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