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