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::getValidParameters(){
  16         try {
  17                 string Array[] =  {"fasta","name", "group", "alignreport", "taxon", "dups", "list","taxonomy","outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "RemoveLineageCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 RemoveLineageCommand::RemoveLineageCommand(){
  28         try {
  29                 abort = true; calledHelp = true;
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32                 outputTypes["taxonomy"] = tempOutNames;
  33                 outputTypes["name"] = tempOutNames;
  34                 outputTypes["group"] = tempOutNames;
  35                 outputTypes["alignreport"] = tempOutNames;
  36                 outputTypes["list"] = tempOutNames;
  37         }
  38         catch(exception& e) {
  39                 m->errorOut(e, "RemoveLineageCommand", "RemoveLineageCommand");
  40                 exit(1);
  41         }
  42 }
  43 //**********************************************************************************************************************
  44 vector<string> RemoveLineageCommand::getRequiredParameters(){
  45         try {
  46                 string Array[] =  {"taxonomy"};
  47                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  48                 return myArray;
  49         }
  50         catch(exception& e) {
  51                 m->errorOut(e, "RemoveLineageCommand", "getRequiredParameters");
  52                 exit(1);
  53         }
  54 }
  55 //**********************************************************************************************************************
  56 vector<string> RemoveLineageCommand::getRequiredFiles(){
  57         try {
  58                 vector<string> myArray;
  59                 return myArray;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "RemoveLineageCommand", "getRequiredFiles");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67 RemoveLineageCommand::RemoveLineageCommand(string option)  {
  68         try {
  69                 abort = false; calledHelp = false;
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; calledHelp = true; }
  73
  74                 else {
  75                         //valid paramters for this command
  76                         string Array[] =  {"fasta","name", "group", "alignreport", "taxon", "dups", "list","taxonomy","outputdir","inputdir"};
  77                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  78
  79                         OptionParser parser(option);
  80                         map<string,string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83                         map<string,string>::iterator it;
  84
  85                         //check to make sure all parameters are valid for command
  86                         for (it = parameters.begin(); it != parameters.end(); it++) {
  87                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["fasta"] = tempOutNames;
  93                         outputTypes["taxonomy"] = tempOutNames;
  94                         outputTypes["name"] = tempOutNames;
  95                         outputTypes["group"] = tempOutNames;
  96                         outputTypes["alignreport"] = tempOutNames;
  97                         outputTypes["list"] = tempOutNames;
  98
  99                         //if the user changes the output directory command factory will send this info to us in the output parameter
 100                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 string path;
 107                                 it = parameters.find("alignreport");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 113                                 }
 114
 115                                 it = parameters.find("fasta");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 121                                 }
 122
 123                                 it = parameters.find("list");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 129                                 }
 130
 131                                 it = parameters.find("name");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 137                                 }
 138
 139                                 it = parameters.find("group");
 140                                 //user has given a template file
 141                                 if(it != parameters.end()){
 142                                         path = m->hasPath(it->second);
 143                                         //if the user has not given a path then, add inputdir. else leave path alone.
 144                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 145                                 }
 146
 147                                 it = parameters.find("taxonomy");
 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["taxonomy"] = inputDir + it->second;         }
 153                                 }
 154                         }
 155
 156
 157                         //check for required parameters
 158                         fastafile = validParameter.validFile(parameters, "fasta", true);
 159                         if (fastafile == "not open") { abort = true; }
 160                         else if (fastafile == "not found") {  fastafile = "";  }
 161
 162                         namefile = validParameter.validFile(parameters, "name", true);
 163                         if (namefile == "not open") { abort = true; }
 164                         else if (namefile == "not found") {  namefile = "";  }
 165
 166                         groupfile = validParameter.validFile(parameters, "group", true);
 167                         if (groupfile == "not open") { abort = true; }
 168                         else if (groupfile == "not found") {  groupfile = "";  }
 169
 170                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 171                         if (alignfile == "not open") { abort = true; }
 172                         else if (alignfile == "not found") {  alignfile = "";  }
 173
 174                         listfile = validParameter.validFile(parameters, "list", true);
 175                         if (listfile == "not open") { abort = true; }
 176                         else if (listfile == "not found") {  listfile = "";  }
 177
 178                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 179                         if (taxfile == "not open") { abort = true; }
 180                         else if (taxfile == "not found") {  taxfile = ""; m->mothurOut("The taxonomy parameter is required for the get.lineage command."); m->mothurOutEndLine();  abort = true; }
 181
 182                         string usedDups = "true";
 183                         string temp = validParameter.validFile(parameters, "dups", false);
 184                         if (temp == "not found") {
 185                                 if (namefile != "") {  temp = "true";                                   }
 186                                 else                            {  temp = "false"; usedDups = "";       }
 187                         }
 188                         dups = m->isTrue(temp);
 189
 190                         taxons = validParameter.validFile(parameters, "taxon", false);
 191                         if (taxons == "not found") { taxons = "";  m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine();  abort = true;  }
 192                         else {
 193                                 //rip off quotes
 194                                 if (taxons[0] == '\'') {  taxons = taxons.substr(1); }
 195                                 if (taxons[(taxons.length()-1)] == '\'') {  taxons = taxons.substr(0, (taxons.length()-1)); }
 196                         }
 197
 198
 199                         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; }
 200
 201                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 202
 203                 }
 204
 205         }
 206         catch(exception& e) {
 207                 m->errorOut(e, "RemoveLineageCommand", "RemoveLineageCommand");
 208                 exit(1);
 209         }
 210 }
 211 //**********************************************************************************************************************
 212
 213 void RemoveLineageCommand::help(){
 214         try {
 215                 m->mothurOut("The remove.lineage command reads a taxonomy file and any of the following file types: fasta, name, group, list or alignreport file.\n");
 216                 m->mothurOut("It outputs a file containing only the sequences from the taxonomy file that are not from the taxon you requested to be removed.\n");
 217                 m->mothurOut("The remove.lineage command parameters are taxon, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide taxonomy and taxon.\n");
 218                 m->mothurOut("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");
 219                 m->mothurOut("The taxon parameter allows you to select the taxons you would like to remove.\n");
 220                 m->mothurOut("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");
 221                 m->mothurOut("If they belong to the taxonomy and have confidences above those you provide the sequence will not be removed.\n");
 222                 m->mothurOut("The remove.lineage command should be in the following format: remove.lineage(taxonomy=yourTaxonomyFile, taxon=yourTaxons).\n");
 223                 m->mothurOut("Example remove.lineage(taxonomy=amazon.silva.taxonomy, taxon=Bacteria;Firmicutes;Bacilli;Lactobacillales;).\n");
 224                 m->mothurOut("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");
 225                 m->mothurOut("Example remove.lineage(taxonomy=amazon.silva.taxonomy, taxon='Bacteria;Firmicutes;Bacilli;Lactobacillales;').\n");
 226                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "RemoveLineageCommand", "help");
 230                 exit(1);
 231         }
 232 }
 233
 234 //**********************************************************************************************************************
 235
 236 int RemoveLineageCommand::execute(){
 237         try {
 238
 239                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 240
 241                 if (m->control_pressed) { return 0; }
 242
 243                 //read through the correct file and output lines you want to keep
 244                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to remove
 245                 if (namefile != "")                     {               readName();             }
 246                 if (fastafile != "")            {               readFasta();    }
 247                 if (groupfile != "")            {               readGroup();    }
 248                 if (alignfile != "")            {               readAlign();    }
 249                 if (listfile != "")                     {               readList();             }
 250
 251
 252                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 253
 254                 if (outputNames.size() != 0) {
 255                         m->mothurOutEndLine();
 256                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 257                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 258                         m->mothurOutEndLine();
 259
 260                         //set fasta file as new current fastafile
 261                         string current = "";
 262                         itTypes = outputTypes.find("fasta");
 263                         if (itTypes != outputTypes.end()) {
 264                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 265                         }
 266
 267                         itTypes = outputTypes.find("name");
 268                         if (itTypes != outputTypes.end()) {
 269                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 270                         }
 271
 272                         itTypes = outputTypes.find("group");
 273                         if (itTypes != outputTypes.end()) {
 274                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 275                         }
 276
 277                         itTypes = outputTypes.find("list");
 278                         if (itTypes != outputTypes.end()) {
 279                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 280                         }
 281
 282                         itTypes = outputTypes.find("taxonomy");
 283                         if (itTypes != outputTypes.end()) {
 284                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 285                         }
 286                 }
 287
 288                 return 0;
 289         }
 290
 291         catch(exception& e) {
 292                 m->errorOut(e, "RemoveLineageCommand", "execute");
 293                 exit(1);
 294         }
 295 }
 296
 297 //**********************************************************************************************************************
 298 int RemoveLineageCommand::readFasta(){
 299         try {
 300                 string thisOutputDir = outputDir;
 301                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 302                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 303
 304                 ofstream out;
 305                 m->openOutputFile(outputFileName, out);
 306
 307                 ifstream in;
 308                 m->openInputFile(fastafile, in);
 309                 string name;
 310
 311                 bool wroteSomething = false;
 312
 313                 while(!in.eof()){
 314                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 315
 316                         Sequence currSeq(in);
 317                         name = currSeq.getName();
 318
 319                         if (name != "") {
 320                                 //if this name is in the accnos file
 321                                 if (names.count(name) == 0) {
 322                                         wroteSomething = true;
 323
 324                                         currSeq.printSequence(out);
 325                                 }
 326                         }
 327                         m->gobble(in);
 328                 }
 329                 in.close();
 330                 out.close();
 331
 332                 if (wroteSomething == false) {  m->mothurOut("Your fasta file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 333                 outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 334
 335                 return 0;
 336
 337         }
 338         catch(exception& e) {
 339                 m->errorOut(e, "RemoveLineageCommand", "readFasta");
 340                 exit(1);
 341         }
 342 }
 343 //**********************************************************************************************************************
 344 int RemoveLineageCommand::readList(){
 345         try {
 346                 string thisOutputDir = outputDir;
 347                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 348                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 349
 350                 ofstream out;
 351                 m->openOutputFile(outputFileName, out);
 352
 353                 ifstream in;
 354                 m->openInputFile(listfile, in);
 355
 356                 bool wroteSomething = false;
 357
 358                 while(!in.eof()){
 359                         //read in list vector
 360                         ListVector list(in);
 361
 362                         //make a new list vector
 363                         ListVector newList;
 364                         newList.setLabel(list.getLabel());
 365
 366                         //for each bin
 367                         for (int i = 0; i < list.getNumBins(); i++) {
 368                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 369
 370                                 //parse out names that are in accnos file
 371                                 string binnames = list.get(i);
 372
 373                                 string newNames = "";
 374                                 while (binnames.find_first_of(',') != -1) {
 375                                         string name = binnames.substr(0,binnames.find_first_of(','));
 376                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 377
 378                                         //if that name is in the .accnos file, add it
 379                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 380                                 }
 381
 382                                 //get last name
 383                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 384
 385                                 //if there are names in this bin add to new list
 386                                 if (newNames != "") {
 387                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 388                                         newList.push_back(newNames);
 389                                 }
 390                         }
 391
 392                         //print new listvector
 393                         if (newList.getNumBins() != 0) {
 394                                 wroteSomething = true;
 395                                 newList.print(out);
 396                         }
 397
 398                         m->gobble(in);
 399                 }
 400                 in.close();
 401                 out.close();
 402
 403                 if (wroteSomething == false) {  m->mothurOut("Your list file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 404                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 405
 406                 return 0;
 407
 408         }
 409         catch(exception& e) {
 410                 m->errorOut(e, "RemoveLineageCommand", "readList");
 411                 exit(1);
 412         }
 413 }
 414 //**********************************************************************************************************************
 415 int RemoveLineageCommand::readName(){
 416         try {
 417                 string thisOutputDir = outputDir;
 418                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 419                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 420
 421                 ofstream out;
 422                 m->openOutputFile(outputFileName, out);
 423
 424                 ifstream in;
 425                 m->openInputFile(namefile, in);
 426                 string name, firstCol, secondCol;
 427
 428                 bool wroteSomething = false;
 429
 430                 while(!in.eof()){
 431                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 432
 433                         in >> firstCol;
 434                         in >> secondCol;
 435
 436                         vector<string> parsedNames;
 437                         //parse second column saving each name
 438                         while (secondCol.find_first_of(',') != -1) {
 439                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 440                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 441                                 parsedNames.push_back(name);
 442                         }
 443
 444                         //get name after last ,
 445                         parsedNames.push_back(secondCol);
 446
 447                         vector<string> validSecond;  validSecond.clear();
 448                         for (int i = 0; i < parsedNames.size(); i++) {
 449                                 if (names.count(parsedNames[i]) == 0) {
 450                                         validSecond.push_back(parsedNames[i]);
 451                                 }
 452                         }
 453
 454                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 455                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 456                         }else {
 457                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 458                                 if (names.count(firstCol) == 0) {
 459
 460                                         wroteSomething = true;
 461
 462                                         out << firstCol << '\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                                         //make first name in set you come to first column and then add the remaining names to second column
 469                                 }else {
 470
 471                                         //you want part of this row
 472                                         if (validSecond.size() != 0) {
 473
 474                                                 wroteSomething = true;
 475
 476                                                 out << validSecond[0] << '\t';
 477
 478                                                 //you know you have at least one valid second since first column is valid
 479                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 480                                                 out << validSecond[validSecond.size()-1] << endl;
 481                                         }
 482                                 }
 483                         }
 484                         m->gobble(in);
 485                 }
 486                 in.close();
 487                 out.close();
 488
 489                 if (wroteSomething == false) {  m->mothurOut("Your name file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 490                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 491
 492                 return 0;
 493         }
 494         catch(exception& e) {
 495                 m->errorOut(e, "RemoveLineageCommand", "readName");
 496                 exit(1);
 497         }
 498 }
 499
 500 //**********************************************************************************************************************
 501 int RemoveLineageCommand::readGroup(){
 502         try {
 503                 string thisOutputDir = outputDir;
 504                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 505                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 506
 507                 ofstream out;
 508                 m->openOutputFile(outputFileName, out);
 509
 510                 ifstream in;
 511                 m->openInputFile(groupfile, in);
 512                 string name, group;
 513
 514                 bool wroteSomething = false;
 515
 516                 while(!in.eof()){
 517                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 518
 519                         in >> name;                             //read from first column
 520                         in >> group;                    //read from second column
 521
 522                         //if this name is in the accnos file
 523                         if (names.count(name) == 0) {
 524                                 wroteSomething = true;
 525                                 out << name << '\t' << group << endl;
 526                         }
 527
 528                         m->gobble(in);
 529                 }
 530                 in.close();
 531                 out.close();
 532
 533                 if (wroteSomething == false) {  m->mothurOut("Your group file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 534                 outputNames.push_back(outputFileName); outputTypes["group"].push_back(outputFileName);
 535
 536                 return 0;
 537         }
 538         catch(exception& e) {
 539                 m->errorOut(e, "RemoveLineageCommand", "readGroup");
 540                 exit(1);
 541         }
 542 }
 543 //**********************************************************************************************************************
 544 int RemoveLineageCommand::readTax(){
 545         try {
 546                 string thisOutputDir = outputDir;
 547                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 548                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 549                 ofstream out;
 550                 m->openOutputFile(outputFileName, out);
 551
 552                 ifstream in;
 553                 m->openInputFile(taxfile, in);
 554                 string name, tax;
 555
 556                 bool wroteSomething = false;
 557
 558                 bool taxonsHasConfidence = false;
 559                 vector< map<string, int> > searchTaxons;
 560                 string noConfidenceTaxons = taxons;
 561                 int hasConPos = taxons.find_first_of('(');
 562                 if (hasConPos != string::npos) {
 563                         taxonsHasConfidence = true;
 564                         searchTaxons = getTaxons(taxons);
 565                         noConfidenceTaxons = removeConfidences(taxons);
 566                 }
 567
 568
 569                 while(!in.eof()){
 570
 571                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 572
 573                         in >> name;                             //read from first column
 574                         in >> tax;                      //read from second column
 575
 576                         string newtax = tax;
 577
 578                         //if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
 579                         if (!taxonsHasConfidence) {
 580
 581                                 int hasConfidences = tax.find_first_of('(');
 582                                 if (hasConfidences != string::npos) {
 583                                         newtax = removeConfidences(tax);
 584                                 }
 585
 586                                 int pos = newtax.find(taxons);
 587
 588                                 if (pos == string::npos) {
 589                                         wroteSomething = true;
 590                                         out << name << '\t' << tax << endl;
 591                                 }else{ //this sequence contains the taxon the user wants to remove
 592                                         names.insert(name);
 593                                 }
 594
 595                         }else{//if taxons has them and you don't them remove taxons
 596                                 int hasConfidences = tax.find_first_of('(');
 597                                 if (hasConfidences == string::npos) {
 598
 599                                         int pos = newtax.find(noConfidenceTaxons);
 600
 601                                         if (pos == string::npos) {
 602                                                 wroteSomething = true;
 603                                                 out << name << '\t' << tax << endl;
 604                                         }else{ //this sequence contains the taxon the user wants to remove
 605                                                 names.insert(name);
 606                                         }
 607                                 }else { //both have confidences so we want to make sure the users confidences are greater then or equal to the taxons
 608                                         //first remove confidences from both and see if the taxonomy exists
 609
 610                                         string noNewTax = tax;
 611                                         int hasConfidences = tax.find_first_of('(');
 612                                         if (hasConfidences != string::npos) {
 613                                                 noNewTax = removeConfidences(tax);
 614                                         }
 615
 616                                         int pos = noNewTax.find(noConfidenceTaxons);
 617
 618                                         if (pos != string::npos) { //if yes, then are the confidences okay
 619
 620                                                 bool remove = false;
 621                                                 vector< map<string, int> > usersTaxon = getTaxons(newtax);
 622
 623                                                 //the usersTaxon is most likely longer than the searchTaxons, and searchTaxon[0] may relate to userTaxon[4]
 624                                                 //we want to "line them up", so we will find the the index where the searchstring starts
 625                                                 int index = 0;
 626                                                 for (int i = 0; i < usersTaxon.size(); i++) {
 627
 628                                                         if (usersTaxon[i].begin()->first == searchTaxons[0].begin()->first) {
 629                                                                 index = i;
 630                                                                 int spot = 0;
 631                                                                 bool goodspot = true;
 632                                                                 //is this really the start, or are we dealing with a taxon of the same name?
 633                                                                 while ((spot < searchTaxons.size()) && ((i+spot) < usersTaxon.size())) {
 634                                                                         if (usersTaxon[i+spot].begin()->first != searchTaxons[spot].begin()->first) { goodspot = false; break; }
 635                                                                         else { spot++; }
 636                                                                 }
 637
 638                                                                 if (goodspot) { break; }
 639                                                         }
 640                                                 }
 641
 642                                                 for (int i = 0; i < searchTaxons.size(); i++) {
 643
 644                                                         if ((i+index) < usersTaxon.size()) { //just in case, should never be false
 645                                                                 if (usersTaxon[i+index].begin()->second < searchTaxons[i].begin()->second) { //is the users cutoff less than the search taxons
 646                                                                         remove = true;
 647                                                                         break;
 648                                                                 }
 649                                                         }else {
 650                                                                 remove = true;
 651                                                                 break;
 652                                                         }
 653                                                 }
 654
 655                                                 //passed the test so remove you
 656                                                 if (remove) {
 657                                                         names.insert(name);
 658                                                 }else {
 659                                                         wroteSomething = true;
 660                                                         out << name << '\t' << tax << endl;
 661                                                 }
 662                                         }else {
 663                                                 wroteSomething = true;
 664                                                 out << name << '\t' << tax << endl;
 665                                         }
 666                                 }
 667                         }
 668
 669
 670
 671
 672                         m->gobble(in);
 673                 }
 674                 in.close();
 675                 out.close();
 676
 677                 if (!wroteSomething) { m->mothurOut("Your taxonomy file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 678                 outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
 679
 680                 return 0;
 681
 682         }
 683         catch(exception& e) {
 684                 m->errorOut(e, "RemoveLineageCommand", "readTax");
 685                 exit(1);
 686         }
 687 }
 688 /**************************************************************************************************/
 689 vector< map<string, int> > RemoveLineageCommand::getTaxons(string tax) {
 690         try {
 691
 692                 vector< map<string, int> > t;
 693                 string taxon = "";
 694                 int taxLength = tax.length();
 695                 for(int i=0;i<taxLength;i++){
 696                         if(tax[i] == ';'){
 697
 698                                 int openParen = taxon.find_first_of('(');
 699                                 int closeParen = taxon.find_last_of(')');
 700
 701                                 string newtaxon, confidence;
 702                                 if ((openParen != string::npos) && (closeParen != string::npos)) {
 703                                         newtaxon = taxon.substr(0, openParen); //rip off confidence
 704                                         confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
 705                                 }else{
 706                                         newtaxon = taxon;
 707                                         confidence = "0";
 708                                 }
 709                                 int con = 0;
 710                                 convert(confidence, con);
 711
 712                                 map<string, int> temp;
 713                                 temp[newtaxon] = con;
 714                                 t.push_back(temp);
 715
 716                                 taxon = "";
 717                         }
 718                         else{
 719                                 taxon += tax[i];
 720                         }
 721                 }
 722
 723                 return t;
 724         }
 725         catch(exception& e) {
 726                 m->errorOut(e, "RemoveLineageCommand", "getTaxons");
 727                 exit(1);
 728         }
 729 }
 730 /**************************************************************************************************/
 731 string RemoveLineageCommand::removeConfidences(string tax) {
 732         try {
 733
 734                 string taxon = "";
 735                 int taxLength = tax.length();
 736                 for(int i=0;i<taxLength;i++){
 737                         if(tax[i] == ';'){
 738                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 739                                 taxon += ";";
 740                         }
 741                         else{
 742                                 taxon += tax[i];
 743                         }
 744                 }
 745
 746                 return taxon;
 747         }
 748         catch(exception& e) {
 749                 m->errorOut(e, "RemoveLineageCommand", "removeConfidences");
 750                 exit(1);
 751         }
 752 }
 753 //**********************************************************************************************************************
 754 //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
 755 int RemoveLineageCommand::readAlign(){
 756         try {
 757                 string thisOutputDir = outputDir;
 758                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 759                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 760
 761                 ofstream out;
 762                 m->openOutputFile(outputFileName, out);
 763
 764                 ifstream in;
 765                 m->openInputFile(alignfile, in);
 766                 string name, junk;
 767
 768                 bool wroteSomething = false;
 769
 770                 //read column headers
 771                 for (int i = 0; i < 16; i++) {
 772                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 773                         else                    {       break;                  }
 774                 }
 775                 out << endl;
 776
 777                 while(!in.eof()){
 778                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 779
 780                         in >> name;                             //read from first column
 781
 782                         //if this name is in the accnos file
 783                         if (names.count(name) == 0) {
 784                                 wroteSomething = true;
 785
 786                                 out << name << '\t';
 787
 788                                 //read rest
 789                                 for (int i = 0; i < 15; i++) {
 790                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 791                                         else                    {       break;                  }
 792                                 }
 793                                 out << endl;
 794
 795                         }else {//still read just don't do anything with it
 796
 797                                 //read rest
 798                                 for (int i = 0; i < 15; i++) {
 799                                         if (!in.eof())  {       in >> junk;             }
 800                                         else                    {       break;                  }
 801                                 }
 802                         }
 803
 804                         m->gobble(in);
 805                 }
 806                 in.close();
 807                 out.close();
 808
 809                 if (wroteSomething == false) {  m->mothurOut("Your align file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 810                 outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
 811
 812                 return 0;
 813
 814         }
 815         catch(exception& e) {
 816                 m->errorOut(e, "RemoveLineageCommand", "readAlign");
 817                 exit(1);
 818         }
 819 }
 820 //**********************************************************************************************************************
 821