "not open") { abort = true; }
 177                         else if (namefile == "not found") {  namefile = "";  }
 178
 179                         groupfile = validParameter.validFile(parameters, "group", true);
 180                         if (groupfile == "not open") { abort = true; }
 181                         else if (groupfile == "not found") {  groupfile = "";  }
 182
 183                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 184                         if (alignfile == "not open") { abort = true; }
 185                         else if (alignfile == "not found") {  alignfile = "";  }
 186
 187                         listfile = validParameter.validFile(parameters, "list", true);
 188                         if (listfile == "not open") { abort = true; }
 189                         else if (listfile == "not found") {  listfile = "";  }
 190
 191                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 192                         if (taxfile == "not open") { abort = true; }
 193                         else if (taxfile == "not found") {
 194                                 taxfile = m->getTaxonomyFile();
 195                                 if (taxfile != "") { m->mothurOut("Using " + taxfile + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
 196                                 else {  m->mothurOut("You have no current taxonomy file and the taxonomy parameter is required."); m->mothurOutEndLine(); abort = true; }
 197                         }
 198
 199                         string usedDups = "true";
 200                         string temp = validParameter.validFile(parameters, "dups", false);
 201                         if (temp == "not found") {
 202                                 if (namefile != "") {  temp = "true";                                   }
 203                                 else                            {  temp = "false"; usedDups = "";       }
 204                         }
 205                         dups = m->isTrue(temp);
 206
 207                         taxons = validParameter.validFile(parameters, "taxon", false);
 208                         if (taxons == "not found") { taxons = "";  m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine();  abort = true;  }
 209                         else {
 210                                 //rip off quotes
 211                                 if (taxons[0] == '\'') {  taxons = taxons.substr(1); }
 212                                 if (taxons[(taxons.length()-1)] == '\'') {  taxons = taxons.substr(0, (taxons.length()-1)); }
 213                         }
 214                         m->splitAtChar(taxons, listOfTaxons, '-');
 215
 216                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport, taxonomy or listfile."); m->mothurOutEndLine(); abort = true; }
 217
 218                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 219
 220                 }
 221
 222         }
 223         catch(exception& e) {
 224                 m->errorOut(e, "RemoveLineageCommand", "RemoveLineageCommand");
 225                 exit(1);
 226         }
 227 }
 228 //**********************************************************************************************************************
 229
 230 int RemoveLineageCommand::execute(){
 231         try {
 232
 233                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 234
 235                 if (m->control_pressed) { return 0; }
 236
 237                 //read through the correct file and output lines you want to keep
 238                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to remove
 239                 if (namefile != "")                     {               readName();             }
 240                 if (fastafile != "")            {               readFasta();    }
 241                 if (groupfile != "")            {               readGroup();    }
 242                 if (alignfile != "")            {               readAlign();    }
 243                 if (listfile != "")                     {               readList();             }
 244
 245
 246                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 247
 248                 if (outputNames.size() != 0) {
 249                         m->mothurOutEndLine();
 250                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 251                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 252                         m->mothurOutEndLine();
 253
 254                         //set fasta file as new current fastafile
 255                         string current = "";
 256                         itTypes = outputTypes.find("fasta");
 257                         if (itTypes != outputTypes.end()) {
 258                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 259                         }
 260
 261                         itTypes = outputTypes.find("name");
 262                         if (itTypes != outputTypes.end()) {
 263                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 264                         }
 265
 266                         itTypes = outputTypes.find("group");
 267                         if (itTypes != outputTypes.end()) {
 268                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 269                         }
 270
 271                         itTypes = outputTypes.find("list");
 272                         if (itTypes != outputTypes.end()) {
 273                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 274                         }
 275
 276                         itTypes = outputTypes.find("taxonomy");
 277                         if (itTypes != outputTypes.end()) {
 278                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 279                         }
 280                 }
 281
 282                 return 0;
 283         }
 284
 285         catch(exception& e) {
 286                 m->errorOut(e, "RemoveLineageCommand", "execute");
 287                 exit(1);
 288         }
 289 }
 290
 291 //**********************************************************************************************************************
 292 int RemoveLineageCommand::readFasta(){
 293         try {
 294                 string thisOutputDir = outputDir;
 295                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 296                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 297
 298                 ofstream out;
 299                 m->openOutputFile(outputFileName, out);
 300
 301                 ifstream in;
 302                 m->openInputFile(fastafile, in);
 303                 string name;
 304
 305                 bool wroteSomething = false;
 306
 307                 while(!in.eof()){
 308                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 309
 310                         Sequence currSeq(in);
 311                         name = currSeq.getName();
 312
 313                         if (name != "") {
 314                                 //if this name is in the accnos file
 315                                 if (names.count(name) == 0) {
 316                                         wroteSomething = true;
 317
 318                                         currSeq.printSequence(out);
 319                                 }
 320                         }
 321                         m->gobble(in);
 322                 }
 323                 in.close();
 324                 out.close();
 325
 326                 if (wroteSomething == false) {  m->mothurOut("Your fasta file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 327                 outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 328
 329                 return 0;
 330
 331         }
 332         catch(exception& e) {
 333                 m->errorOut(e, "RemoveLineageCommand", "readFasta");
 334                 exit(1);
 335         }
 336 }
 337 //**********************************************************************************************************************
 338 int RemoveLineageCommand::readList(){
 339         try {
 340                 string thisOutputDir = outputDir;
 341                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 342                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 343
 344                 ofstream out;
 345                 m->openOutputFile(outputFileName, out);
 346
 347                 ifstream in;
 348                 m->openInputFile(listfile, in);
 349
 350                 bool wroteSomething = false;
 351
 352                 while(!in.eof()){
 353                         //read in list vector
 354                         ListVector list(in);
 355
 356                         //make a new list vector
 357                         ListVector newList;
 358                         newList.setLabel(list.getLabel());
 359
 360                         //for each bin
 361                         for (int i = 0; i < list.getNumBins(); i++) {
 362                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 363
 364                                 //parse out names that are in accnos file
 365                                 string binnames = list.get(i);
 366
 367                                 string newNames = "";
 368                                 while (binnames.find_first_of(',') != -1) {
 369                                         string name = binnames.substr(0,binnames.find_first_of(','));
 370                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 371
 372                                         //if that name is in the .accnos file, add it
 373                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 374                                 }
 375
 376                                 //get last name
 377                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 378
 379                                 //if there are names in this bin add to new list
 380                                 if (newNames != "") {
 381                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 382                                         newList.push_back(newNames);
 383                                 }
 384                         }
 385
 386                         //print new listvector
 387                         if (newList.getNumBins() != 0) {
 388                                 wroteSomething = true;
 389                                 newList.print(out);
 390                         }
 391
 392                         m->gobble(in);
 393                 }
 394                 in.close();
 395                 out.close();
 396
 397                 if (wroteSomething == false) {  m->mothurOut("Your list file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 398                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 399
 400                 return 0;
 401
 402         }
 403         catch(exception& e) {
 404                 m->errorOut(e, "RemoveLineageCommand", "readList");
 405                 exit(1);
 406         }
 407 }
 408 //**********************************************************************************************************************
 409 int RemoveLineageCommand::readName(){
 410         try {
 411                 string thisOutputDir = outputDir;
 412                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 413                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 414
 415                 ofstream out;
 416                 m->openOutputFile(outputFileName, out);
 417
 418                 ifstream in;
 419                 m->openInputFile(namefile, in);
 420                 string name, firstCol, secondCol;
 421
 422                 bool wroteSomething = false;
 423
 424                 while(!in.eof()){
 425                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 426
 427                         in >> firstCol;
 428                         in >> secondCol;
 429
 430                         vector<string> parsedNames;
 431                         m->splitAtComma(secondCol, parsedNames);
 432
 433                         vector<string> validSecond;  validSecond.clear();
 434                         for (int i = 0; i < parsedNames.size(); i++) {
 435                                 if (names.count(parsedNames[i]) == 0) {
 436                                         validSecond.push_back(parsedNames[i]);
 437                                 }
 438                         }
 439
 440                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 441                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 442                         }else {
 443                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 444                                 if (names.count(firstCol) == 0) {
 445
 446                                         wroteSomething = true;
 447
 448                                         out << firstCol << '\t';
 449
 450                                         //you know you have at least one valid second since first column is valid
 451                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 452                                         out << validSecond[validSecond.size()-1] << endl;
 453
 454                                         //make first name in set you come to first column and then add the remaining names to second column
 455                                 }else {
 456
 457                                         //you want part of this row
 458                                         if (validSecond.size() != 0) {
 459
 460                                                 wroteSomething = true;
 461
 462                                                 out << validSecond[0] << '\t';
 463
 464                                                 //you know you have at least one valid second since first column is valid
 465                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 466                                                 out << validSecond[validSecond.size()-1] << endl;
 467                                         }
 468                                 }
 469                         }
 470                         m->gobble(in);
 471                 }
 472                 in.close();
 473                 out.close();
 474
 475                 if (wroteSomething == false) {  m->mothurOut("Your name file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 476                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 477
 478                 return 0;
 479         }
 480         catch(exception& e) {
 481                 m->errorOut(e, "RemoveLineageCommand", "readName");
 482                 exit(1);
 483         }
 484 }
 485
 486 //**********************************************************************************************************************
 487 int RemoveLineageCommand::readGroup(){
 488         try {
 489                 string thisOutputDir = outputDir;
 490                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 491                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 492
 493                 ofstream out;
 494                 m->openOutputFile(outputFileName, out);
 495
 496                 ifstream in;
 497                 m->openInputFile(groupfile, in);
 498                 string name, group;
 499
 500                 bool wroteSomething = false;
 501
 502                 while(!in.eof()){
 503                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 504
 505                         in >> name;                             //read from first column
 506                         in >> group;                    //read from second column
 507
 508                         //if this name is in the accnos file
 509                         if (names.count(name) == 0) {
 510                                 wroteSomething = true;
 511                                 out << name << '\t' << group << endl;
 512                         }
 513
 514                         m->gobble(in);
 515                 }
 516                 in.close();
 517                 out.close();
 518
 519                 if (wroteSomething == false) {  m->mothurOut("Your group file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 520                 outputNames.push_back(outputFileName); outputTypes["group"].push_back(outputFileName);
 521
 522                 return 0;
 523         }
 524         catch(exception& e) {
 525                 m->errorOut(e, "RemoveLineageCommand", "readGroup");
 526                 exit(1);
 527         }
 528 }
 529 //**********************************************************************************************************************
 530 int RemoveLineageCommand::readTax(){
 531         try {
 532                 string thisOutputDir = outputDir;
 533                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 534                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 535                 ofstream out;
 536                 m->openOutputFile(outputFileName, out);
 537
 538                 ifstream in;
 539                 m->openInputFile(taxfile, in);
 540                 string name, tax;
 541
 542                 bool wroteSomething = false;
 543
 544                 vector<bool> taxonsHasConfidence; taxonsHasConfidence.resize(listOfTaxons.size(), false);
 545                 vector< vector< map<string, float> > > searchTaxons; searchTaxons.resize(listOfTaxons.size());
 546                 vector<string> noConfidenceTaxons; noConfidenceTaxons.resize(listOfTaxons.size(), "");
 547
 548                 for (int i = 0; i < listOfTaxons.size(); i++) {
 549                         noConfidenceTaxons[i] = listOfTaxons[i];
 550                         int hasConPos = listOfTaxons[i].find_first_of('(');
 551                         if (hasConPos != string::npos) {
 552                                 taxonsHasConfidence[i] = true;
 553                                 searchTaxons[i] = getTaxons(listOfTaxons[i]);
 554                                 noConfidenceTaxons[i] = removeConfidences(listOfTaxons[i]);
 555                         }
 556                 }
 557
 558
 559                 while(!in.eof()){
 560
 561                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 562
 563                         in >> name;                             //read from first column
 564                         in >> tax;                      //read from second column
 565
 566                         bool remove = false;
 567
 568                         for (int j = 0; j < listOfTaxons.size(); j++) {
 569                                 string newtax = tax;
 570
 571                                 //if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
 572                                 if (!taxonsHasConfidence[j]) {
 573
 574                                         int hasConfidences = tax.find_first_of('(');
 575                                         if (hasConfidences != string::npos) {
 576                                                 newtax = removeConfidences(tax);
 577                                         }
 578
 579                                         int pos = newtax.find(noConfidenceTaxons[j]);
 580
 581                                         if (pos == string::npos) {
 582                                                 //wroteSomething = true;
 583                                                 //out << name << '\t' << tax << endl;
 584                                         }else{ //this sequence contains the taxon the user wants to remove
 585                                                 names.insert(name);
 586                                                 remove=true; break;
 587                                         }
 588
 589                                 }else{//if taxons has them and you don't them remove taxons
 590                                         int hasConfidences = tax.find_first_of('(');
 591                                         if (hasConfidences == string::npos) {
 592
 593                                                 int pos = newtax.find(noConfidenceTaxons[j]);
 594
 595                                                 if (pos == string::npos) {
 596                                                         //wroteSomething = true;
 597                                                         //out << name << '\t' << tax << endl;
 598                                                 }else{ //this sequence contains the taxon the user wants to remove
 599                                                         names.insert(name);
 600                                                         remove=true; break;
 601                                                 }
 602                                         }else { //both have confidences so we want to make sure the users confidences are greater then or equal to the taxons
 603                                                 //first remove confidences from both and see if the taxonomy exists
 604
 605                                                 string noNewTax = tax;
 606                                                 int hasConfidences = tax.find_first_of('(');
 607                                                 if (hasConfidences != string::npos) {
 608                                                         noNewTax = removeConfidences(tax);
 609                                                 }
 610
 611                                                 int pos = noNewTax.find(noConfidenceTaxons[j]);
 612
 613                                                 if (pos != string::npos) { //if yes, then are the confidences okay
 614
 615                                                         vector< map<string, float> > usersTaxon = getTaxons(newtax);
 616
 617                                                         //the usersTaxon is most likely longer than the searchTaxons, and searchTaxon[0] may relate to userTaxon[4]
 618                                                         //we want to "line them up", so we will find the the index where the searchstring starts
 619                                                         int index = 0;
 620                                                         for (int i = 0; i < usersTaxon.size(); i++) {
 621
 622                                                                 if (usersTaxon[i].begin()->first == searchTaxons[j][0].begin()->first) {
 623                                                                         index = i;
 624                                                                         int spot = 0;
 625                                                                         bool goodspot = true;
 626                                                                         //is this really the start, or are we dealing with a taxon of the same name?
 627                                                                         while ((spot < searchTaxons[j].size()) && ((i+spot) < usersTaxon.size())) {
 628                                                                                 if (usersTaxon[i+spot].begin()->first != searchTaxons[j][spot].begin()->first) { goodspot = false; break; }
 629                                                                                 else { spot++; }
 630                                                                         }
 631
 632                                                                         if (goodspot) { break; }
 633                                                                 }
 634                                                         }
 635
 636                                                         for (int i = 0; i < searchTaxons[j].size(); i++) {
 637
 638                                                                 if ((i+index) < usersTaxon.size()) { //just in case, should never be false
 639                                                                         if (usersTaxon[i+index].begin()->second < searchTaxons[j][i].begin()->second) { //is the users cutoff less than the search taxons
 640                                                                                 remove = true;
 641                                                                                 break;
 642                                                                         }
 643                                                                 }else {
 644                                                                         remove = true;
 645                                                                         break;
 646                                                                 }
 647                                                         }
 648
 649                                                         //passed the test so remove you
 650                                                         if (remove) {
 651                                                                 names.insert(name);
 652                                                                 remove=true; break;
 653                                                         }else {
 654                                                                 //wroteSomething = true;
 655                                                                 //out << name << '\t' << tax << endl;
 656                                                         }
 657                                                 }else {
 658                                                         //wroteSomething = true;
 659                                                         //out << name << '\t' << tax << endl;
 660                                                 }
 661                                         }
 662                                 }
 663
 664                         }
 665
 666                         if (!remove) {  wroteSomething = true; out << name << '\t' << tax << endl; }
 667                         m->gobble(in);
 668                 }
 669                 in.close();
 670                 out.close();
 671
 672                 if (!wroteSomething) { m->mothurOut("Your taxonomy file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 673                 outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
 674
 675                 return 0;
 676
 677         }
 678         catch(exception& e) {
 679                 m->errorOut(e, "RemoveLineageCommand", "readTax");
 680                 exit(1);
 681         }
 682 }
 683 /**************************************************************************************************/
 684 vector< map<string, float> > RemoveLineageCommand::getTaxons(string tax) {
 685         try {
 686
 687                 vector< map<string, float> > t;
 688                 string taxon = "";
 689                 int taxLength = tax.length();
 690                 for(int i=0;i<taxLength;i++){
 691                         if(tax[i] == ';'){
 692
 693                                 int openParen = taxon.find_first_of('(');
 694                                 int closeParen = taxon.find_last_of(')');
 695
 696                                 string newtaxon, confidence;
 697                                 if ((openParen != string::npos) && (closeParen != string::npos)) {
 698                                         newtaxon = taxon.substr(0, openParen); //rip off confidence
 699                                         confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
 700                                 }else{
 701                                         newtaxon = taxon;
 702                                         confidence = "0";
 703                                 }
 704                                 float con = 0;
 705                                 convert(confidence, con);
 706
 707                                 map<string, float> temp;
 708                                 temp[newtaxon] = con;
 709                                 t.push_back(temp);
 710
 711                                 taxon = "";
 712                         }
 713                         else{
 714                                 taxon += tax[i];
 715                         }
 716                 }
 717
 718                 return t;
 719         }
 720         catch(exception& e) {
 721                 m->errorOut(e, "RemoveLineageCommand", "getTaxons");
 722                 exit(1);
 723         }
 724 }
 725 /**************************************************************************************************/
 726 string RemoveLineageCommand::removeConfidences(string tax) {
 727         try {
 728
 729                 string taxon = "";
 730                 int taxLength = tax.length();
 731                 for(int i=0;i<taxLength;i++){
 732                         if(tax[i] == ';'){
 733                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 734                                 taxon += ";";
 735                         }
 736                         else{
 737                                 taxon += tax[i];
 738                         }
 739                 }
 740
 741                 return taxon;
 742         }
 743         catch(exception& e) {
 744                 m->errorOut(e, "RemoveLineageCommand", "removeConfidences");
 745                 exit(1);
 746         }
 747 }
 748 //**********************************************************************************************************************
 749 //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
 750 int RemoveLineageCommand::readAlign(){
 751         try {
 752                 string thisOutputDir = outputDir;
 753                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 754                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 755
 756                 ofstream out;
 757                 m->openOutputFile(outputFileName, out);
 758
 759                 ifstream in;
 760                 m->openInputFile(alignfile, in);
 761                 string name, junk;
 762
 763                 bool wroteSomething = false;
 764
 765                 //read column headers
 766                 for (int i = 0; i < 16; i++) {
 767                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 768                         else                    {       break;                  }
 769                 }
 770                 out << endl;
 771
 772                 while(!in.eof()){
 773                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 774
 775                         in >> name;                             //read from first column
 776
 777                         //if this name is in the accnos file
 778                         if (names.count(name) == 0) {
 779                                 wroteSomething = true;
 780
 781                                 out << name << '\t';
 782
 783                                 //read rest
 784                                 for (int i = 0; i < 15; i++) {
 785                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 786                                         else                    {       break;                  }
 787                                 }
 788                                 out << endl;
 789
 790                         }else {//still read just don't do anything with it
 791
 792                                 //read rest
 793                                 for (int i = 0; i < 15; i++) {
 794                                         if (!in.eof())  {       in >> junk;             }
 795                                         else                    {       break;                  }
 796                                 }
 797                         }
 798
 799                         m->gobble(in);
 800                 }
 801                 in.close();
 802                 out.close();
 803
 804                 if (wroteSomething == false) {  m->mothurOut("Your align file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 805                 outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
 806
 807                 return 0;
 808
 809         }
 810         catch(exception& e) {
 811                 m->errorOut(e, "RemoveLineageCommand", "readAlign");
 812                 exit(1);
 813         }
 814 }
 815 //**********************************************************************************************************************
 816