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; calledHelp = true;
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["taxonomy"] = tempOutNames;
  34                 outputTypes["name"] = tempOutNames;
  35                 outputTypes["group"] = tempOutNames;
  36                 outputTypes["alignreport"] = tempOutNames;
  37                 outputTypes["list"] = tempOutNames;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 vector<string> GetLineageCommand::getRequiredParameters(){
  46         try {
  47                 string Array[] =  {"taxonomy","taxon"};
  48                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  49                 return myArray;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "GetLineageCommand", "getRequiredParameters");
  53                 exit(1);
  54         }
  55 }
  56 //**********************************************************************************************************************
  57 vector<string> GetLineageCommand::getRequiredFiles(){
  58         try {
  59                 vector<string> myArray;
  60                 return myArray;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "GetLineageCommand", "getRequiredFiles");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 GetLineageCommand::GetLineageCommand(string option)  {
  69         try {
  70                 abort = false; calledHelp = false;
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; calledHelp = true; }
  74
  75                 else {
  76                         //valid paramters for this command
  77                         string Array[] =  {"fasta","name", "group", "alignreport", "taxon", "dups", "list","taxonomy","outputdir","inputdir"};
  78                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82
  83                         ValidParameters validParameter;
  84                         map<string,string>::iterator it;
  85
  86                         //check to make sure all parameters are valid for command
  87                         for (it = parameters.begin(); it != parameters.end(); it++) {
  88                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  89                         }
  90
  91                         //initialize outputTypes
  92                         vector<string> tempOutNames;
  93                         outputTypes["fasta"] = tempOutNames;
  94                         outputTypes["taxonomy"] = tempOutNames;
  95                         outputTypes["name"] = tempOutNames;
  96                         outputTypes["group"] = tempOutNames;
  97                         outputTypes["alignreport"] = tempOutNames;
  98                         outputTypes["list"] = tempOutNames;
  99
 100                         //if the user changes the output directory command factory will send this info to us in the output parameter
 101                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 102
 103                         //if the user changes the input directory command factory will send this info to us in the output parameter
 104                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 105                         if (inputDir == "not found"){   inputDir = "";          }
 106                         else {
 107                                 string path;
 108                                 it = parameters.find("alignreport");
 109                                 //user has given a template file
 110                                 if(it != parameters.end()){
 111                                         path = m->hasPath(it->second);
 112                                         //if the user has not given a path then, add inputdir. else leave path alone.
 113                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 114                                 }
 115
 116                                 it = parameters.find("fasta");
 117                                 //user has given a template file
 118                                 if(it != parameters.end()){
 119                                         path = m->hasPath(it->second);
 120                                         //if the user has not given a path then, add inputdir. else leave path alone.
 121                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 122                                 }
 123
 124                                 it = parameters.find("list");
 125                                 //user has given a template file
 126                                 if(it != parameters.end()){
 127                                         path = m->hasPath(it->second);
 128                                         //if the user has not given a path then, add inputdir. else leave path alone.
 129                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 130                                 }
 131
 132                                 it = parameters.find("name");
 133                                 //user has given a template file
 134                                 if(it != parameters.end()){
 135                                         path = m->hasPath(it->second);
 136                                         //if the user has not given a path then, add inputdir. else leave path alone.
 137                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 138                                 }
 139
 140                                 it = parameters.find("group");
 141                                 //user has given a template file
 142                                 if(it != parameters.end()){
 143                                         path = m->hasPath(it->second);
 144                                         //if the user has not given a path then, add inputdir. else leave path alone.
 145                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 146                                 }
 147
 148                                 it = parameters.find("taxonomy");
 149                                 //user has given a template file
 150                                 if(it != parameters.end()){
 151                                         path = m->hasPath(it->second);
 152                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 154                                 }
 155                         }
 156
 157
 158                         //check for required parameters
 159                         fastafile = validParameter.validFile(parameters, "fasta", true);
 160                         if (fastafile == "not open") { abort = true; }
 161                         else if (fastafile == "not found") {  fastafile = "";  }
 162
 163                         namefile = validParameter.validFile(parameters, "name", true);
 164                         if (namefile == "not open") { abort = true; }
 165                         else if (namefile == "not found") {  namefile = "";  }
 166
 167                         groupfile = validParameter.validFile(parameters, "group", true);
 168                         if (groupfile == "not open") { abort = true; }
 169                         else if (groupfile == "not found") {  groupfile = "";  }
 170
 171                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 172                         if (alignfile == "not open") { abort = true; }
 173                         else if (alignfile == "not found") {  alignfile = "";  }
 174
 175                         listfile = validParameter.validFile(parameters, "list", true);
 176                         if (listfile == "not open") { abort = true; }
 177                         else if (listfile == "not found") {  listfile = "";  }
 178
 179                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 180                         if (taxfile == "not open") { abort = true; }
 181                         else if (taxfile == "not found") {  taxfile = ""; m->mothurOut("The taxonomy parameter is required for the get.lineage command."); m->mothurOutEndLine();  abort = true; }
 182
 183                         string usedDups = "true";
 184                         string temp = validParameter.validFile(parameters, "dups", false);
 185                         if (temp == "not found") {
 186                                 if (namefile != "") {  temp = "true";                                   }
 187                                 else                            {  temp = "false"; usedDups = "";       }
 188                         }
 189                         dups = m->isTrue(temp);
 190
 191                         taxons = validParameter.validFile(parameters, "taxon", false);
 192                         if (taxons == "not found") { taxons = "";  m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine();  abort = true;  }
 193                         else {
 194                                 //rip off quotes
 195                                 if (taxons[0] == '\'') {  taxons = taxons.substr(1); }
 196                                 if (taxons[(taxons.length()-1)] == '\'') {  taxons = taxons.substr(0, (taxons.length()-1)); }
 197                         }
 198
 199
 200                         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; }
 201
 202                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 203
 204                 }
 205
 206         }
 207         catch(exception& e) {
 208                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
 209                 exit(1);
 210         }
 211 }
 212 //**********************************************************************************************************************
 213
 214 void GetLineageCommand::help(){
 215         try {
 216                 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");
 217                 m->mothurOut("It outputs a file containing only the sequences from the taxonomy file that are from the taxon requested.\n");
 218                 m->mothurOut("The get.lineage command parameters are taxon, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide taxonomy and taxon.\n");
 219                 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");
 220                 m->mothurOut("The taxon parameter allows you to select the taxons you would like to get.\n");
 221                 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");
 222                 m->mothurOut("If they belong to the taxonomy and have confidences below those you provide the sequence will not be selected.\n");
 223                 m->mothurOut("The get.lineage command should be in the following format: get.lineage(taxonomy=yourTaxonomyFile, taxon=yourTaxons).\n");
 224                 m->mothurOut("Example get.lineage(taxonomy=amazon.silva.taxonomy, taxon=Bacteria;Firmicutes;Bacilli;Lactobacillales;).\n");
 225                 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");
 226                 m->mothurOut("Example get.lineage(taxonomy=amazon.silva.taxonomy, taxon='Bacteria;Firmicutes;Bacilli;Lactobacillales;').\n");
 227                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 228         }
 229         catch(exception& e) {
 230                 m->errorOut(e, "GetLineageCommand", "help");
 231                 exit(1);
 232         }
 233 }
 234
 235 //**********************************************************************************************************************
 236
 237 int GetLineageCommand::execute(){
 238         try {
 239
 240                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 241
 242                 if (m->control_pressed) { return 0; }
 243
 244                 //read through the correct file and output lines you want to keep
 245                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to get
 246                 if (namefile != "")                     {               readName();             }
 247                 if (fastafile != "")            {               readFasta();    }
 248                 if (groupfile != "")            {               readGroup();    }
 249                 if (alignfile != "")            {               readAlign();    }
 250                 if (listfile != "")                     {               readList();             }
 251
 252
 253                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str());  } return 0; }
 254
 255                 if (outputNames.size() != 0) {
 256                         m->mothurOutEndLine();
 257                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 258                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 259                         m->mothurOutEndLine();
 260
 261                         //set fasta file as new current fastafile
 262                         string current = "";
 263                         itTypes = outputTypes.find("fasta");
 264                         if (itTypes != outputTypes.end()) {
 265                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 266                         }
 267
 268                         itTypes = outputTypes.find("name");
 269                         if (itTypes != outputTypes.end()) {
 270                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 271                         }
 272
 273                         itTypes = outputTypes.find("group");
 274                         if (itTypes != outputTypes.end()) {
 275                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 276                         }
 277
 278                         itTypes = outputTypes.find("list");
 279                         if (itTypes != outputTypes.end()) {
 280                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 281                         }
 282
 283                         itTypes = outputTypes.find("taxonomy");
 284                         if (itTypes != outputTypes.end()) {
 285                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 286                         }
 287                 }
 288
 289                 return 0;
 290         }
 291
 292         catch(exception& e) {
 293                 m->errorOut(e, "GetLineageCommand", "execute");
 294                 exit(1);
 295         }
 296 }
 297
 298 //**********************************************************************************************************************
 299 int GetLineageCommand::readFasta(){
 300         try {
 301                 string thisOutputDir = outputDir;
 302                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 303                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 304                 ofstream out;
 305                 m->openOutputFile(outputFileName, out);
 306
 307
 308                 ifstream in;
 309                 m->openInputFile(fastafile, in);
 310                 string name;
 311
 312                 bool wroteSomething = false;
 313
 314                 while(!in.eof()){
 315
 316                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 317
 318                         Sequence currSeq(in);
 319                         name = currSeq.getName();
 320
 321                         if (name != "") {
 322                                 //if this name is in the accnos file
 323                                 if (names.count(name) != 0) {
 324                                         wroteSomething = true;
 325
 326                                         currSeq.printSequence(out);
 327                                 }
 328                         }
 329                         m->gobble(in);
 330                 }
 331                 in.close();
 332                 out.close();
 333
 334                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 335                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 336
 337                 return 0;
 338
 339         }
 340         catch(exception& e) {
 341                 m->errorOut(e, "GetLineageCommand", "readFasta");
 342                 exit(1);
 343         }
 344 }
 345 //**********************************************************************************************************************
 346 int GetLineageCommand::readList(){
 347         try {
 348                 string thisOutputDir = outputDir;
 349                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 350                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 351                 ofstream out;
 352                 m->openOutputFile(outputFileName, out);
 353
 354                 ifstream in;
 355                 m->openInputFile(listfile, in);
 356
 357                 bool wroteSomething = false;
 358
 359                 while(!in.eof()){
 360
 361                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 362
 363                         //read in list vector
 364                         ListVector list(in);
 365
 366                         //make a new list vector
 367                         ListVector newList;
 368                         newList.setLabel(list.getLabel());
 369
 370                         //for each bin
 371                         for (int i = 0; i < list.getNumBins(); i++) {
 372
 373                                 //parse out names that are in accnos file
 374                                 string binnames = list.get(i);
 375
 376                                 string newNames = "";
 377                                 while (binnames.find_first_of(',') != -1) {
 378                                         string name = binnames.substr(0,binnames.find_first_of(','));
 379                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 380
 381                                         //if that name is in the .accnos file, add it
 382                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 383                                 }
 384
 385                                 //get last name
 386                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 387
 388                                 //if there are names in this bin add to new list
 389                                 if (newNames != "") {
 390                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 391                                         newList.push_back(newNames);
 392                                 }
 393                         }
 394
 395                         //print new listvector
 396                         if (newList.getNumBins() != 0) {
 397                                 wroteSomething = true;
 398                                 newList.print(out);
 399                         }
 400
 401                         m->gobble(in);
 402                 }
 403                 in.close();
 404                 out.close();
 405
 406                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 407                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 408
 409                 return 0;
 410
 411         }
 412         catch(exception& e) {
 413                 m->errorOut(e, "GetLineageCommand", "readList");
 414                 exit(1);
 415         }
 416 }
 417 //**********************************************************************************************************************
 418 int GetLineageCommand::readName(){
 419         try {
 420                 string thisOutputDir = outputDir;
 421                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 422                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 423                 ofstream out;
 424                 m->openOutputFile(outputFileName, out);
 425
 426
 427                 ifstream in;
 428                 m->openInputFile(namefile, in);
 429                 string name, firstCol, secondCol;
 430
 431                 bool wroteSomething = false;
 432
 433
 434                 while(!in.eof()){
 435
 436                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 437
 438                         in >> firstCol;
 439                         in >> secondCol;
 440
 441                         string hold = "";
 442                         if (dups) { hold = secondCol; }
 443
 444                         vector<string> parsedNames;
 445                         //parse second column saving each name
 446                         while (secondCol.find_first_of(',') != -1) {
 447                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 448                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 449                                 parsedNames.push_back(name);
 450                         }
 451
 452                         //get name after last ,
 453                         parsedNames.push_back(secondCol);
 454
 455                         vector<string> validSecond;
 456                         for (int i = 0; i < parsedNames.size(); i++) {
 457                                 if (names.count(parsedNames[i]) != 0) {
 458                                         validSecond.push_back(parsedNames[i]);
 459                                 }
 460                         }
 461
 462                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 463                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 464                                 out << firstCol << '\t' << hold << endl;
 465                                 wroteSomething = true;
 466                         }else {
 467                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 468                                 if (names.count(firstCol) != 0) {
 469
 470                                         wroteSomething = true;
 471
 472                                         out << firstCol << '\t';
 473
 474                                         //you know you have at least one valid second since first column is valid
 475                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 476                                         out << validSecond[validSecond.size()-1] << endl;
 477
 478
 479                                 //make first name in set you come to first column and then add the remaining names to second column
 480                                 }else {
 481                                         //you want part of this row
 482                                         if (validSecond.size() != 0) {
 483
 484                                                 wroteSomething = true;
 485
 486                                                 out << validSecond[0] << '\t';
 487
 488                                                 //you know you have at least one valid second since first column is valid
 489                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 490                                                 out << validSecond[validSecond.size()-1] << endl;
 491                                         }
 492                                 }
 493                         }
 494                         m->gobble(in);
 495                 }
 496                 in.close();
 497                 out.close();
 498
 499                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 500                 outputNames.push_back(outputFileName);  outputTypes["name"].push_back(outputFileName);
 501
 502                 return 0;
 503
 504         }
 505         catch(exception& e) {
 506                 m->errorOut(e, "GetLineageCommand", "readName");
 507                 exit(1);
 508         }
 509 }
 510
 511 //**********************************************************************************************************************
 512 int GetLineageCommand::readGroup(){
 513         try {
 514                 string thisOutputDir = outputDir;
 515                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 516                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 517                 ofstream out;
 518                 m->openOutputFile(outputFileName, out);
 519
 520
 521                 ifstream in;
 522                 m->openInputFile(groupfile, in);
 523                 string name, group;
 524
 525                 bool wroteSomething = false;
 526
 527                 while(!in.eof()){
 528
 529                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 530
 531
 532                         in >> name;                             //read from first column
 533                         in >> group;                    //read from second column
 534
 535                         //if this name is in the accnos file
 536                         if (names.count(name) != 0) {
 537                                 wroteSomething = true;
 538
 539                                 out << name << '\t' << group << endl;
 540                         }
 541
 542                         m->gobble(in);
 543                 }
 544                 in.close();
 545                 out.close();
 546
 547                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 548                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 549
 550                 return 0;
 551
 552         }
 553         catch(exception& e) {
 554                 m->errorOut(e, "GetLineageCommand", "readGroup");
 555                 exit(1);
 556         }
 557 }
 558 //**********************************************************************************************************************
 559 int GetLineageCommand::readTax(){
 560         try {
 561                 string thisOutputDir = outputDir;
 562                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 563                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 564                 ofstream out;
 565                 m->openOutputFile(outputFileName, out);
 566
 567                 ifstream in;
 568                 m->openInputFile(taxfile, in);
 569                 string name, tax;
 570
 571                 //bool wroteSomething = false;
 572
 573                 bool taxonsHasConfidence = false;
 574                 vector< map<string, float> > searchTaxons;
 575                 string noConfidenceTaxons = taxons;
 576                 int hasConPos = taxons.find_first_of('(');
 577                 if (hasConPos != string::npos) {
 578                         taxonsHasConfidence = true;
 579                         searchTaxons = getTaxons(taxons);
 580                         noConfidenceTaxons = removeConfidences(taxons);
 581                 }
 582
 583
 584                 while(!in.eof()){
 585
 586                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 587
 588                         in >> name;                             //read from first column
 589                         in >> tax;                      //read from second column
 590
 591                         string newtax = tax;
 592
 593
 594                         //if the users file contains confidence scores we want to ignore them when searching for the taxons, unless the taxon has them
 595                         if (!taxonsHasConfidence) {
 596                                 int hasConfidences = tax.find_first_of('(');
 597                                 if (hasConfidences != string::npos) {
 598                                         newtax = removeConfidences(tax);
 599                                 }
 600
 601                                 int pos = newtax.find(taxons);
 602
 603                                 if (pos != string::npos) { //this sequence contains the taxon the user wants
 604                                         names.insert(name);
 605                                         out << name << '\t' << tax << endl;
 606                                 }
 607
 608                         }else{//if taxons has them and you don't them remove taxons
 609                                 int hasConfidences = tax.find_first_of('(');
 610                                 if (hasConfidences == string::npos) {
 611
 612                                         int pos = newtax.find(noConfidenceTaxons);
 613
 614                                         if (pos != string::npos) { //this sequence contains the taxon the user wants
 615                                                 names.insert(name);
 616                                                 out << name << '\t' << tax << endl;
 617                                         }
 618                                 }else { //both have confidences so we want to make sure the users confidences are greater then or equal to the taxons
 619                                         //first remove confidences from both and see if the taxonomy exists
 620
 621                                         string noNewTax = tax;
 622                                         int hasConfidences = tax.find_first_of('(');
 623                                         if (hasConfidences != string::npos) {
 624                                                 noNewTax = removeConfidences(tax);
 625                                         }
 626
 627                                         int pos = noNewTax.find(noConfidenceTaxons);
 628
 629                                         if (pos != string::npos) { //if yes, then are the confidences okay
 630
 631                                                 bool good = true;
 632                                                 vector< map<string, float> > usersTaxon = getTaxons(newtax);
 633
 634                                                 //the usersTaxon is most likely longer than the searchTaxons, and searchTaxon[0] may relate to userTaxon[4]
 635                                                 //we want to "line them up", so we will find the the index where the searchstring starts
 636                                                 int index = 0;
 637                                                 for (int i = 0; i < usersTaxon.size(); i++) {
 638
 639                                                         if (usersTaxon[i].begin()->first == searchTaxons[0].begin()->first) {
 640                                                                 index = i;
 641                                                                 int spot = 0;
 642                                                                 bool goodspot = true;
 643                                                                 //is this really the start, or are we dealing with a taxon of the same name?
 644                                                                 while ((spot < searchTaxons.size()) && ((i+spot) < usersTaxon.size())) {
 645                                                                         if (usersTaxon[i+spot].begin()->first != searchTaxons[spot].begin()->first) { goodspot = false; break; }
 646                                                                         else { spot++; }
 647                                                                 }
 648
 649                                                                 if (goodspot) { break; }
 650                                                         }
 651                                                 }
 652
 653                                                 for (int i = 0; i < searchTaxons.size(); i++) {
 654
 655                                                         if ((i+index) < usersTaxon.size()) { //just in case, should never be false
 656                                                                 if (usersTaxon[i+index].begin()->second < searchTaxons[i].begin()->second) { //is the users cutoff less than the search taxons
 657                                                                         good = false;
 658                                                                         break;
 659                                                                 }
 660                                                         }else {
 661                                                                 good = false;
 662                                                                 break;
 663                                                         }
 664                                                 }
 665
 666                                                 //passed the test so add you
 667                                                 if (good) {
 668                                                         names.insert(name);
 669                                                         out << name << '\t' << tax << endl;
 670                                                 }
 671                                         }
 672                                 }
 673                         }
 674
 675
 676
 677                         m->gobble(in);
 678                 }
 679                 in.close();
 680                 out.close();
 681
 682                 if (names.size() == 0) { m->mothurOut("Your taxonomy file does not contain any sequences from " + taxons + "."); m->mothurOutEndLine();  }
 683                 outputNames.push_back(outputFileName); outputTypes["taxonomy"].push_back(outputFileName);
 684
 685                 return 0;
 686
 687         }
 688         catch(exception& e) {
 689                 m->errorOut(e, "GetLineageCommand", "readTax");
 690                 exit(1);
 691         }
 692 }
 693 /**************************************************************************************************/
 694 vector< map<string, float> > GetLineageCommand::getTaxons(string tax) {
 695         try {
 696
 697                 vector< map<string, float> > t;
 698                 string taxon = "";
 699                 int taxLength = tax.length();
 700                 for(int i=0;i<taxLength;i++){
 701                         if(tax[i] == ';'){
 702
 703                                 int openParen = taxon.find_first_of('(');
 704                                 int closeParen = taxon.find_last_of(')');
 705
 706                                 string newtaxon, confidence;
 707                                 if ((openParen != string::npos) && (closeParen != string::npos)) {
 708                                         newtaxon = taxon.substr(0, openParen); //rip off confidence
 709                                         confidence = taxon.substr((openParen+1), (closeParen-openParen-1));
 710                                 }else{
 711                                         newtaxon = taxon;
 712                                         confidence = "0";
 713                                 }
 714                                 float con = 0;
 715                                 convert(confidence, con);
 716
 717                                 map<string, float> temp;
 718                                 temp[newtaxon] = con;
 719                                 t.push_back(temp);
 720
 721                                 taxon = "";
 722                         }
 723                         else{
 724                                 taxon += tax[i];
 725                         }
 726                 }
 727
 728                 return t;
 729         }
 730         catch(exception& e) {
 731                 m->errorOut(e, "GetLineageCommand", "getTaxons");
 732                 exit(1);
 733         }
 734 }
 735 /**************************************************************************************************/
 736 string GetLineageCommand::removeConfidences(string tax) {
 737         try {
 738
 739                 string taxon = "";
 740                 int taxLength = tax.length();
 741                 for(int i=0;i<taxLength;i++){
 742                         if(tax[i] == ';'){
 743                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 744                                 taxon += ";";
 745                         }
 746                         else{
 747                                 taxon += tax[i];
 748                         }
 749                 }
 750
 751                 return taxon;
 752         }
 753         catch(exception& e) {
 754                 m->errorOut(e, "GetLineageCommand", "removeConfidences");
 755                 exit(1);
 756         }
 757 }
 758 //**********************************************************************************************************************
 759 //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
 760 int GetLineageCommand::readAlign(){
 761         try {
 762                 string thisOutputDir = outputDir;
 763                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 764                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 765                 ofstream out;
 766                 m->openOutputFile(outputFileName, out);
 767
 768
 769                 ifstream in;
 770                 m->openInputFile(alignfile, in);
 771                 string name, junk;
 772
 773                 bool wroteSomething = false;
 774
 775                 //read column headers
 776                 for (int i = 0; i < 16; i++) {
 777                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 778                         else                    {       break;                  }
 779                 }
 780                 out << endl;
 781
 782                 while(!in.eof()){
 783
 784                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 785
 786
 787                         in >> name;                             //read from first column
 788
 789                         //if this name is in the accnos file
 790                         if (names.count(name) != 0) {
 791                                 wroteSomething = true;
 792
 793                                 out << name << '\t';
 794
 795                                 //read rest
 796                                 for (int i = 0; i < 15; i++) {
 797                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 798                                         else                    {       break;                  }
 799                                 }
 800                                 out << endl;
 801
 802                         }else {//still read just don't do anything with it
 803                                 //read rest
 804                                 for (int i = 0; i < 15; i++) {
 805                                         if (!in.eof())  {       in >> junk;             }
 806                                         else                    {       break;                  }
 807                                 }
 808                         }
 809
 810                         m->gobble(in);
 811                 }
 812                 in.close();
 813                 out.close();
 814
 815                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 816                 outputNames.push_back(outputFileName); outputTypes["alignreport"].push_back(outputFileName);
 817
 818                 return 0;
 819
 820         }
 821         catch(exception& e) {
 822                 m->errorOut(e, "GetLineageCommand", "readAlign");
 823                 exit(1);
 824         }
 825 }
 826 //**********************************************************************************************************************
 827
 828
 829
 830