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