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