readfeat.cpp

来自「ncbi源码」· C++ 代码 · 共 1,648 行 · 第 1/5 页

{
    string line;
    string feat, qual, val;
    Int4 start, stop;
    bool partial5, partial3, ispoint;
    Int4 offset = 0;
    CSeqFeatData::ESubtype sbtyp = CSeqFeatData::eSubtype_bad;
    CSeqFeatData::E_Choice typ = CSeqFeatData::e_not_set;
    CRef<CSeq_annot> sap(new CSeq_annot);
    CSeq_annot::C_Data::TFtable& ftable = sap->SetData().SetFtable();
    CRef<CSeq_feat> sfp;
    CSeq_loc_mix *mix = 0;
    CT_POS_TYPE pos(0);

    if (! annotname.empty ()) {
      CAnnot_descr& descr = sap->SetDesc ();
      CRef<CAnnotdesc> annot(new CAnnotdesc);
      annot->SetName (annotname);
      descr.Set().push_back (annot);
    }

    while (ifs.good ()) {

        pos = ifs.tellg ();
        NcbiGetlineEOL (ifs, line);

        if (! line.empty ()) {
            if (line [0] == '>') {

                // if next feature table, reposition and return current sap

                ifs.seekg (pos);
                return sap;

            } else if (line [0] == '[') {

                // set offset !!!!!!!!

            } else if (x_ParseFeatureTableLine (line, &start, &stop, &partial5, &partial3,
                                                &ispoint, feat, qual, val, offset)) {

                // process line in feature table

                if ((! feat.empty ()) && start >= 0 && stop >= 0) {

                    // process start - stop - feature line

                    if (m_FeatKeys.find (feat) != m_FeatKeys.end ()) {
                        sbtyp = m_FeatKeys [feat];
                        if (sbtyp != CSeqFeatData::eSubtype_bad) {

                            // populate *sfp here...

                            sfp.Reset (new CSeq_feat);
                            sfp->ResetLocation ();

                            typ = CSeqFeatData::GetTypeFromSubtype (sbtyp);
                            sfp->SetData ().Select (typ);
                            CSeqFeatData& sfdata = sfp->SetData ();
                            if (typ == CSeqFeatData::e_Rna) {
                                CRNA_ref& rrp = sfdata.SetRna ();
                                CRNA_ref::EType rnatyp = CRNA_ref::eType_unknown;
                                switch (sbtyp) {
                                    case CSeqFeatData::eSubtype_preRNA :
                                        rnatyp = CRNA_ref::eType_premsg;
                                        break;
                                    case CSeqFeatData::eSubtype_mRNA :
                                        rnatyp = CRNA_ref::eType_mRNA;
                                        break;
                                    case CSeqFeatData::eSubtype_tRNA :
                                        rnatyp = CRNA_ref::eType_tRNA;
                                        break;
                                    case CSeqFeatData::eSubtype_rRNA :
                                        rnatyp = CRNA_ref::eType_rRNA;
                                        break;
                                    case CSeqFeatData::eSubtype_snRNA :
                                        rnatyp = CRNA_ref::eType_snRNA;
                                        break;
                                    case CSeqFeatData::eSubtype_scRNA :
                                        rnatyp = CRNA_ref::eType_scRNA;
                                        break;
                                    case CSeqFeatData::eSubtype_snoRNA :
                                        rnatyp = CRNA_ref::eType_snoRNA;
                                        break;
                                    case CSeqFeatData::eSubtype_otherRNA :
                                        rnatyp = CRNA_ref::eType_other;
                                        break;
                                    default :
                                        break;
                                }
                                rrp.SetType (rnatyp);

                            } else if (typ == CSeqFeatData::e_Imp) {
                                CImp_feat_Base& imp = sfdata.SetImp ();
                                imp.SetKey (feat);
                            }

                            ftable.push_back (sfp);

                            // now create location

                            CRef<CSeq_loc> location (new CSeq_loc);
                            mix = &(location->SetMix ());
                            sfp->SetLocation (*location);

                            // and add first interval

                            x_AddIntervalToFeature (sfp, mix, seqid, start, stop, partial5, partial3);

                        }
                    } else {

                        // unrecognized feature key

                        if ((flags & CFeature_table_reader::fReportBadKey) != 0) {
                            ERR_POST (Warning << "Unrecognized feature " << feat);
                        }

                        if ((flags & CFeature_table_reader::fKeepBadKey) != 0) {
                            sfp.Reset (new CSeq_feat);
                            sfp->ResetLocation ();
                            sfp->SetData ().Select (CSeqFeatData::e_Imp);
                            CSeqFeatData& sfdata = sfp->SetData ();
                            CImp_feat_Base& imp = sfdata.SetImp ();
                            imp.SetKey (feat);
                            ftable.push_back (sfp);
                            CRef<CSeq_loc> location (new CSeq_loc);
                            mix = &(location->SetMix ());
                            sfp->SetLocation (*location);
                            x_AddIntervalToFeature (sfp, mix, seqid, start, stop, partial5, partial3);
                        }
                    }

                } else if (start >= 0 && stop >= 0 && feat.empty () && qual.empty () && val.empty ()) {

                    // process start - stop multiple interval line

                    x_AddIntervalToFeature (sfp, mix, seqid, start, stop, partial5, partial3);

                } else if ((! qual.empty ()) && (! val.empty ())) {

                    // process qual - val qualifier line

                    if (! x_AddQualifierToFeature (sfp, qual, val)) {

                        // unrecognized qualifier key

                        if ((flags & CFeature_table_reader::fReportBadKey) != 0) {
                            ERR_POST (Warning << "Unrecognized qualifier " << qual);
                        }

                        if ((flags & CFeature_table_reader::fKeepBadKey) != 0) {
                            CSeq_feat::TQual& qlist = sfp->SetQual ();
                            CRef<CGb_qual> gbq (new CGb_qual);
                            gbq->SetQual (qual);
                            gbq->SetVal (val);
                            qlist.push_back (gbq);
                        }
                    }

                } else if ((! qual.empty ()) && (val.empty ())) {

                    // check for the few qualifiers that do not need a value

                    if (find (m_SingleKeys.begin (), m_SingleKeys.end (), qual) != m_SingleKeys.end ()) {

                        x_AddQualifierToFeature (sfp, qual, val);

                    }
                } else if (! feat.empty ()) {
                
                    // unrecognized location

                    if ((flags & CFeature_table_reader::fReportBadKey) != 0) {
                        ERR_POST (Warning << "Bad location on feature " << feat <<
                                 " (start " << start << ", stop " << stop << ")");
                    }
                }
            }
        }
    }

    return sap;
}


CRef<CSeq_feat> CFeature_table_reader_imp::CreateSeqFeat (
    const string& feat,
    CSeq_loc& location,
    const CFeature_table_reader::TFlags flags
)

{
    CRef<CSeq_feat> sfp (new CSeq_feat);

    if (! feat.empty ()) {
        if (m_FeatKeys.find (feat) != m_FeatKeys.end ()) {
            CSeqFeatData::ESubtype sbtyp = m_FeatKeys [feat];
            CSeqFeatData::E_Choice typ = CSeqFeatData::GetTypeFromSubtype (sbtyp);
            sfp->SetData ().Select (typ);
            CSeqFeatData& sfdata = sfp->SetData ();
            sfp->SetLocation (location);
            if (typ == CSeqFeatData::e_Rna) {
                CRNA_ref& rrp = sfdata.SetRna ();
                CRNA_ref::EType rnatyp = CRNA_ref::eType_unknown;
                switch (sbtyp) {
                    case CSeqFeatData::eSubtype_preRNA :
                        rnatyp = CRNA_ref::eType_premsg;
                        break;
                    case CSeqFeatData::eSubtype_mRNA :
                        rnatyp = CRNA_ref::eType_mRNA;
                        break;
                    case CSeqFeatData::eSubtype_tRNA :
                        rnatyp = CRNA_ref::eType_tRNA;
                        break;
                    case CSeqFeatData::eSubtype_rRNA :
                        rnatyp = CRNA_ref::eType_rRNA;
                        break;
                    case CSeqFeatData::eSubtype_snRNA :
                        rnatyp = CRNA_ref::eType_snRNA;
                        break;
                    case CSeqFeatData::eSubtype_scRNA :
                        rnatyp = CRNA_ref::eType_scRNA;
                        break;
                    case CSeqFeatData::eSubtype_snoRNA :
                        rnatyp = CRNA_ref::eType_snoRNA;
                        break;
                    case CSeqFeatData::eSubtype_otherRNA :
                        rnatyp = CRNA_ref::eType_other;
                        break;
                    default :
                        break;
                }
                rrp.SetType (rnatyp);

            } else if (typ == CSeqFeatData::e_Imp) {
                CImp_feat_Base& imp = sfdata.SetImp ();
                imp.SetKey (feat);
            }
            sfp->SetLocation (location);
 
        } else {

            // unrecognized feature key

            if ((flags & CFeature_table_reader::fReportBadKey) != 0) {
                ERR_POST (Warning << "Unrecognized feature " << feat);
            }

            if ((flags & CFeature_table_reader::fKeepBadKey) != 0) {
                sfp.Reset (new CSeq_feat);
                sfp->ResetLocation ();
                sfp->SetData ().Select (CSeqFeatData::e_Imp);
                CSeqFeatData& sfdata = sfp->SetData ();
                CImp_feat_Base& imp = sfdata.SetImp ();
                imp.SetKey (feat);
                sfp->SetLocation (location);
            }
        }
    }

    return sfp;
}


void CFeature_table_reader_imp::AddFeatQual (
    CRef<CSeq_feat> sfp,
    const string& qual,
    const string& val,
    const CFeature_table_reader::TFlags flags
)

{
    if ((! qual.empty ()) && (! val.empty ())) {

        if (! x_AddQualifierToFeature (sfp, qual, val)) {

            // unrecognized qualifier key

            if ((flags & CFeature_table_reader::fReportBadKey) != 0) {
                ERR_POST (Warning << "Unrecognized qualifier " << qual);
            }

            if ((flags & CFeature_table_reader::fKeepBadKey) != 0) {
                CSeq_feat::TQual& qlist = sfp->SetQual ();
                CRef<CGb_qual> gbq (new CGb_qual);
                gbq->SetQual (qual);
                gbq->SetVal (val);
                qlist.push_back (gbq);
            }
        }

    } else if ((! qual.empty ()) && (val.empty ())) {

        // check for the few qualifiers that do not need a value

        if (find (m_SingleKeys.begin (), m_SingleKeys.end (), qual) != m_SingleKeys.end ()) {

            x_AddQualifierToFeature (sfp, qual, val);

        }
    }
}


// public access functions

CRef<CSeq_annot> CFeature_table_reader::ReadSequinFeatureTable (
    CNcbiIstream& ifs,
    const string& seqid,
    const string& annotname,
    const TFlags flags
)

{
    // just read features from 5-column table

    CRef<CSeq_annot> sap = x_GetImplementation ().ReadSequinFeatureTable (ifs, seqid, annotname, flags);

    // go through all features and demote single interval seqlocmix to seqlocint

    for (CTypeIterator<CSeq_feat> fi(*sap); fi; ++fi) {
        CSeq_feat& feat = *fi;
        CSeq_loc& location = feat.SetLocation ();
        if (location.IsMix ()) {
            CSeq_loc_mix& mx = location.SetMix ();
            switch (mx.Get ().size ()) {