classify.cpp

   1 /*
   2  *  classify.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/3/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "classify.h"
  11 #include "sequence.hpp"
  12 #include "kmerdb.hpp"
  13 #include "suffixdb.hpp"
  14 #include "blastdb.hpp"
  15
  16 /**************************************************************************************************/
  17
  18 Classify::Classify(string tfile, string tempFile, string method, int kmerSize, int gapOpen, int gapExtend, int match, int misMatch) : taxFile(tfile), templateFile(tempFile) {
  19         try {
  20
  21                 readTaxonomy(taxFile);
  22
  23                 int numSeqs = 0;
  24                 //need to know number of template seqs for suffixdb
  25                 if (method == "suffix") {
  26                         ifstream inFASTA;
  27                         openInputFile(tempFile, inFASTA);
  28                         numSeqs = count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
  29                         inFASTA.close();
  30                 }
  31
  32                 mothurOut("Generating search database...    "); cout.flush();
  33
  34                 bool needToGenerate = true;
  35                 string kmerDBName;
  36                 if(method == "kmer")                    {
  37                         database = new KmerDB(tempFile, kmerSize);
  38
  39                         kmerDBName = tempFile.substr(0,tempFile.find_last_of(".")+1) + char('0'+ kmerSize) + "mer";
  40                         ifstream kmerFileTest(kmerDBName.c_str());
  41                         if(kmerFileTest){       needToGenerate = false;         }
  42                 }
  43                 else if(method == "suffix")             {       database = new SuffixDB(numSeqs);                                                               }
  44                 else if(method == "blast")              {       database = new BlastDB(gapOpen, gapExtend, match, misMatch);    }
  45                 else {
  46                         mothurOut(method + " is not a valid search option. I will run the command using kmer, ksize=8.");
  47                         mothurOutEndLine();
  48                         database = new KmerDB(tempFile, 8);
  49                 }
  50
  51                 if (needToGenerate) {
  52                         ifstream fastaFile;
  53                         openInputFile(tempFile, fastaFile);
  54
  55                         while (!fastaFile.eof()) {
  56                                 Sequence temp(fastaFile);
  57                                 gobble(fastaFile);
  58
  59                                 names.push_back(temp.getName());
  60                                 database->addSequence(temp);
  61                         }
  62                         fastaFile.close();
  63
  64                         database->generateDB();
  65
  66                 }else if ((method == "kmer") && (!needToGenerate)) {
  67                         ifstream kmerFileTest(kmerDBName.c_str());
  68                         database->readKmerDB(kmerFileTest);
  69
  70                         ifstream fastaFile;
  71                         openInputFile(tempFile, fastaFile);
  72
  73                         while (!fastaFile.eof()) {
  74                                 Sequence temp(fastaFile);
  75                                 gobble(fastaFile);
  76
  77                                 names.push_back(temp.getName());
  78                         }
  79                         fastaFile.close();
  80                 }
  81
  82                 database->setNumSeqs(names.size());
  83
  84                 mothurOut("DONE."); mothurOutEndLine();
  85
  86         }
  87         catch(exception& e) {
  88                 errorOut(e, "Classify", "Classify");
  89                 exit(1);
  90         }
  91 }
  92 /**************************************************************************************************/
  93
  94 void Classify::readTaxonomy(string file) {
  95         try {
  96
  97                 ifstream inTax;
  98                 openInputFile(file, inTax);
  99
 100                 mothurOutEndLine();
 101                 mothurOut("Reading in the " + file + " taxonomy...\t"); cout.flush();
 102
 103                 string name, taxInfo;
 104                 //read template seqs and save
 105                 while (!inTax.eof()) {
 106                         inTax >> name >> taxInfo;
 107
 108                         taxonomy[name] = taxInfo;
 109
 110                         gobble(inTax);
 111                 }
 112                 inTax.close();
 113
 114                 mothurOut("DONE.");
 115                 mothurOutEndLine();     cout.flush();
 116
 117         }
 118         catch(exception& e) {
 119                 errorOut(e, "Classify", "readTaxonomy");
 120                 exit(1);
 121         }
 122 }
 123 /**************************************************************************************************/
 124