getlineagecommand.cpp

   1 /*
   2  *  getlineagecommand.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 "getlineagecommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15 vector<string> GetLineageCommand::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, "GetLineageCommand", "setParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 string GetLineageCommand::getHelpString(){
  39         try {
  40                 string helpString = "";
  41                 helpString += "The get.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 from the taxon requested.\n";
  43                 helpString += "The get.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 get and is required.\n";
  46                 helpString += "You may enter your taxons with confidence scores, doing so will get only those sequences that belong to the taxonomy and whose cofidence scores is above the scores you give.\n";
  47                 helpString += "If they belong to the taxonomy and have confidences below those you provide the sequence will not be selected.\n";
  48                 helpString += "The get.lineage command should be in the following format: get.lineage(taxonomy=yourTaxonomyFile, taxon=yourTaxons).\n";
  49                 helpString += "Example get.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 get.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, "GetLineageCommand", "getHelpString");
  57                 exit(1);
  58         }
  59 }
  60
  61 //**********************************************************************************************************************
  62 GetLineageCommand::GetLineageCommand(){
  63         try {
  64                 abort = true; calledHelp = true;
  65                 setParameters();
  66                 vector<string> tempOutNames;
  67                 outputTypes["fasta"] = tempOutNames;
  68                 outputTypes["taxonomy"] = tempOutNames;
  69                 outputTypes["name"] = tempOutNames;
  70                 outputTypes["group"] = tempOutNames;
  71                 outputTypes["alignreport"] = tempOutNames;
  72                 outputTypes["list"] = tempOutNames;
  73         }
  74         catch(exception& e) {
  75                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
  76                 exit(1);
  77         }
  78 }
  79 //**********************************************************************************************************************
  80 GetLineageCommand::GetLineageCommand(string option)  {
  81         try {
  82                 abort = false; calledHelp = false;
  83
  84                 //allow user to run help
  85                 if(option == "help") { help(); abort = true; calledHelp = true; }
  86                 else if(option == "citation") { citation(); 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                         m->splitAtChar(taxons, listOfTaxons, '-');
 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
 218         }
 219         catch(exception& e) {
 220                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
 221                 exit(1);
 222         }
 223 }
 224 //**********************************************************************************************************************
 225
 226 int GetLineageCommand::execute(){
 227         try {
 228
 229                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 230
 231                 if (m->control_pressed) { return 0; }
 232
 233                 //read through the correct file and output lines you want to keep
 234                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to get
 235                 if (namefile != "")                     {               readName();             }
 236                 if (fastafile != "")            {               readFasta();    }
 237                 if (groupfile != "")            {               readGroup();    }
 238                 if (alignfile != "")            {               readAlign();    }
 239                 if (listfile != "")                     {               readList();             }
 240
 241
 242                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str());  } return 0; }
 243
 244                 if (outputNames.size() != 0) {
 245                         m->mothurOutEndLine();
 246                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 247                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 248                         m->mothurOutEndLine();
 249
 250                         //set fasta file as new current fastafile
 251                         string current = "";
 252                         itTypes = outputTypes.find("fasta");
 253                         if (itTypes != outputTypes.end()) {
 254                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 255                         }
 256
 257                         itTypes = outputTypes.find("name");
 258                         if (itTypes != outputTypes.end()) {
 259                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 260                         }
 261
 262                         itTypes = outputTypes.find("group");
 263                         if (itTypes != outputTypes.end()) {
 264                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 265                         }
 266
 267                         itTypes = outputTypes.find("list");
 268                         if (itTypes != outputTypes.end()) {
 269                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 270                         }
 271
 272                         itTypes = outputTypes.find("taxonomy");
 273                         if (itTypes != outputTypes.end()) {
 274                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 275                         }
 276                 }
 277
 278                 return 0;
 279         }
 280
 281         catch(exception& e) {
 282                 m->errorOut(e, "GetLineageCommand", "execute");
 283                 exit(1);
 284         }
 285 }
 286
 287 //**********************************************************************************************************************
 288 int GetLineageCommand::readFasta(){
 289         try {
 290                 string thisOutputDir = outputDir;
 291                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 292                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 293                 ofstream out;
 294                 m->openOutputFile(outputFileName, out);
 295
 296
 297                 ifstream in;
 298                 m->openInputFile(fastafile, in);
 299                 string name;
 300
 301                 bool wroteSomething = false;
 302
 303                 while(!in.eof()){
 304
 305                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 306
 307                         Sequence currSeq(in);
 308                         name = currSeq.getName();
 309
 310                         if (name != "") {
 311                                 //if this name is in the accnos file
 312                                 if (names.count(name) != 0) {
 313                                         wroteSomething = true;
 314
 315                                         currSeq.printSequence(out);
 316                                 }
 317                         }
 318                         m->gobble(in);
 319                 }
 320                 in.close();
 321                 out.close();
 322
 323                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 324                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 325
 326                 return 0;
 327
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "GetLineageCommand", "readFasta");
 331                 exit(1);
 332         }
 333 }
 334 //**********************************************************************************************************************
 335 int GetLineageCommand::readList(){
 336         try {
 337                 string thisOutputDir = outputDir;
 338                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 339                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 340                 ofstream out;
 341                 m->openOutputFile(outputFileName, out);
 342
 343                 ifstream in;
 344                 m->openInputFile(listfile, in);
 345
 346                 bool wroteSomething = false;
 347
 348                 while(!in.eof()){
 349
 350                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 351
 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
 362                                 //parse out names that are in accnos file
 363                                 string binnames = list.get(i);
 364
 365                                 string newNames = "";
 366                                 while (binnames.find_first_of(',') != -1) {
 367                                         string name = binnames.substr(0,binnames.find_first_of(','));
 368                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 369
 370                                         //if that name is in the .accnos file, add it
 371                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 372                                 }
 373
 374                                 //get last name
 375                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 376
 377                                 //if there are names in this bin add to new list
 378                                 if (newNames != "") {
 379                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 380                                         newList.push_back(newNames);
 381                                 }
 382                         }
 383
 384                         //print new listvector
 385                         if (newList.getNumBins() != 0) {
 386                                 wroteSomething = true;
 387                                 newList.print(out);
 388                         }
 389
 390                         m->gobble(in);
 391                 }
 392                 in.close();
 393                 out.close();
 394
 395                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 396                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 397
 398                 return 0;
 399
 400         }
 401         catch(exception& e) {
 402                 m->errorOut(e, "GetLineageCommand", "readList");
 403                 exit(1);
 404         }
 405 }
 406 //**********************************************************************************************************************
 407 int GetLineageCommand::readName(){
 408         try {
 409                 string thisOutputDir = outputDir;
 410                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 411                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 412                 ofstream out;
 413                 m->openOutputFile(outputFileName, out);
 414
 415
 416                 ifstream in;
 417                 m->openInputFile(namefile, in);
 418                 string name, firstCol, secondCol;
 419
 420                 bool wroteSomething = false;
 421
 422
 423                 while(!in.eof()){
 424
 425                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 426
 427                         in >> firstCol;
 428                         in >> secondCol;
 429
 430                         string hold = "";
 431                         if (dups) { hold = secondCol; }
 432
 433                         vector<string> parsedNames;
 434                         m->splitAtComma(secondCol, parsedNames);
 435
 436                         vector<string> validSecond;
 437                         for (int i = 0; i < parsedNames.size(); i++) {
 438                                 if (names.count(parsedNames[i]) != 0) {
 439                                         validSecond.push_back(parsedNames[i]);
 440                                 }
 441                         }
 442
 443                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 444                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 445                                 out << firstCol << '\t' << hold << endl;
 446                                 wroteSomething = true;
 447                         }else {
 448                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 449                                 if (names.count(firstCol) != 0) {
 450
 451                                         wroteSomething = true;
 452
 453                                         out << firstCol << '\t';
 454
 455                                         //you know you have at least one valid second since first column is valid
 456                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 457                                         out << validSecond[validSecond.size()-1] << endl;
 458
 459
 460                                 //make first name in set you come to first column and then add the remaining names to second column
 461                                 }else {
 462                                         //you want part of this row
 463                                         if (validSecond.size() != 0) {
 464
 465                                                 wroteSomething = true;
 466
 467                                                 out << validSecond[0] << '\t';
 468
 469                                                 //you know you have at least one valid second since first column is valid
 470                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 471                                                 out << validSecond[validSecond.size()-1] << endl;
 472                                         }
 473                                 }
 474                         }
 475                         m->gobble(in);
 476                 }
 477                 in.close();
 478                 out.close();
 479
 480                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 481                 outputNames.push_back(outputFileName);  outputTypes["name"].push_back(outputFileName);
 482
 483                 return 0;
 484
 485         }
 486         catch(exception& e) {
 487                 m->errorOut(e, "GetLineageCommand", "readName");
 488                 exit(1);
 489         }
 490 }
 491
 492 //**********************************************************************************************************************
 493 int GetLineageCommand::readGroup(){
 494         try {
 495                 string thisOutputDir = outputDir;
 496                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 497                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 498                 ofstream out;
 499                 m->openOutputFile(outputFileName, out);
 500
 501
 502                 ifstream in;
 503                 m->openInputFile(groupfile, in);
 504                 string name, group;
 505
 506                 bool wroteSomething = false;
 507
 508                 while(!in.eof()){
 509
 510                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 511
 512
 513                         in >> name;                             //read from first column
 514                         in >> group;                    //read from second column
 515
 516                         //if this name is in the accnos file
 517                         if (names.count(name) != 0) {
 518                                 wroteSomething = true;
 519
 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 file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 529                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 530
 531                 return 0;
 532
 533         }
 534         catch(exception& e) {
 535                 m->errorOut(e, "GetLineageCommand", "readGroup");
 536                 exit(1);
 537         }
 538 }
 539 //**********************************************************************************************************************
 540 int GetLineageCommand::readTax(){
 541         try {
 542                 string thisOutputDir = outputDir;
 543                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 544                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 545                 ofstream out;
 546                 m->openOutputFile(outputFileName, out);
 547
 548                 ifstream in;
 549                 m->openInputFile(taxfile, in);
 550                 string name, tax;
 551
 552                 //bool wroteSomething = false;
 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] = removeConfidences(listOfTaxons[i]);
 564                         }
 565                 }
 566
 567
 568                 while(!in.eof()){
 569
 570                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 571
 572                         in >> name;                             //read from first column
 573                         in >> tax;                      //read from second column
 574
 575                         for (int j = 0; j < listOfTaxons.size(); j++) {
 576
 577                                 string newtax = tax;
 578
 579                                 //if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
 580                                 if (!taxonsHasConfidence[j]) {
 581                                         int hasConfidences = tax.find_first_of('(');
 582                                         if (hasConfidences != string::npos) {
 583                                                 newtax = removeConfidences(tax);
 584                                         }
 585
 586                                         int pos = newtax.find(noConfidenceTaxons[j]);
 587
 588                                         if (pos != string::npos) { //this sequence contains the taxon the user wants
 589                                                 names.insert(name);
 590                                                 out << name << '\t' << tax << endl;
 591                                                 //since you belong to at least one of the taxons we want you are included so no need to search for other
 592                                                 break;
 593                                         }
 594                                 }else{//if listOfTaxons[i] has them and you don't them remove taxons
 595                                         int hasConfidences = tax.find_first_of('(');
 596                                         if (hasConfidences == string::npos) {
 597
 598                                                 int pos = newtax.find(noConfidenceTaxons[j]);
 599
 600                                                 if (pos != string::npos) { //this sequence contains the taxon the user wants
 601                                                         names.insert(name);
 602                                                         out << name << '\t' << tax << endl;
 603                                                         //since you belong to at least one of the taxons we want you are included so no need to search for other
 604                                                         break;
 605                                                 }
 606                                         }else { //both have confidences so we want to make sure the users confidences are greater then or equal to the taxons
 607                                                 //first remove confidences from both and see if the taxonomy exists
 608
 609                                                 string noNewTax = tax;
 610                                                 int hasConfidences = tax.find_first_of('(');
 611                                                 if (hasConfidences != string::npos) {
 612                                                         noNewTax = removeConfidences(tax);
 613                                                 }
 614
 615                                                 int pos = noNewTax.find(noConfidenceTaxons[j]);
 616
 617                                                 if (pos != string::npos) { //if yes, then are the confidences okay
 618
 619                                                         bool good = true;
 620                                                         vector< map<string, float> > usersTaxon = getTaxons(newtax);
 621
 622                                                         //the usersTaxon is most likely longer than the searchTaxons, and searchTaxon[0] may relate to userTaxon[4]
 623                                                         //we want to "line them up", so we will find the the index where the searchstring starts
 624                                                         int index = 0;
 625                                                         for (int i = 0; i < usersTaxon.size(); i++) {
 626
 627                                                                 if (usersTaxon[i].begin()->first == searchTaxons[j][0].begin()->first) {
 628                                                                         index = i;
 629                                                                         int spot = 0;
 630                                                                         bool goodspot = true;
 631                                                                         //is this really the start, or are we dealing with a taxon of the same name?
 632                                                                         while ((spot < searchTaxons[j].size()) && ((i+spot) < usersTaxon.size())) {
 633                                                                                 if (usersTaxon[i+spot].begin()->first != searchTaxons[j][spot].begin()->first) { goodspot = false; break; }
 634                                                                                 else { spot++; }
 635                                                                         }
 636
 637                                                                         if (goodspot) { break; }
 638                                                                 }
 639                                                         }
 640
 641                                                         for (int i = 0; i < searchTaxons[j].size(); i++) {
 642
 643                                                                 if ((i+index) < usersTaxon.size()) { //just in case, should never be false
 644                                                                         if (usersTaxon[i+index].begin()->second < searchTaxons[j][i].begin()->second) { //is the users cutoff less than the search taxons
 645                                                                                 good = false;
 646                                                                                 break;
 647                                                                         }
 648                                                                 }else {
 649                                                                         good = false;
 650                                                                         break;
 651                                                                 }
 652                                                         }
 653
 654                                                         //passed the test so add you
 655                                                         if (good) {
 656                                                                 names.insert(name);
 657                                                                 out << name << '\t' << tax << endl;
 658                                                                 break;
 659                                                         }
 660                                                 }
 661                                         }
 662                                 }
 663
 664                         }
 665
 666                         m->gobble(in);
 667                 }
 668                 in.close();
 669                 out.close();
 670
 671                 if (names.size() == 0) { m->mothurOut("Your taxonomy file does not contain any sequences from " + taxons + "."); m->mothurOutEndLine();  }
 672                 outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
 673
 674                 return 0;
 675
 676         }
 677         catch(exception& e) {
 678                 m->errorOut(e, "GetLineageCommand", "readTax");
 679                 exit(1);
 680         }
 681 }
 682 /**************************************************************************************************/
 683 vector< map<string, float> > GetLineageCommand::getTaxons(string tax) {
 684         try {
 685
 686                 vector< map<string, float> > t;
 687                 string taxon = "";
 688                 int taxLength = tax.length();
 689                 for(int i=0;i<taxLength;i++){
 690                         if(tax[i] == ';'){
 691
 692                                 int openParen = taxon.find_first_of('(');
 693                                 int closeParen = taxon.find_last_of(')');
 694
 695                                 string newtaxon, confidence;
 696                                 if ((openParen != string::npos) && (closeParen != string::npos)) {
 697                                         newtaxon = taxon.substr(0, openParen); //rip off confidence
 698                                         confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
 699                                 }else{
 700                                         newtaxon = taxon;
 701                                         confidence = "0";
 702                                 }
 703                                 float con = 0;
 704                                 convert(confidence, con);
 705
 706                                 map<string, float> temp;
 707                                 temp[newtaxon] = con;
 708                                 t.push_back(temp);
 709
 710                                 taxon = "";
 711                         }
 712                         else{
 713                                 taxon += tax[i];
 714                         }
 715                 }
 716
 717                 return t;
 718         }
 719         catch(exception& e) {
 720                 m->errorOut(e, "GetLineageCommand", "getTaxons");
 721                 exit(1);
 722         }
 723 }
 724 /**************************************************************************************************/
 725 string GetLineageCommand::removeConfidences(string tax) {
 726         try {
 727
 728                 string taxon = "";
 729                 int taxLength = tax.length();
 730                 for(int i=0;i<taxLength;i++){
 731                         if(tax[i] == ';'){
 732                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 733                                 taxon += ";";
 734                         }
 735                         else{
 736                                 taxon += tax[i];
 737                         }
 738                 }
 739
 740                 return taxon;
 741         }
 742         catch(exception& e) {
 743                 m->errorOut(e, "GetLineageCommand", "removeConfidences");
 744                 exit(1);
 745         }
 746 }
 747 //**********************************************************************************************************************
 748 //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
 749 int GetLineageCommand::readAlign(){
 750         try {
 751                 string thisOutputDir = outputDir;
 752                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 753                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 754                 ofstream out;
 755                 m->openOutputFile(outputFileName, out);
 756
 757
 758                 ifstream in;
 759                 m->openInputFile(alignfile, in);
 760                 string name, junk;
 761
 762                 bool wroteSomething = false;
 763
 764                 //read column headers
 765                 for (int i = 0; i < 16; i++) {
 766                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 767                         else                    {       break;                  }
 768                 }
 769                 out << endl;
 770
 771                 while(!in.eof()){
 772
 773                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 774
 775
 776                         in >> name;                             //read from first column
 777
 778                         //if this name is in the accnos file
 779                         if (names.count(name) != 0) {
 780                                 wroteSomething = true;
 781
 782                                 out << name << '\t';
 783
 784                                 //read rest
 785                                 for (int i = 0; i < 15; i++) {
 786                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 787                                         else                    {       break;                  }
 788                                 }
 789                                 out << endl;
 790
 791                         }else {//still read just don't do anything with it
 792                                 //read rest
 793                                 for (int i = 0; i < 15; i++) {
 794                                         if (!in.eof())  {       in >> junk;             }
 795                                         else                    {       break;                  }
 796                                 }
 797                         }
 798
 799                         m->gobble(in);
 800                 }
 801                 in.close();
 802                 out.close();
 803
 804                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 805                 outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
 806
 807                 return 0;
 808
 809         }
 810         catch(exception& e) {
 811                 m->errorOut(e, "GetLineageCommand", "readAlign");
 812                 exit(1);
 813         }
 814 }
 815 //**********************************************************************************************************************
 816
 817
 818
 819