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                         else { m->setFastaFile(fastafile); }
 175
 176                         namefile = validParameter.validFile(parameters, "name", true);
 177                         if (namefile == "not open") { abort = true; }
 178                         else if (namefile == "not found") {  namefile = "";  }
 179                         else { m->setNameFile(namefile); }
 180
 181                         groupfile = validParameter.validFile(parameters, "group", true);
 182                         if (groupfile == "not open") { abort = true; }
 183                         else if (groupfile == "not found") {  groupfile = "";  }
 184                         else { m->setGroupFile(groupfile); }
 185
 186                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 187                         if (alignfile == "not open") { abort = true; }
 188                         else if (alignfile == "not found") {  alignfile = "";  }
 189
 190                         listfile = validParameter.validFile(parameters, "list", true);
 191                         if (listfile == "not open") { abort = true; }
 192                         else if (listfile == "not found") {  listfile = "";  }
 193                         else { m->setListFile(listfile); }
 194
 195                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 196                         if (taxfile == "not open") { abort = true; }
 197                         else if (taxfile == "not found") {
 198                                 taxfile = m->getTaxonomyFile();
 199                                 if (taxfile != "") { m->mothurOut("Using " + taxfile + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
 200                                 else {  m->mothurOut("You have no current taxonomy file and the taxonomy parameter is required."); m->mothurOutEndLine(); abort = true; }
 201                         }else { m->setTaxonomyFile(taxfile); }
 202
 203                         string usedDups = "true";
 204                         string temp = validParameter.validFile(parameters, "dups", false);
 205                         if (temp == "not found") {
 206                                 if (namefile != "") {  temp = "true";                                   }
 207                                 else                            {  temp = "false"; usedDups = "";       }
 208                         }
 209                         dups = m->isTrue(temp);
 210
 211                         taxons = validParameter.validFile(parameters, "taxon", false);
 212                         if (taxons == "not found") { taxons = "";  m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine();  abort = true;  }
 213                         else {
 214                                 //rip off quotes
 215                                 if (taxons[0] == '\'') {  taxons = taxons.substr(1); }
 216                                 if (taxons[(taxons.length()-1)] == '\'') {  taxons = taxons.substr(0, (taxons.length()-1)); }
 217                         }
 218                         m->splitAtChar(taxons, listOfTaxons, '-');
 219
 220                         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; }
 221
 222                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 223
 224                 }
 225
 226         }
 227         catch(exception& e) {
 228                 m->errorOut(e, "RemoveLineageCommand", "RemoveLineageCommand");
 229                 exit(1);
 230         }
 231 }
 232 //**********************************************************************************************************************
 233
 234 int RemoveLineageCommand::execute(){
 235         try {
 236
 237                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 238
 239                 if (m->control_pressed) { return 0; }
 240
 241                 //read through the correct file and output lines you want to keep
 242                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to remove
 243                 if (namefile != "")                     {               readName();             }
 244                 if (fastafile != "")            {               readFasta();    }
 245                 if (groupfile != "")            {               readGroup();    }
 246                 if (alignfile != "")            {               readAlign();    }
 247                 if (listfile != "")                     {               readList();             }
 248
 249
 250                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 251
 252                 if (outputNames.size() != 0) {
 253                         m->mothurOutEndLine();
 254                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 255                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 256                         m->mothurOutEndLine();
 257
 258                         //set fasta file as new current fastafile
 259                         string current = "";
 260                         itTypes = outputTypes.find("fasta");
 261                         if (itTypes != outputTypes.end()) {
 262                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 263                         }
 264
 265                         itTypes = outputTypes.find("name");
 266                         if (itTypes != outputTypes.end()) {
 267                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 268                         }
 269
 270                         itTypes = outputTypes.find("group");
 271                         if (itTypes != outputTypes.end()) {
 272                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 273                         }
 274
 275                         itTypes = outputTypes.find("list");
 276                         if (itTypes != outputTypes.end()) {
 277                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 278                         }
 279
 280                         itTypes = outputTypes.find("taxonomy");
 281                         if (itTypes != outputTypes.end()) {
 282                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 283                         }
 284                 }
 285
 286                 return 0;
 287         }
 288
 289         catch(exception& e) {
 290                 m->errorOut(e, "RemoveLineageCommand", "execute");
 291                 exit(1);
 292         }
 293 }
 294
 295 //**********************************************************************************************************************
 296 int RemoveLineageCommand::readFasta(){
 297         try {
 298                 string thisOutputDir = outputDir;
 299                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 300                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 301
 302                 ofstream out;
 303                 m->openOutputFile(outputFileName, out);
 304
 305                 ifstream in;
 306                 m->openInputFile(fastafile, in);
 307                 string name;
 308
 309                 bool wroteSomething = false;
 310
 311                 while(!in.eof()){
 312                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 313
 314                         Sequence currSeq(in);
 315                         name = currSeq.getName();
 316
 317                         if (name != "") {
 318                                 //if this name is in the accnos file
 319                                 if (names.count(name) == 0) {
 320                                         wroteSomething = true;
 321
 322                                         currSeq.printSequence(out);
 323                                 }
 324                         }
 325                         m->gobble(in);
 326                 }
 327                 in.close();
 328                 out.close();
 329
 330                 if (wroteSomething == false) {  m->mothurOut("Your fasta file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 331                 outputNames.push_back(outputFileName); outputTypes["fasta"].push_back(outputFileName);
 332
 333                 return 0;
 334
 335         }
 336         catch(exception& e) {
 337                 m->errorOut(e, "RemoveLineageCommand", "readFasta");
 338                 exit(1);
 339         }
 340 }
 341 //**********************************************************************************************************************
 342 int RemoveLineageCommand::readList(){
 343         try {
 344                 string thisOutputDir = outputDir;
 345                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 346                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 347
 348                 ofstream out;
 349                 m->openOutputFile(outputFileName, out);
 350
 351                 ifstream in;
 352                 m->openInputFile(listfile, in);
 353
 354                 bool wroteSomething = false;
 355
 356                 while(!in.eof()){
 357                         //read in list vector
 358                         ListVector list(in);
 359
 360                         //make a new list vector
 361                         ListVector newList;
 362                         newList.setLabel(list.getLabel());
 363
 364                         //for each bin
 365                         for (int i = 0; i < list.getNumBins(); i++) {
 366                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 367
 368                                 //parse out names that are in accnos file
 369                                 string binnames = list.get(i);
 370
 371                                 string newNames = "";
 372                                 while (binnames.find_first_of(',') != -1) {
 373                                         string name = binnames.substr(0,binnames.find_first_of(','));
 374                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 375
 376                                         //if that name is in the .accnos file, add it
 377                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 378                                 }
 379
 380                                 //get last name
 381                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 382
 383                                 //if there are names in this bin add to new list
 384                                 if (newNames != "") {
 385                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 386                                         newList.push_back(newNames);
 387                                 }
 388                         }
 389
 390                         //print new listvector
 391                         if (newList.getNumBins() != 0) {
 392                                 wroteSomething = true;
 393                                 newList.print(out);
 394                         }
 395
 396                         m->gobble(in);
 397                 }
 398                 in.close();
 399                 out.close();
 400
 401                 if (wroteSomething == false) {  m->mothurOut("Your list file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 402                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 403
 404                 return 0;
 405
 406         }
 407         catch(exception& e) {
 408                 m->errorOut(e, "RemoveLineageCommand", "readList");
 409                 exit(1);
 410         }
 411 }
 412 //**********************************************************************************************************************
 413 int RemoveLineageCommand::readName(){
 414         try {
 415                 string thisOutputDir = outputDir;
 416                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 417                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 418
 419                 ofstream out;
 420                 m->openOutputFile(outputFileName, out);
 421
 422                 ifstream in;
 423                 m->openInputFile(namefile, in);
 424                 string name, firstCol, secondCol;
 425
 426                 bool wroteSomething = false;
 427
 428                 while(!in.eof()){
 429                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 430
 431                         in >> firstCol;
 432                         in >> secondCol;
 433
 434                         vector<string> parsedNames;
 435                         m->splitAtComma(secondCol, parsedNames);
 436
 437                         vector<string> validSecond;  validSecond.clear();
 438                         for (int i = 0; i < parsedNames.size(); i++) {
 439                                 if (names.count(parsedNames[i]) == 0) {
 440                                         validSecond.push_back(parsedNames[i]);
 441                                 }
 442                         }
 443
 444                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 445                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 446                         }else {
 447                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 448                                 if (names.count(firstCol) == 0) {
 449
 450                                         wroteSomething = true;
 451
 452                                         out << firstCol << '\t';
 453
 454                                         //you know you have at least one valid second since first column is valid
 455                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 456                                         out << validSecond[validSecond.size()-1] << endl;
 457
 458                                         //make first name in set you come to first column and then add the remaining names to second column
 459                                 }else {
 460
 461                                         //you want part of this row
 462                                         if (validSecond.size() != 0) {
 463
 464                                                 wroteSomething = true;
 465
 466                                                 out << validSecond[0] << '\t';
 467
 468                                                 //you know you have at least one valid second since first column is valid
 469                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 470                                                 out << validSecond[validSecond.size()-1] << endl;
 471                                         }
 472                                 }
 473                         }
 474                         m->gobble(in);
 475                 }
 476                 in.close();
 477                 out.close();
 478
 479                 if (wroteSomething == false) {  m->mothurOut("Your name file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 480                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 481
 482                 return 0;
 483         }
 484         catch(exception& e) {
 485                 m->errorOut(e, "RemoveLineageCommand", "readName");
 486                 exit(1);
 487         }
 488 }
 489
 490 //**********************************************************************************************************************
 491 int RemoveLineageCommand::readGroup(){
 492         try {
 493                 string thisOutputDir = outputDir;
 494                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 495                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 496
 497                 ofstream out;
 498                 m->openOutputFile(outputFileName, out);
 499
 500                 ifstream in;
 501                 m->openInputFile(groupfile, in);
 502                 string name, group;
 503
 504                 bool wroteSomething = false;
 505
 506                 while(!in.eof()){
 507                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 508
 509                         in >> name;                             //read from first column
 510                         in >> group;                    //read from second column
 511
 512                         //if this name is in the accnos file
 513                         if (names.count(name) == 0) {
 514                                 wroteSomething = true;
 515                                 out << name << '\t' << group << endl;
 516                         }
 517
 518                         m->gobble(in);
 519                 }
 520                 in.close();
 521                 out.close();
 522
 523                 if (wroteSomething == false) {  m->mothurOut("Your group file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 524                 outputNames.push_back(outputFileName); outputTypes["group"].push_back(outputFileName);
 525
 526                 return 0;
 527         }
 528         catch(exception& e) {
 529                 m->errorOut(e, "RemoveLineageCommand", "readGroup");
 530                 exit(1);
 531         }
 532 }
 533 //**********************************************************************************************************************
 534 int RemoveLineageCommand::readTax(){
 535         try {
 536                 string thisOutputDir = outputDir;
 537                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 538                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 539                 ofstream out;
 540                 m->openOutputFile(outputFileName, out);
 541
 542                 ifstream in;
 543                 m->openInputFile(taxfile, in);
 544                 string name, tax;
 545
 546                 bool wroteSomething = false;
 547
 548                 vector<bool> taxonsHasConfidence; taxonsHasConfidence.resize(listOfTaxons.size(), false);
 549                 vector< vector< map<string, float> > > searchTaxons; searchTaxons.resize(listOfTaxons.size());
 550                 vector<string> noConfidenceTaxons; noConfidenceTaxons.resize(listOfTaxons.size(), "");
 551
 552                 for (int i = 0; i < listOfTaxons.size(); i++) {
 553                         noConfidenceTaxons[i] = listOfTaxons[i];
 554                         int hasConPos = listOfTaxons[i].find_first_of('(');
 555                         if (hasConPos != string::npos) {
 556                                 taxonsHasConfidence[i] = true;
 557                                 searchTaxons[i] = getTaxons(listOfTaxons[i]);
 558                                 noConfidenceTaxons[i] = removeConfidences(listOfTaxons[i]);
 559                         }
 560                 }
 561
 562
 563                 while(!in.eof()){
 564
 565                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 566
 567                         in >> name;                             //read from first column
 568                         in >> tax;                      //read from second column
 569
 570                         bool remove = false;
 571
 572                         for (int j = 0; j < listOfTaxons.size(); j++) {
 573                                 string newtax = tax;
 574
 575                                 //if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
 576                                 if (!taxonsHasConfidence[j]) {
 577
 578                                         int hasConfidences = tax.find_first_of('(');
 579                                         if (hasConfidences != string::npos) {
 580                                                 newtax = removeConfidences(tax);
 581                                         }
 582
 583                                         int pos = newtax.find(noConfidenceTaxons[j]);
 584
 585                                         if (pos == string::npos) {
 586                                                 //wroteSomething = true;
 587                                                 //out << name << '\t' << tax << endl;
 588                                         }else{ //this sequence contains the taxon the user wants to remove
 589                                                 names.insert(name);
 590                                                 remove=true; break;
 591                                         }
 592
 593                                 }else{//if taxons has them and you don't them remove taxons
 594                                         int hasConfidences = tax.find_first_of('(');
 595                                         if (hasConfidences == string::npos) {
 596
 597                                                 int pos = newtax.find(noConfidenceTaxons[j]);
 598
 599                                                 if (pos == string::npos) {
 600                                                         //wroteSomething = true;
 601                                                         //out << name << '\t' << tax << endl;
 602                                                 }else{ //this sequence contains the taxon the user wants to remove
 603                                                         names.insert(name);
 604                                                         remove=true; break;
 605                                                 }
 606                                         }else { //both have confidences so we want to make sure the users confidences are greater then or equal to the taxons
 607                                                 //first remove confidences from both and see if the taxonomy exists
 608
 609                                                 string noNewTax = tax;
 610                                                 int hasConfidences = tax.find_first_of('(');
 611                                                 if (hasConfidences != string::npos) {
 612                                                         noNewTax = removeConfidences(tax);
 613                                                 }
 614
 615                                                 int pos = noNewTax.find(noConfidenceTaxons[j]);
 616
 617                                                 if (pos != string::npos) { //if yes, then are the confidences okay
 618
 619                                                         vector< map<string, float> > usersTaxon = getTaxons(newtax);
 620
 621                                                         //the usersTaxon is most likely longer than the searchTaxons, and searchTaxon[0] may relate to userTaxon[4]
 622                                                         //we want to "line them up", so we will find the the index where the searchstring starts
 623                                                         int index = 0;
 624                                                         for (int i = 0; i < usersTaxon.size(); i++) {
 625
 626                                                                 if (usersTaxon[i].begin()->first == searchTaxons[j][0].begin()->first) {
 627                                                                         index = i;
 628                                                                         int spot = 0;
 629                                                                         bool goodspot = true;
 630                                                                         //is this really the start, or are we dealing with a taxon of the same name?
 631                                                                         while ((spot < searchTaxons[j].size()) && ((i+spot) < usersTaxon.size())) {
 632                                                                                 if (usersTaxon[i+spot].begin()->first != searchTaxons[j][spot].begin()->first) { goodspot = false; break; }
 633                                                                                 else { spot++; }
 634                                                                         }
 635
 636                                                                         if (goodspot) { break; }
 637                                                                 }
 638                                                         }
 639
 640                                                         for (int i = 0; i < searchTaxons[j].size(); i++) {
 641
 642                                                                 if ((i+index) < usersTaxon.size()) { //just in case, should never be false
 643                                                                         if (usersTaxon[i+index].begin()->second < searchTaxons[j][i].begin()->second) { //is the users cutoff less than the search taxons
 644                                                                                 remove = true;
 645                                                                                 break;
 646                                                                         }
 647                                                                 }else {
 648                                                                         remove = true;
 649                                                                         break;
 650                                                                 }
 651                                                         }
 652
 653                                                         //passed the test so remove you
 654                                                         if (remove) {
 655                                                                 names.insert(name);
 656                                                                 remove=true; break;
 657                                                         }else {
 658                                                                 //wroteSomething = true;
 659                                                                 //out << name << '\t' << tax << endl;
 660                                                         }
 661                                                 }else {
 662                                                         //wroteSomething = true;
 663                                                         //out << name << '\t' << tax << endl;
 664                                                 }
 665                                         }
 666                                 }
 667
 668                         }
 669
 670                         if (!remove) {  wroteSomething = true; out << name << '\t' << tax << endl; }
 671                         m->gobble(in);
 672                 }
 673                 in.close();
 674                 out.close();
 675
 676                 if (!wroteSomething) { m->mothurOut("Your taxonomy file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 677                 outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
 678
 679                 return 0;
 680
 681         }
 682         catch(exception& e) {
 683                 m->errorOut(e, "RemoveLineageCommand", "readTax");
 684                 exit(1);
 685         }
 686 }
 687 /**************************************************************************************************/
 688 vector< map<string, float> > RemoveLineageCommand::getTaxons(string tax) {
 689         try {
 690
 691                 vector< map<string, float> > t;
 692                 string taxon = "";
 693                 int taxLength = tax.length();
 694                 for(int i=0;i<taxLength;i++){
 695                         if(tax[i] == ';'){
 696
 697                                 int openParen = taxon.find_first_of('(');
 698                                 int closeParen = taxon.find_last_of(')');
 699
 700                                 string newtaxon, confidence;
 701                                 if ((openParen != string::npos) && (closeParen != string::npos)) {
 702                                         newtaxon = taxon.substr(0, openParen); //rip off confidence
 703                                         confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
 704                                 }else{
 705                                         newtaxon = taxon;
 706                                         confidence = "0";
 707                                 }
 708                                 float con = 0;
 709                                 convert(confidence, con);
 710
 711                                 map<string, float> temp;
 712                                 temp[newtaxon] = con;
 713                                 t.push_back(temp);
 714
 715                                 taxon = "";
 716                         }
 717                         else{
 718                                 taxon += tax[i];
 719                         }
 720                 }
 721
 722                 return t;
 723         }
 724         catch(exception& e) {
 725                 m->errorOut(e, "RemoveLineageCommand", "getTaxons");
 726                 exit(1);
 727         }
 728 }
 729 /**************************************************************************************************/
 730 string RemoveLineageCommand::removeConfidences(string tax) {
 731         try {
 732
 733                 string taxon = "";
 734                 int taxLength = tax.length();
 735                 for(int i=0;i<taxLength;i++){
 736                         if(tax[i] == ';'){
 737                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 738                                 taxon += ";";
 739                         }
 740                         else{
 741                                 taxon += tax[i];
 742                         }
 743                 }
 744
 745                 return taxon;
 746         }
 747         catch(exception& e) {
 748                 m->errorOut(e, "RemoveLineageCommand", "removeConfidences");
 749                 exit(1);
 750         }
 751 }
 752 //**********************************************************************************************************************
 753 //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
 754 int RemoveLineageCommand::readAlign(){
 755         try {
 756                 string thisOutputDir = outputDir;
 757                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 758                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 759
 760                 ofstream out;
 761                 m->openOutputFile(outputFileName, out);
 762
 763                 ifstream in;
 764                 m->openInputFile(alignfile, in);
 765                 string name, junk;
 766
 767                 bool wroteSomething = false;
 768
 769                 //read column headers
 770                 for (int i = 0; i < 16; i++) {
 771                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 772                         else                    {       break;                  }
 773                 }
 774                 out << endl;
 775
 776                 while(!in.eof()){
 777                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 778
 779                         in >> name;                             //read from first column
 780
 781                         //if this name is in the accnos file
 782                         if (names.count(name) == 0) {
 783                                 wroteSomething = true;
 784
 785                                 out << name << '\t';
 786
 787                                 //read rest
 788                                 for (int i = 0; i < 15; i++) {
 789                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 790                                         else                    {       break;                  }
 791                                 }
 792                                 out << endl;
 793
 794                         }else {//still read just don't do anything with it
 795
 796                                 //read rest
 797                                 for (int i = 0; i < 15; i++) {
 798                                         if (!in.eof())  {       in >> junk;             }
 799                                         else                    {       break;                  }
 800                                 }
 801                         }
 802
 803                         m->gobble(in);
 804                 }
 805                 in.close();
 806                 out.close();
 807
 808                 if (wroteSomething == false) {  m->mothurOut("Your align file contains only sequences from " + taxons + "."); m->mothurOutEndLine();  }
 809                 outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
 810
 811                 return 0;
 812
 813         }
 814         catch(exception& e) {
 815                 m->errorOut(e, "RemoveLineageCommand", "readAlign");
 816                 exit(1);
 817         }
 818 }
 819 //**********************************************************************************************************************
 820