gnGBKSource.h Source File

00001 
00002 // File:            gnGBKSource.h
00003 // Purpose:         Implements gnBaseSource for GenBank files
00004 // Description:     
00005 // Changes:        
00006 // Version:         libGenome 0.1.0 
00007 // Author:          Aaron Darling 
00008 // Last Edited:     April 15, 2001, 10:34:50pm 
00009 // Modified by:     
00010 // Copyright:       (c) Aaron Darling 
00011 // Licenses:        Proprietary 
00013 
00014 #ifndef _gnGBKSource_h_
00015 #define _gnGBKSource_h_
00016 
00017 #include "gn/gnDefs.h"
00018 
00019 #include <string>
00020 #include <fstream>
00021 #include <vector>
00022 #include "gn/gnFileSource.h"
00023 #include "gn/gnFileContig.h"
00024 #include "gn/gnSourceSpec.h"
00025 #include "gn/gnSequence.h"
00026 
00027 const uint32 SEQ_COLUMN_WIDTH = 80;
00028 const uint32 SEQ_HEADER_NAME_LENGTH = 11;
00029 const uint32 SEQ_SUBTAG_COLUMN = 5;
00030 const uint32 SEQ_LOCUS_CIRCULAR_COLUMN = 43;
00031 const uint32 SEQ_LOCUS_NAME_COLUMN = 13;
00032 const uint32 SEQ_LOCUS_NAME_LENGTH = 10;
00033 const uint32 SEQ_LOCUS_SIZE_LENGTH = 7;
00034 const uint32 SEQ_LOCUS_DNATYPE_OFFSET = 33;
00035 const uint32 SEQ_LOCUS_DNATYPE_LENGTH = 7;
00036 const uint32 SEQ_LOCUS_DIVCODE_OFFSET = 52;
00037 const uint32 SEQ_LOCUS_DIVCODE_LENGTH = 3;
00038 const uint32 SEQ_LOCUS_DATE_OFFSET = 62;
00039 const uint32 SEQ_LOCUS_DATE_LENGTH = 11;
00040 const uint32 SEQ_FEATURE_LOC_OFFSET = 21;
00041 const uint32 SEQ_BASES_INDEX_END = 9;
00042 
00050 
00051 class GNDLLEXPORT gnGBKSource : public gnFileSource
00052 {
00053 public:
00057         gnGBKSource();  
00062         gnGBKSource( const gnGBKSource& s );
00066         ~gnGBKSource();
00070         gnGBKSource* Clone() const;
00071 // Contig Access methods        
00072         uint32 GetContigListLength() const;
00073         boolean HasContig( const string& name ) const;
00074         uint32 GetContigID( const string& name ) const;
00075         string GetContigName( const uint32 i ) const;
00076         gnSeqI GetContigSeqLength( const uint32 i ) const;
00077 
00078         boolean SeqRead( const gnSeqI start, char* buf, uint32& bufLen, const uint32 contigI=ALL_CONTIGS );
00079 
00086         static boolean Write(gnSequence& seq, const string& filename);
00093         static boolean Write(gnBaseSource *source, const string& filename);
00094         gnGenomeSpec *GetSpec() const;
00095         gnFileContig* GetFileContig( const uint32 contigI ) const;
00096 private:
00097         boolean SeqSeek( const gnSeqI start, const uint32& contigI, uint64& startPos, uint64& readableBytes );
00098         boolean SeqStartPos( const gnSeqI start, gnFileContig& contig, uint64& startPos, uint64& readableBytes );
00099         boolean ParseStream( istream& fin );
00100 
00101         static string& Filler(uint32 length);
00102         static void FormatString(string& data, uint32 offset, uint32 width);
00103         static void WriteHeader(gnMultiSpec* spec, const string& hdr, ofstream& m_ofstream);
00104 //      gnSeqI m_seqLength;
00105         
00106         gnGenomeSpec *m_spec;
00107         vector< gnFileContig* > m_contigList;   
00108 };// class gnGBKSource
00109 // Clone        
00110 inline
00111 gnGBKSource* gnGBKSource::Clone() const
00112 {
00113         return new gnGBKSource( *this );
00114 }
00115 // Contig Access methods        
00116 inline
00117 uint32 gnGBKSource::GetContigListLength() const
00118 {
00119         return m_contigList.size();
00120 }
00121 inline
00122 boolean gnGBKSource::Write(gnBaseSource *source, const string& filename){
00123         gnSequence gns(*source->GetSpec());
00124         return Write(gns, filename);
00125 }
00126 inline
00127 gnGenomeSpec *gnGBKSource::GetSpec() const{
00128         return m_spec->Clone();
00129 }
00130 
00131 #endif
00132         // _gnGBKSource_h_