genemark_loader.cpp

来自「ncbi源码」· C++ 代码 · 共 564 行 · 第 1/2 页

    annot->AddName("GeneMark.hmm predictions");

    list< CRef<CSeq_feat> >& ftable = annot->SetData().SetFtable(); 

    char buf[ GENEMARK_MAXLINE +1 ];

    while( istr.getline(buf, GENEMARK_MAXLINE) )  // find and skip file header
    {
        if(strstr(buf, "Strand") && strstr(buf,"RightEnd"))
        {
            istr.getline(buf, GENEMARK_MAXLINE); // skip another one line
            break;
        }
    }

    while( istr.getline(buf, GENEMARK_MAXLINE) )
    {
        CNcbiIstrstream Lstr( buf );      // LOG_POST(Info << "Parsing line:" << buf);

        int     gene_number = 0;
        Lstr >> gene_number;

        char strand = 'U';
        for(; (strand != '-') && (strand != '+'); Lstr >> strand);

        string aLeftEnd;
        TSeqPos LeftEnd, RightEnd;

        Lstr >> aLeftEnd;
        Lstr >>  RightEnd;

        if(aLeftEnd[0] == '<') aLeftEnd[0] = ' '; 

        LeftEnd = NStr::StringToInt(aLeftEnd);

        //  _TRACE("parsed line: "<< gene_number <<" from: "<< LeftEnd <<" to: "<< RightEnd <<" starnd: "<< strand);

        CRef<CSeq_feat> feat(new CSeq_feat);

        feat->SetComment() = "GeneMark.hmm pred #" + NStr::IntToString(gene_number) ;

        CSeq_interval& floc = feat->SetLocation().SetInt();

        floc.SetFrom(LeftEnd- 1);
        floc.SetTo(RightEnd - 1); 
        floc.SetStrand((strand == '-') ? eNa_strand_minus : eNa_strand_plus);

        floc.SetId().Assign(id);    // floc.SetId().SetGi( NStr::StringToInt(m_SeqId) );

        CSeqFeatData& fdata = feat->SetData();
        CCdregion& cdreg = fdata.SetCdregion();
        cdreg.SetFrame(CCdregion::eFrame_one);

        list< CRef< CGenetic_code::C_E > >& gcode = cdreg.SetCode().Set();
        CRef< CGenetic_code::C_E > ce(new CGenetic_code::C_E);
        ce->SetId(11);                                        // TSE=1; seq=1; feat=1
        gcode.push_back(ce);

        ftable.push_back(feat);
    }
    return annot;
}

CRef<CSeq_annot>
CGeneMarkLoader::x_LoadGeneMarkFile(const string& fname, const CSeq_id& id)
{
    char buf[ GENEMARK_MAXLINE +1 ];
    CNcbiIfstream istr(fname.c_str());

    while(istr.getline(buf, GENEMARK_MAXLINE))  // find and skip file header
    {
        if(strstr(buf, "Strand") && strstr(buf,"Frame"))
        {
            istr.getline(buf, GENEMARK_MAXLINE); // skip  ----- -----
            // istr.getline(buf, GENEMARK_MAXLINE); // sometimes NOT empty !
            break;
        }
    }

    CRef<CSeq_annot> annot( new CSeq_annot() );

    annot->AddName("GeneMark predictions");

    list< CRef<CSeq_feat> >& ftable = annot->SetData().SetFtable();

    TSeqPos prevStop = 0;
    int gene_number  = 0;
    CRef<CSeq_feat> feat;

    CSeq_interval *floc;

    while( istr.getline(buf, GENEMARK_MAXLINE) )
    {
        if(strstr(buf, "interest")) break; // we have reached "List of Regions of interest"
        if(strlen(buf) < 10) continue;

        // LOG_POST(Info << "CGeneMarkLoader::Load: parsing line: " << buf );

        CNcbiIstrstream Lstr( buf );

        string  strand;
        TSeqPos LeftEnd, RightEnd, curStop;

        Lstr >> LeftEnd;
        Lstr >> RightEnd;
        Lstr >> strand;

        bool  is_complementary = (strand.find("complement") != string::npos);

        curStop = is_complementary ? LeftEnd : RightEnd;

        if(curStop != prevStop)  //  new lines-group
        {
            prevStop = curStop;
            ++gene_number;

            feat = new CSeq_feat();
            feat->SetComment() = "GeneMark pred #" + NStr::IntToString(gene_number) ;
            ftable.push_back(feat);

            CSeqFeatData& fdata = feat->SetData();
            CCdregion& cdreg = fdata.SetCdregion();
            cdreg.SetFrame(CCdregion::eFrame_one);

            list< CRef< CGenetic_code::C_E > >& gcode = cdreg.SetCode().Set();
            CRef< CGenetic_code::C_E > ce(new CGenetic_code::C_E);
            ce->SetId(11);                                        // TSE=1; seq=1; feat=1
            gcode.push_back(ce);

            floc = & feat->SetLocation().SetInt();

            floc->SetFrom(LeftEnd-1);
            floc->SetTo(RightEnd -1); 
            floc->SetStrand(is_complementary ? eNa_strand_minus : eNa_strand_plus);

            floc->SetId().Assign(id);   // floc->SetId().SetGi( NStr::StringToInt(m_SeqId) );
        }
        else    // another start of previous gene
        {    
            if(is_complementary)
            {
                floc->SetFuzz_to().SetAlt().push_back( RightEnd -1);
            }
            else
            {
                floc->SetFuzz_from().SetAlt().push_back( LeftEnd -1); // LOG_POST(Info << "Left:"<<LeftEnd);
            }  
        }
    }
    return annot;
}

END_NCBI_SCOPE


/*
 * =====================================================================
 * $Log: genemark_loader.cpp,v $
 * Revision 1000.5  2004/06/01 20:58:40  gouriano
 * PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.35
 *
 * Revision 1.35  2004/05/21 22:27:48  gorelenk
 * Added PCH ncbi_pch.hpp
 *
 * Revision 1.34  2004/03/11 17:44:00  dicuccio
 * Use new file loader dialog
 *
 * Revision 1.33  2003/12/22 20:29:47  dernovoy
 * skip Glimmer's prediction over zero-point
 *
 * Revision 1.32  2003/12/16 20:24:55  dernovoy
 * Path to precomputed predictions should work on both Windows and UNIX inside NCBI
 *
 * Revision 1.31  2003/12/10 22:53:24  dernovoy
 * defaults directory and file extensions added
 *
 * Revision 1.30  2003/12/09 23:22:14  dernovoy
 * menu string changed : coma instead of slash
 *
 * Revision 1.29  2003/12/09 21:46:00  dernovoy
 * Glimmer2 output loader was added
 *
 * Revision 1.28  2003/11/24 15:45:40  dicuccio
 * Renamed CVersion to CPluginVersion
 *
 * Revision 1.27  2003/11/18 17:49:26  dicuccio
 * Added standard processing of return values
 *
 * Revision 1.26  2003/11/04 17:49:25  dicuccio
 * Changed calling parameters for plugins - pass CPluginMessage instead of paired
 * CPluginCommand/CPluginReply
 *
 * Revision 1.25  2003/10/10 17:19:33  dicuccio
 * Added Import() interface.  Removed dead Save() interfaces
 *
 * Revision 1.24  2003/10/07 13:47:06  dicuccio
 * Renamed CPluginURL* to CPluginValue*
 *
 * Revision 1.23  2003/09/17 16:27:28  dicuccio
 * Removed load command
 *
 * Revision 1.22  2003/09/04 14:51:59  dicuccio
 * Use IDocument instead of CDocument
 *
 * Revision 1.21  2003/07/14 11:17:25  shomrat
 * Plugin messageing system related changes
 *
 * Revision 1.20  2003/06/25 17:02:59  dicuccio
 * Split CPluginHandle into a handle (pointer-to-implementation) and
 * implementation file.  Lots of #include file clean-ups.
 *
 * Revision 1.19  2003/06/20 14:52:58  dicuccio
 * Revised plugin registration - moved GetInfo() into the plugin handler
 *
 * Revision 1.18  2003/05/19 13:40:45  dicuccio
 * Moved gui/core/plugin/ -> gui/plugin/.  Merged core libraries into libgui_core.
 * Removed old, unused dialog box.
 *
 * Revision 1.17  2003/04/24 16:39:29  dicuccio
 * Updated to reflect changes in plugin API
 *
 * Revision 1.16  2003/02/24 13:03:16  dicuccio
 * Renamed classes in plugin spec:
 *     CArgSeg --> CPluginArgSet
 *     CArgument --> CPluginArg
 *     CPluginArgs --> CPluginCommand
 *     CPluginCommands --> CPluginCommandSet
 *
 * Revision 1.15  2003/02/20 19:49:56  dicuccio
 * Created new plugin architecture, based on ASN.1 spec.  Moved GBENCH frameowrk
 * over to use new plugin architecture.
 *
 * Revision 1.14  2003/02/06 18:48:36  dicuccio
 * Made 'catch (...)' conditional for non-debug builds
 *
 * Revision 1.13  2003/01/15 19:47:37  dernovoy
* accession can be used for seq_id in features (was: only gi).
*
* Revision 1.12  2003/01/13 13:10:07  dicuccio
* Namespace clean-up.  Retired namespace gui -> converted all to namespace ncbi.
* Moved all FLUID-generated code into namespace ncbi.
*
* Revision 1.11  2003/01/08 21:17:59  dernovoy
* Feature's location fixed (from local to gi),
* GeneMark coordinates translated to ncbi format (0 - (Len-1))
    *
    * Revision 1.10  2003/01/06 21:03:12  dernovoy
    * Support for Alternative starts (fuzz-from/to) of GeneMark's output
    *
    * Revision 1.9  2003/01/02 21:41:37  dernovoy
    * Load of genemark output added, the farthest starts taken for features
    *
    * Revision 1.8  2003/01/02 19:58:38  dernovoy
    * fix comparing stream with int
    *
    * Revision 1.7  2002/12/31 19:45:20  dernovoy
    * Addded support for genemarkHMM start positions started with symbol '<'
    *
    * Revision 1.6  2002/12/30 17:48:29  dicuccio
* Added mechanism for data loader plugins to announce supported modes of
* operation (load, import, save currently)
    *
    * Revision 1.5  2002/12/30 15:47:10  dernovoy
    * TRACE outputs added, any doc's updates comments out
    *
    * Revision 1.4  2002/12/26 20:50:25  dernovoy
    * *** empty log message ***
    *
    * Revision 1.3  2002/12/26 20:48:59  dernovoy
    * Log output
    *
* Revision 1.2  2002/12/26 17:45:55  dicuccio
* Reformatted code (reindent)
    *
    * Revision 1.1  2002/12/26 17:12:38  dernovoy
    * Initial revision.
    *
    * =====================================================================
    */