seq_id.hpp

ncbi源码

    int WorstRankScore(void) const
        { return AdjustScore(BaseWorstRankScore()); }

    // Wrappers for use with FindBestChoice from <corelib/ncbiutil.hpp>
    static int Score(const CRef<CSeq_id>& id)
        { return id ? id->TextScore() : kMax_Int; }
    static int BestRank(const CRef<CSeq_id>& id)
        { return id ? id->BestRankScore() : kMax_Int; }
    static int WorstRank(const CRef<CSeq_id>& id)
        { return id ? id->WorstRankScore() : kMax_Int; }

    virtual void Assign(const CSerialObject& source,
                        ESerialRecursionMode how = eRecursive);
    virtual bool Equals(const CSerialObject& object,
                        ESerialRecursionMode how = eRecursive) const;

private:
    void x_Init
    (CSeq_id_Base::E_Choice the_type,
     // Just first string, as in text seqid, for unusual
     // cases (patents, pdb) not really an acc
     const string&          acc_in,
     const string&          name_in    = kEmptyStr,
     int                    version    = 0,
     const string&          release_in = kEmptyStr);

    // Prohibit copy constructor & assignment operator
    CSeq_id(const CSeq_id&);
    CSeq_id& operator= (const CSeq_id&);

    //CRef<CAbstractObjectManager> m_ObjectManager;

};


// Search the container of CRef<CSeq_id> for the id of given type.
// Return the id of requested type, or null CRef.
template<class container>
CRef<CSeq_id> GetSeq_idByType(const container& ids,
                                    CSeq_id::E_Choice choice)
{
    ITERATE (typename container, iter, ids) {
        if ((*iter)->Which() == choice){
            return *iter;
        }
    }
    return CRef<CSeq_id>(0);
}

//return gi from id list if exists, return 0 otherwise
template<class container>
int FindGi(const container& ids)
{
    CRef<CSeq_id> id = GetSeq_idByType(ids, CSeq_id::e_Gi);
    return id ? id->GetGi() : 0;
}


//return text seq-id from id list if exists, return 0 otherwise
template<class container>
CRef<CSeq_id> FindTextseq_id(const container& ids)
{
    ITERATE (typename container, iter, ids) {
        if ( (*iter)->GetTextseq_Id() ) {
            return *iter;
        }
    }
    return CRef<CSeq_id>(0);
}


/////////////////// CSeq_id inline methods

// Match - just uses Compare
inline
bool CSeq_id::Match (const CSeq_id& sid2) const
{
    return Compare(sid2) == e_YES;
}


inline
int CSeq_id::AdjustScore(int base_score) const
{
    int score = base_score * 10;
    const CTextseq_id* text_id = GetTextseq_Id();
    if (text_id) {
        if ( !text_id->IsSetVersion() ) {
            score += 4;
        }
        if ( !text_id->IsSetAccession() ) {
            score += 3;
        }
        if ( !text_id->IsSetName() ) {
            score += 2;
        }
    }
    return score;
}


inline
int CSeq_id::BaseTextScore(void) const
{
    switch (Which()) {
    case e_not_set:                                return kMax_Int;
    case e_Giim:    case e_Gi:                     return 20;
    case e_General: case e_Gibbsq: case e_Gibbmt:  return 15;
    case e_Local:   case e_Patent:                 return 10;
    case e_Other:                                  return 8;
    default:                                       return 5;
    }
}


inline
int CSeq_id::BaseBestRankScore(void) const
{
    switch (Which()) {
    case e_not_set:                               return 83;
    case e_General: case e_Local:                 return 80;
    case e_Gibbsq: case e_Gibbmt: case e_Giim:    return 70;
    case e_Patent:                                return 67;
    case e_Other:                                 return 65;

    case e_Ddbj: case e_Prf: case e_Pdb:
    case e_Tpe:  case e_Tpd: case e_Embl:
    case e_Pir:  case e_Swissprot:
    case e_Tpg:  case e_Genbank:                  return 60;

    case e_Gi:                                    return 51;
    default:                                      return 5;
    }
}


inline
int CSeq_id::BaseWorstRankScore(void) const
{
    switch (Which()) {
    case e_not_set:                               return 83;
    case e_Gi: case e_Giim:                       return 20;
    case e_General: case e_Gibbsq: case e_Gibbmt: return 15;
    case e_Local: case e_Patent:                  return 10;
    case e_Other:                                 return 8;

    case e_Ddbj: case e_Prf: case e_Pdb:
    case e_Tpe:  case e_Tpd: case e_Embl:
    case e_Pir:  case e_Swissprot:
    case e_Tpg:  case e_Genbank:                  return 5;

    default:                                      return 3;
    }
}


/////////////////// end of CSeq_id inline methods


END_objects_SCOPE // namespace ncbi::objects::
END_NCBI_SCOPE

/*
 * ===========================================================================
 *
 * $Log: Seq_id.hpp,v $
 * Revision 1000.2  2004/06/01 19:30:49  gouriano
 * PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.47
 *
 * Revision 1.47  2004/05/11 14:34:12  ucko
 * Factored out and refined Textseq-id bonus calculation.
 *
 * Revision 1.46  2004/04/28 14:13:17  grichenk
 * Added FindTextseq_id()
 *
 * Revision 1.45  2004/04/19 18:20:16  grichenk
 * Added GetSeq_idByType() and FindGi()
 *
 * Revision 1.44  2004/04/01 16:36:34  ucko
 * Make the scoring functions wrappers around const methods (with slightly
 * different names to avoid ambiguity when passing CSeq_id::Score et al. to
 * FindBestChoice) and add a more verbose comment.  Also, WorstRank,
 * despite being "Worst," should still favor versioned accessions.
 *
 * Revision 1.43  2004/03/30 20:46:08  ucko
 * BestRank: demote gibb* and patent to 70 and 67 respectively per the C Toolkit.
 *
 * Revision 1.42  2004/03/25 15:58:41  gouriano
 * Added possibility to copy and compare serial object non-recursively
 *
 * Revision 1.41  2004/03/01 18:26:04  dicuccio
 * Modified CSeq_id::Score(), CSeq_id::BestRank(), and CSeq_id::WorstRank to
 * consider an accession's version if present.
 *
 * Revision 1.40  2004/01/22 21:03:58  dicuccio
 * Separated functionality of enums in GetLabel() into discrete mode and flags
 *
 * Revision 1.39  2004/01/22 18:45:45  dicuccio
 * Added new API: CSeq_id::GetLabel().  Rewired GetSeqIdString() to feed into
 * GetLabel().  Rewired GetStringDescr() to feed into GetLabel() directly instead
 * of feeding through GetSeqIdString().
 *
 * Revision 1.38  2004/01/21 18:04:20  dicuccio
 * Added ctor to create a seq-id from a given dbtag, performing conversion to
 * specific seq-id types where possible
 *
 * Revision 1.37  2004/01/16 22:11:15  ucko
 * Update for new CSerializable interface.
 *
 * Revision 1.36  2004/01/09 19:54:13  ucko
 * Give EXAMPLE prefixes to ease migration from hardcoded checks.
 *
 * Revision 1.35  2003/10/24 14:55:10  ucko
 * Merged eAcc_con and eAcc_segset (had been mistakenly separated);
 * renamed eAcc_gb_segset to eAcc_gb_con, but left the old name as a synoynm.
 *
 * Revision 1.34  2003/02/06 22:23:29  vasilche
 * Added CSeq_id::Assign(), CSeq_loc::Assign().
 * Added int CSeq_id::Compare() (not safe).
 * Added caching of CSeq_loc::GetTotalRange().
 *
 * Revision 1.33  2003/02/04 15:15:11  grichenk
 * Overrided Assign() for CSeq_loc and CSeq_id
 *
 * Revision 1.32  2003/01/18 08:40:04  kimelman
 * addes seqid constructor for numeric types
 *
 * Revision 1.31  2003/01/07 19:52:07  ucko
 * Add more refseq types (NR_, NS_, NW_) and list the prefixes as comments.
 *
 * Revision 1.30  2002/12/26 16:39:22  vasilche
 * Object manager class CSeqMap rewritten.
 *
 * Revision 1.29  2002/12/26 12:43:42  dicuccio
 * Added Win32 export specifiers
 *
 * Revision 1.28  2002/11/26 15:12:15  dicuccio
 * Added general processing function to retrieve sequence id descriptions in a
 * number of formats.
 *
 * Revision 1.27  2002/10/23 18:22:57  ucko
 * Add self-classification (using known type information) and expand
 * EAccessionInfo accordingly.
 *
 * Revision 1.26  2002/10/22 20:18:30  jianye
 * Added GetSeqIdString()
 *
 * Revision 1.25  2002/10/03 18:51:11  clausen
 * Removed extra whitespace
 *
 * Revision 1.24  2002/10/03 17:06:13  clausen
 * Added BestRank() and WorstRank()
 *
 * Revision 1.23  2002/08/22 21:25:26  ucko
 * Added a standard score function for FindBestChoice corresponding to
 * the table in SeqIdFindWorst.
 *
 * Revision 1.22  2002/08/16 19:25:14  ucko
 * +eAcc_refseq_wgs_*
 *
 * Revision 1.21  2002/08/14 15:52:05  ucko
 * Add eAcc_refseq_ncrna(_predicted) [non-coding RNA, XR_].
 *
 * Revision 1.20  2002/08/01 20:32:52  ucko
 * s_IdentifyAccession -> IdentifyAccession; s_ is only for module-static names.
 *
 * Revision 1.19  2002/07/30 19:41:33  ucko
 * Add s_IdentifyAccession (along with EAccessionInfo and GetAccType).
 * Move CVS log to end.
 *
 * Revision 1.18  2002/06/07 11:16:57  clausen
 * Fixed comments
 *
 * Revision 1.17  2002/06/07 11:13:01  clausen
 * Added comment about util/sequence.hpp
 *
 * Revision 1.16  2002/06/06 20:32:01  clausen
 * Moved methods using object manager to objects/util
 *
 * Revision 1.15  2002/05/22 14:03:34  grichenk
 * CSerialUserOp -- added prefix UserOp_ to Assign() and Equals()
 *
 * Revision 1.14  2002/05/03 21:28:04  ucko
 * Introduce T(Signed)SeqPos.
 *
 * Revision 1.13  2002/01/10 18:43:34  clausen
 * Added GetLength
 *
 * Revision 1.12  2001/08/31 15:59:59  clausen
 * Added new constructors for FastA construction and added tpg, tpd, and tpe id types
 *
 * Revision 1.11  2001/07/25 19:11:09  grichenk
 * Equals() and Assign() re-declared as protected
 *
 * Revision 1.10  2001/07/16 16:22:44  grichenk
 * Added CSerialUserOp class to create Assign() and Equals() methods for
 * user-defind classes.
 * Added SerialAssign<>() and SerialEquals<>() functions.
 *
 * Revision 1.9  2001/06/25 18:52:01  grichenk
 * Prohibited copy constructor and assignment operator
 *
 * Revision 1.8  2001/04/17 04:13:03  vakatov
 * Utilize the redesigned "CSerializable" base class.
 * Completely get rid of the non-standard "AsFastaString()" method in
 * favor of more standard "DumpAsFasta()" one.
 *
 * Revision 1.7  2001/04/16 16:55:19  kholodov
 * Modified: Added implementation for the ISerializable interface.
 *
 * Revision 1.6  2001/01/03 16:38:53  vasilche
 * Added CAbstractObjectManager - stub for object manager.
 * CRange extracted to separate file.
 *
 * Revision 1.5  2000/12/26 17:28:34  vasilche
 * Simplified and formatted code.
 *
 * Revision 1.4  2000/12/08 22:18:41  ostell
 * changed MakeFastString to AsFastaString and to use ostream instead of string
 *
 * Revision 1.3  2000/12/08 20:45:56  ostell
 * added MakeFastaString()
 *
 * Revision 1.2  2000/11/27 20:36:39  vasilche
 * Enum should be defined in public area.
 *
 * Revision 1.1  2000/11/21 18:58:12  vasilche
 * Added Match() methods for CSeq_id, CObject_id and CDbtag.
 *
 * ===========================================================================
 */

#endif // OBJECTS_SEQLOC_SEQ_ID_HPP