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 //**********************************************************************************************************************
  16 vector<string> GetLineageCommand::getValidParameters(){
  17         try {
  18                 string Array[] =  {"fasta","name", "group", "alignreport", "taxon", "dups", "list","taxonomy","outputdir","inputdir"};
  19                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  20                 return myArray;
  21         }
  22         catch(exception& e) {
  23                 m->errorOut(e, "GetLineageCommand", "getValidParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 GetLineageCommand::GetLineageCommand(){
  29         try {
  30                 abort = true;
  31                 //initialize outputTypes
  32                 vector<string> tempOutNames;
  33                 outputTypes["fasta"] = tempOutNames;
  34                 outputTypes["taxonomy"] = tempOutNames;
  35                 outputTypes["name"] = tempOutNames;
  36                 outputTypes["group"] = tempOutNames;
  37                 outputTypes["alignreport"] = tempOutNames;
  38                 outputTypes["list"] = tempOutNames;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 vector<string> GetLineageCommand::getRequiredParameters(){
  47         try {
  48                 string Array[] =  {"taxonomy","taxon"};
  49                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "GetLineageCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> GetLineageCommand::getRequiredFiles(){
  59         try {
  60                 vector<string> myArray;
  61                 return myArray;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "GetLineageCommand", "getRequiredFiles");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 GetLineageCommand::GetLineageCommand(string option)  {
  70         try {
  71                 abort = false;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; }
  75
  76                 else {
  77                         //valid paramters for this command
  78                         string Array[] =  {"fasta","name", "group", "alignreport", "taxon", "dups", "list","taxonomy","outputdir","inputdir"};
  79                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83
  84                         ValidParameters validParameter;
  85                         map<string,string>::iterator it;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95                         outputTypes["taxonomy"] = tempOutNames;
  96                         outputTypes["name"] = tempOutNames;
  97                         outputTypes["group"] = tempOutNames;
  98                         outputTypes["alignreport"] = tempOutNames;
  99                         outputTypes["list"] = tempOutNames;
 100
 101                         //if the user changes the output directory command factory will send this info to us in the output parameter
 102                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 103
 104                         //if the user changes the input directory command factory will send this info to us in the output parameter
 105                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 106                         if (inputDir == "not found"){   inputDir = "";          }
 107                         else {
 108                                 string path;
 109                                 it = parameters.find("alignreport");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 115                                 }
 116
 117                                 it = parameters.find("fasta");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 123                                 }
 124
 125                                 it = parameters.find("list");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 131                                 }
 132
 133                                 it = parameters.find("name");
 134                                 //user has given a template file
 135                                 if(it != parameters.end()){
 136                                         path = m->hasPath(it->second);
 137                                         //if the user has not given a path then, add inputdir. else leave path alone.
 138                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 139                                 }
 140
 141                                 it = parameters.find("group");
 142                                 //user has given a template file
 143                                 if(it != parameters.end()){
 144                                         path = m->hasPath(it->second);
 145                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 147                                 }
 148
 149                                 it = parameters.find("taxonomy");
 150                                 //user has given a template file
 151                                 if(it != parameters.end()){
 152                                         path = m->hasPath(it->second);
 153                                         //if the user has not given a path then, add inputdir. else leave path alone.
 154                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 155                                 }
 156                         }
 157
 158
 159                         //check for required parameters
 160                         fastafile = validParameter.validFile(parameters, "fasta", true);
 161                         if (fastafile == "not open") { abort = true; }
 162                         else if (fastafile == "not found") {  fastafile = "";  }
 163
 164                         namefile = validParameter.validFile(parameters, "name", true);
 165                         if (namefile == "not open") { abort = true; }
 166                         else if (namefile == "not found") {  namefile = "";  }
 167
 168                         groupfile = validParameter.validFile(parameters, "group", true);
 169                         if (groupfile == "not open") { abort = true; }
 170                         else if (groupfile == "not found") {  groupfile = "";  }
 171
 172                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 173                         if (alignfile == "not open") { abort = true; }
 174                         else if (alignfile == "not found") {  alignfile = "";  }
 175
 176                         listfile = validParameter.validFile(parameters, "list", true);
 177                         if (listfile == "not open") { abort = true; }
 178                         else if (listfile == "not found") {  listfile = "";  }
 179
 180                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 181                         if (taxfile == "not open") { abort = true; }
 182                         else if (taxfile == "not found") {  taxfile = ""; m->mothurOut("The taxonomy parameter is required for the get.lineage command."); m->mothurOutEndLine();  abort = true; }
 183
 184                         string usedDups = "true";
 185                         string temp = validParameter.validFile(parameters, "dups", false);
 186                         if (temp == "not found") {
 187                                 if (namefile != "") {  temp = "true";                                   }
 188                                 else                            {  temp = "false"; usedDups = "";       }
 189                         }
 190                         dups = m->isTrue(temp);
 191
 192                         taxons = validParameter.validFile(parameters, "taxon", false);
 193                         if (taxons == "not found") { taxons = "";  m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine();  abort = true;  }
 194                         else {
 195                                 //rip off quotes
 196                                 if (taxons[0] == '\'') {  taxons = taxons.substr(1); }
 197                                 if (taxons[(taxons.length()-1)] == '\'') {  taxons = taxons.substr(0, (taxons.length()-1)); }
 198                         }
 199
 200
 201                         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; }
 202
 203                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 204
 205                 }
 206
 207         }
 208         catch(exception& e) {
 209                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
 210                 exit(1);
 211         }
 212 }
 213 //**********************************************************************************************************************
 214
 215 void GetLineageCommand::help(){
 216         try {
 217                 m->mothurOut("The get.lineage command reads a taxonomy file and any of the following file types: fasta, name, group, list or alignreport file.\n");
 218                 m->mothurOut("It outputs a file containing only the sequences from the taxonomy file that are from the taxon requested.\n");
 219                 m->mothurOut("The get.lineage command parameters are taxon, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide taxonomy and taxon.\n");
 220                 m->mothurOut("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");
 221                 m->mothurOut("The taxon parameter allows you to select the taxons you would like to get.\n");
 222                 m->mothurOut("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");
 223                 m->mothurOut("If they belong to the taxonomy and have confidences below those you provide the sequence will not be selected.\n");
 224                 m->mothurOut("The get.lineage command should be in the following format: get.lineage(taxonomy=yourTaxonomyFile, taxon=yourTaxons).\n");
 225                 m->mothurOut("Example get.lineage(taxonomy=amazon.silva.taxonomy, taxon=Bacteria;Firmicutes;Bacilli;Lactobacillales;).\n");
 226                 m->mothurOut("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");
 227                 m->mothurOut("Example get.lineage(taxonomy=amazon.silva.taxonomy, taxon='Bacteria;Firmicutes;Bacilli;Lactobacillales;').\n");
 228                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 229         }
 230         catch(exception& e) {
 231                 m->errorOut(e, "GetLineageCommand", "help");
 232                 exit(1);
 233         }
 234 }
 235
 236 //**********************************************************************************************************************
 237
 238 int GetLineageCommand::execute(){
 239         try {
 240
 241                 if (abort == true) { return 0; }
 242
 243                 if (m->control_pressed) { return 0; }
 244
 245                 //read through the correct file and output lines you want to keep
 246                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to get
 247                 if (namefile != "")                     {               readName();             }
 248                 if (fastafile != "")            {               readFasta();    }
 249                 if (groupfile != "")            {               readGroup();    }
 250                 if (alignfile != "")            {               readAlign();    }
 251                 if (listfile != "")                     {               readList();             }
 252
 253
 254                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str());  } return 0; }
 255
 256                 if (outputNames.size() != 0) {
 257                         m->mothurOutEndLine();
 258                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 259                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 260                         m->mothurOutEndLine();
 261                 }
 262
 263                 return 0;
 264         }
 265
 266         catch(exception& e) {
 267                 m->errorOut(e, "GetLineageCommand", "execute");
 268                 exit(1);
 269         }
 270 }
 271
 272 //**********************************************************************************************************************
 273 int GetLineageCommand::readFasta(){
 274         try {
 275                 string thisOutputDir = outputDir;
 276                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 277                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 278                 ofstream out;
 279                 m->openOutputFile(outputFileName, out);
 280
 281
 282                 ifstream in;
 283                 m->openInputFile(fastafile, in);
 284                 string name;
 285
 286                 bool wroteSomething = false;
 287
 288                 while(!in.eof()){
 289
 290                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 291
 292                         Sequence currSeq(in);
 293                         name = currSeq.getName();
 294
 295                         if (name != "") {
 296                                 //if this name is in the accnos file
 297                                 if (names.count(name) != 0) {
 298                                         wroteSomething = true;
 299
 300                                         currSeq.printSequence(out);
 301                                 }
 302                         }
 303                         m->gobble(in);
 304                 }
 305                 in.close();
 306                 out.close();
 307
 308                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 309                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 310
 311                 return 0;
 312
 313         }
 314         catch(exception& e) {
 315                 m->errorOut(e, "GetLineageCommand", "readFasta");
 316                 exit(1);
 317         }
 318 }
 319 //**********************************************************************************************************************
 320 int GetLineageCommand::readList(){
 321         try {
 322                 string thisOutputDir = outputDir;
 323                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 324                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 325                 ofstream out;
 326                 m->openOutputFile(outputFileName, out);
 327
 328                 ifstream in;
 329                 m->openInputFile(listfile, in);
 330
 331                 bool wroteSomething = false;
 332
 333                 while(!in.eof()){
 334
 335                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 336
 337                         //read in list vector
 338                         ListVector list(in);
 339
 340                         //make a new list vector
 341                         ListVector newList;
 342                         newList.setLabel(list.getLabel());
 343
 344                         //for each bin
 345                         for (int i = 0; i < list.getNumBins(); i++) {
 346
 347                                 //parse out names that are in accnos file
 348                                 string binnames = list.get(i);
 349
 350                                 string newNames = "";
 351                                 while (binnames.find_first_of(',') != -1) {
 352                                         string name = binnames.substr(0,binnames.find_first_of(','));
 353                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 354
 355                                         //if that name is in the .accnos file, add it
 356                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 357                                 }
 358
 359                                 //get last name
 360                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 361
 362                                 //if there are names in this bin add to new list
 363                                 if (newNames != "") {
 364                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 365                                         newList.push_back(newNames);
 366                                 }
 367                         }
 368
 369                         //print new listvector
 370                         if (newList.getNumBins() != 0) {
 371                                 wroteSomething = true;
 372                                 newList.print(out);
 373                         }
 374
 375                         m->gobble(in);
 376                 }
 377                 in.close();
 378                 out.close();
 379
 380                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 381                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 382
 383                 return 0;
 384
 385         }
 386         catch(exception& e) {
 387                 m->errorOut(e, "GetLineageCommand", "readList");
 388                 exit(1);
 389         }
 390 }
 391 //**********************************************************************************************************************
 392 int GetLineageCommand::readName(){
 393         try {
 394                 string thisOutputDir = outputDir;
 395                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 396                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 397                 ofstream out;
 398                 m->openOutputFile(outputFileName, out);
 399
 400
 401                 ifstream in;
 402                 m->openInputFile(namefile, in);
 403                 string name, firstCol, secondCol;
 404
 405                 bool wroteSomething = false;
 406
 407
 408                 while(!in.eof()){
 409
 410                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 411
 412                         in >> firstCol;
 413                         in >> secondCol;
 414
 415                         string hold = "";
 416                         if (dups) { hold = secondCol; }
 417
 418                         vector<string> parsedNames;
 419                         //parse second column saving each name
 420                         while (secondCol.find_first_of(',') != -1) {
 421                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 422                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 423                                 parsedNames.push_back(name);
 424                         }
 425
 426                         //get name after last ,
 427                         parsedNames.push_back(secondCol);
 428
 429                         vector<string> validSecond;
 430                         for (int i = 0; i < parsedNames.size(); i++) {
 431                                 if (names.count(parsedNames[i]) != 0) {
 432                                         validSecond.push_back(parsedNames[i]);
 433                                 }
 434                         }
 435
 436                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 437                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 438                                 out << firstCol << '\t' << hold << endl;
 439                                 wroteSomething = true;
 440                         }else {
 441                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 442                                 if (names.count(firstCol) != 0) {
 443
 444                                         wroteSomething = true;
 445
 446                                         out << firstCol << '\t';
 447
 448                                         //you know you have at least one valid second since first column is valid
 449                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 450                                         out << validSecond[validSecond.size()-1] << endl;
 451
 452
 453                                 //make first name in set you come to first column and then add the remaining names to second column
 454                                 }else {
 455                                         //you want part of this row
 456                                         if (validSecond.size() != 0) {
 457
 458                                                 wroteSomething = true;
 459
 460                                                 out << validSecond[0] << '\t';
 461
 462                                                 //you know you have at least one valid second since first column is valid
 463                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 464                                                 out << validSecond[validSecond.size()-1] << endl;
 465                                         }
 466                                 }
 467                         }
 468                         m->gobble(in);
 469                 }
 470                 in.close();
 471                 out.close();
 472
 473                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 474                 outputNames.push_back(outputFileName);  outputTypes["name"].push_back(outputFileName);
 475
 476                 return 0;
 477
 478         }
 479         catch(exception& e) {
 480                 m->errorOut(e, "GetLineageCommand", "readName");
 481                 exit(1);
 482         }
 483 }
 484
 485 //**********************************************************************************************************************
 486 int GetLineageCommand::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                 ofstream out;
 492                 m->openOutputFile(outputFileName, out);
 493
 494
 495                 ifstream in;
 496                 m->openInputFile(groupfile, in);
 497                 string name, group;
 498
 499                 bool wroteSomething = false;
 500
 501                 while(!in.eof()){
 502
 503                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 504
 505
 506                         in >> name;                             //read from first column
 507                         in >> group;                    //read from second column
 508
 509                         //if this name is in the accnos file
 510                         if (names.count(name) != 0) {
 511                                 wroteSomething = true;
 512
 513                                 out << name << '\t' << group << endl;
 514                         }
 515
 516                         m->gobble(in);
 517                 }
 518                 in.close();
 519                 out.close();
 520
 521                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 522                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 523
 524                 return 0;
 525
 526         }
 527         catch(exception& e) {
 528                 m->errorOut(e, "GetLineageCommand", "readGroup");
 529                 exit(1);
 530         }
 531 }
 532 //**********************************************************************************************************************
 533 int GetLineageCommand::readTax(){
 534         try {
 535                 string thisOutputDir = outputDir;
 536                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 537                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 538                 ofstream out;
 539                 m->openOutputFile(outputFileName, out);
 540
 541                 ifstream in;
 542                 m->openInputFile(taxfile, in);
 543                 string name, tax;
 544
 545                 //bool wroteSomething = false;
 546
 547                 bool taxonsHasConfidence = false;
 548                 vector< map<string, int> > searchTaxons;
 549                 string noConfidenceTaxons = taxons;
 550                 int hasConPos = taxons.find_first_of('(');
 551                 if (hasConPos != string::npos) {
 552                         taxonsHasConfidence = true;
 553                         searchTaxons = getTaxons(taxons);
 554                         noConfidenceTaxons = removeConfidences(taxons);
 555                 }
 556
 557
 558                 while(!in.eof()){
 559
 560                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 561
 562                         in >> name;                             //read from first column
 563                         in >> tax;                      //read from second column
 564
 565                         string newtax = tax;
 566
 567
 568                         //if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
 569                         if (!taxonsHasConfidence) {
 570                                 int hasConfidences = tax.find_first_of('(');
 571                                 if (hasConfidences != string::npos) {
 572                                         newtax = removeConfidences(tax);
 573                                 }
 574
 575                                 int pos = newtax.find(taxons);
 576
 577                                 if (pos != string::npos) { //this sequence contains the taxon the user wants
 578                                         names.insert(name);
 579                                         out << name << '\t' << tax << endl;
 580                                 }
 581
 582                         }else{//if taxons has them and you don't them remove taxons
 583                                 int hasConfidences = tax.find_first_of('(');
 584                                 if (hasConfidences == string::npos) {
 585
 586                                         int pos = newtax.find(noConfidenceTaxons);
 587
 588                                         if (pos != string::npos) { //this sequence contains the taxon the user wants
 589                                                 names.insert(name);
 590                                                 out << name << '\t' << tax << endl;
 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 good = true;
 606                                                 vector< map<string, int> > 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                                                                         good = false;
 632                                                                         break;
 633                                                                 }
 634                                                         }else {
 635                                                                 good = false;
 636                                                                 break;
 637                                                         }
 638                                                 }
 639
 640                                                 //passed the test so add you
 641                                                 if (good) {
 642                                                         names.insert(name);
 643                                                         out << name << '\t' << tax << endl;
 644                                                 }
 645                                         }
 646                                 }
 647                         }
 648
 649
 650
 651                         m->gobble(in);
 652                 }
 653                 in.close();
 654                 out.close();
 655
 656                 if (names.size() == 0) { m->mothurOut("Your taxonomy file does not contain any sequences from " + taxons + "."); m->mothurOutEndLine();  }
 657                 outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
 658
 659                 return 0;
 660
 661         }
 662         catch(exception& e) {
 663                 m->errorOut(e, "GetLineageCommand", "readTax");
 664                 exit(1);
 665         }
 666 }
 667 /**************************************************************************************************/
 668 vector< map<string, int> > GetLineageCommand::getTaxons(string tax) {
 669         try {
 670
 671                 vector< map<string, int> > t;
 672                 string taxon = "";
 673                 int taxLength = tax.length();
 674                 for(int i=0;i<taxLength;i++){
 675                         if(tax[i] == ';'){
 676
 677                                 int openParen = taxon.find_first_of('(');
 678                                 int closeParen = taxon.find_last_of(')');
 679
 680                                 string newtaxon = taxon.substr(0, openParen); //rip off confidence
 681                                 string confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
 682                                 int con = 0;
 683                                 convert(confidence, con);
 684
 685                                 map<string, int> temp;
 686                                 temp[newtaxon] = con;
 687                                 t.push_back(temp);
 688
 689                                 taxon = "";
 690                         }
 691                         else{
 692                                 taxon += tax[i];
 693                         }
 694                 }
 695
 696                 return t;
 697         }
 698         catch(exception& e) {
 699                 m->errorOut(e, "GetLineageCommand", "getTaxons");
 700                 exit(1);
 701         }
 702 }
 703 /**************************************************************************************************/
 704 string GetLineageCommand::removeConfidences(string tax) {
 705         try {
 706
 707                 string taxon = "";
 708                 int taxLength = tax.length();
 709                 for(int i=0;i<taxLength;i++){
 710                         if(tax[i] == ';'){
 711                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 712                                 taxon += ";";
 713                         }
 714                         else{
 715                                 taxon += tax[i];
 716                         }
 717                 }
 718
 719                 return taxon;
 720         }
 721         catch(exception& e) {
 722                 m->errorOut(e, "GetLineageCommand", "removeConfidences");
 723                 exit(1);
 724         }
 725 }
 726 //**********************************************************************************************************************
 727 //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
 728 int GetLineageCommand::readAlign(){
 729         try {
 730                 string thisOutputDir = outputDir;
 731                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 732                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 733                 ofstream out;
 734                 m->openOutputFile(outputFileName, out);
 735
 736
 737                 ifstream in;
 738                 m->openInputFile(alignfile, in);
 739                 string name, junk;
 740
 741                 bool wroteSomething = false;
 742
 743                 //read column headers
 744                 for (int i = 0; i < 16; i++) {
 745                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 746                         else                    {       break;                  }
 747                 }
 748                 out << endl;
 749
 750                 while(!in.eof()){
 751
 752                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 753
 754
 755                         in >> name;                             //read from first column
 756
 757                         //if this name is in the accnos file
 758                         if (names.count(name) != 0) {
 759                                 wroteSomething = true;
 760
 761                                 out << name << '\t';
 762
 763                                 //read rest
 764                                 for (int i = 0; i < 15; i++) {
 765                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 766                                         else                    {       break;                  }
 767                                 }
 768                                 out << endl;
 769
 770                         }else {//still read just don't do anything with it
 771                                 //read rest
 772                                 for (int i = 0; i < 15; i++) {
 773                                         if (!in.eof())  {       in >> junk;             }
 774                                         else                    {       break;                  }
 775                                 }
 776                         }
 777
 778                         m->gobble(in);
 779                 }
 780                 in.close();
 781                 out.close();
 782
 783                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 784                 outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
 785
 786                 return 0;
 787
 788         }
 789         catch(exception& e) {
 790                 m->errorOut(e, "GetLineageCommand", "readAlign");
 791                 exit(1);
 792         }
 793 }
 794 //**********************************************************************************************************************
 795
 796
 797
 798