readfeat.cpp

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

    { "terminator",         CSeqFeatData::eSubtype_terminator         },
    { "transit_peptide",    CSeqFeatData::eSubtype_transit_peptide_aa },
    { "transit_peptide_nt", CSeqFeatData::eSubtype_transit_peptide    },
    { "tRNA",               CSeqFeatData::eSubtype_tRNA               },
    { "TxInit",             CSeqFeatData::eSubtype_txinit             },
    { "unsure",             CSeqFeatData::eSubtype_unsure             },
    { "User",               CSeqFeatData::eSubtype_user               },
    { "variation",          CSeqFeatData::eSubtype_variation          },
    { "virion",             CSeqFeatData::eSubtype_virion             },
    { "V_region",           CSeqFeatData::eSubtype_V_region           },
    { "V_segment",          CSeqFeatData::eSubtype_V_segment          },
    { "Xref",               CSeqFeatData::eSubtype_seq                }
};

typedef struct qualinit {
    const char *                         key;
    CFeature_table_reader_imp::EQual subtype;
} QualInit;

static QualInit qual_key_to_subtype [] = {
    { "allele",               CFeature_table_reader_imp::eQual_allele               },
    { "anticodon",            CFeature_table_reader_imp::eQual_anticodon            },
    { "bond_type",            CFeature_table_reader_imp::eQual_bond_type            },
    { "bound_moiety",         CFeature_table_reader_imp::eQual_bound_moiety         },
    { "citation",             CFeature_table_reader_imp::eQual_citation             },
    { "clone",                CFeature_table_reader_imp::eQual_clone                },
    { "codon_start",          CFeature_table_reader_imp::eQual_codon_start          },
    { "cons_splice",          CFeature_table_reader_imp::eQual_cons_splice          },
    { "db_xref",              CFeature_table_reader_imp::eQual_db_xref              },
    { "direction",            CFeature_table_reader_imp::eQual_direction            },
    { "EC_number",            CFeature_table_reader_imp::eQual_EC_number            },
    { "evidence",             CFeature_table_reader_imp::eQual_evidence             },
    { "exception",            CFeature_table_reader_imp::eQual_exception            },
    { "frequency",            CFeature_table_reader_imp::eQual_frequency            },
    { "function",             CFeature_table_reader_imp::eQual_function             },
    { "gene",                 CFeature_table_reader_imp::eQual_gene                 },
    { "gene_desc",            CFeature_table_reader_imp::eQual_gene_desc            },
    { "gene_syn",             CFeature_table_reader_imp::eQual_gene_syn             },
    { "go_component",         CFeature_table_reader_imp::eQual_go_component         },
    { "go_function",          CFeature_table_reader_imp::eQual_go_function          },
    { "go_process",           CFeature_table_reader_imp::eQual_go_process           },
    { "insertion_seq",        CFeature_table_reader_imp::eQual_insertion_seq        },
    { "label",                CFeature_table_reader_imp::eQual_label                },
    { "locus_tag",            CFeature_table_reader_imp::eQual_locus_tag            },
    { "macronuclear",         CFeature_table_reader_imp::eQual_macronuclear         },
    { "map",                  CFeature_table_reader_imp::eQual_map                  },
    { "mod_base",             CFeature_table_reader_imp::eQual_mod_base             },
    { "note",                 CFeature_table_reader_imp::eQual_note                 },
    { "number",               CFeature_table_reader_imp::eQual_number               },
    { "operon",               CFeature_table_reader_imp::eQual_operon               },
    { "organism",             CFeature_table_reader_imp::eQual_organism             },
    { "partial",              CFeature_table_reader_imp::eQual_partial              },
    { "PCR_conditions",       CFeature_table_reader_imp::eQual_PCR_conditions       },
    { "phenotype",            CFeature_table_reader_imp::eQual_phenotype            },
    { "product",              CFeature_table_reader_imp::eQual_product              },
    { "protein_id",           CFeature_table_reader_imp::eQual_protein_id           },
    { "prot_desc",            CFeature_table_reader_imp::eQual_prot_desc            },
    { "pseudo",               CFeature_table_reader_imp::eQual_pseudo               },
    { "replace",              CFeature_table_reader_imp::eQual_replace              },
    { "rpt_family",           CFeature_table_reader_imp::eQual_rpt_family           },
    { "rpt_type",             CFeature_table_reader_imp::eQual_rpt_type             },
    { "rpt_unit",             CFeature_table_reader_imp::eQual_rpt_unit             },
    { "standard_name",        CFeature_table_reader_imp::eQual_standard_name        },
    { "transcript_id",        CFeature_table_reader_imp::eQual_transcript_id        },
    { "transl_except",        CFeature_table_reader_imp::eQual_transl_except        },
    { "transl_table",         CFeature_table_reader_imp::eQual_transl_table         },
    { "translation",          CFeature_table_reader_imp::eQual_translation          },
    { "transposon",           CFeature_table_reader_imp::eQual_transposon           },
    { "usedin",               CFeature_table_reader_imp::eQual_usedin               }
};

typedef struct orgrefinit {
    const char *                       key;
    CFeature_table_reader_imp::EOrgRef subtype;
} OrgRefInit;

static OrgRefInit orgref_key_to_subtype [] = {
    { "div",        CFeature_table_reader_imp::eOrgRef_div       },
    { "gcode",      CFeature_table_reader_imp::eOrgRef_gcode     },
    { "lineage",    CFeature_table_reader_imp::eOrgRef_lineage   },
    { "mgcode",     CFeature_table_reader_imp::eOrgRef_mgcode    },
    { "organelle",  CFeature_table_reader_imp::eOrgRef_organelle },
    { "organism",   CFeature_table_reader_imp::eOrgRef_organism  }
};

typedef struct genomeinit {
    const char *        key;
    CBioSource::EGenome subtype;
} GenomeInit;

static GenomeInit genome_key_to_subtype [] = {
    { "unknown",                   CBioSource::eGenome_unknown          },
    { "genomic",                   CBioSource::eGenome_genomic          },
    { "chloroplast",               CBioSource::eGenome_chloroplast      },
    { "plastid:chloroplast",       CBioSource::eGenome_chloroplast      },
    { "chromoplast",               CBioSource::eGenome_chromoplast      },
    { "plastid:chromoplast",       CBioSource::eGenome_chromoplast      },
    { "kinetoplast",               CBioSource::eGenome_kinetoplast      },
    { "mitochondrion:kinetoplast", CBioSource::eGenome_kinetoplast      },
    { "mitochondrion",             CBioSource::eGenome_mitochondrion    },
    { "plastid",                   CBioSource::eGenome_plastid          },
    { "macronuclear",              CBioSource::eGenome_macronuclear     },
    { "extrachrom",                CBioSource::eGenome_extrachrom       },
    { "plasmid",                   CBioSource::eGenome_plasmid          },
    { "transposon",                CBioSource::eGenome_transposon       },
    { "insertion_seq",             CBioSource::eGenome_insertion_seq    },
    { "cyanelle",                  CBioSource::eGenome_cyanelle         },
    { "plastid:cyanelle",          CBioSource::eGenome_cyanelle         },
    { "proviral",                  CBioSource::eGenome_proviral         },
    { "virion",                    CBioSource::eGenome_virion           },
    { "nucleomorph",               CBioSource::eGenome_nucleomorph      },
    { "apicoplast",                CBioSource::eGenome_apicoplast       },
    { "plastid:apicoplast",        CBioSource::eGenome_apicoplast       },
    { "plastid:leucoplast",        CBioSource::eGenome_leucoplast       },
    { "plastid:proplastid",        CBioSource::eGenome_proplastid       },
    { "endogenous_virus",          CBioSource::eGenome_endogenous_virus }
};

typedef struct subsrcinit {
    const char *         key;
    CSubSource::ESubtype subtype;
} SubSrcInit;

static SubSrcInit subsrc_key_to_subtype [] = {
    { "cell_line",            CSubSource::eSubtype_cell_line             },
    { "cell_type",            CSubSource::eSubtype_cell_type             },
    { "chromosome",           CSubSource::eSubtype_chromosome            },
    { "clone",                CSubSource::eSubtype_clone                 },
    { "clone_lib",            CSubSource::eSubtype_clone_lib             },
    { "country",              CSubSource::eSubtype_country               },
    { "dev_stage",            CSubSource::eSubtype_dev_stage             },
    { "endogenous_virus",     CSubSource::eSubtype_endogenous_virus_name },
    { "environmental_sample", CSubSource::eSubtype_environmental_sample  },
    { "frequency",            CSubSource::eSubtype_frequency             },
    { "genotype",             CSubSource::eSubtype_genotype              },
    { "germline",             CSubSource::eSubtype_germline              },
    { "haplotype",            CSubSource::eSubtype_haplotype             },
    { "insertion_seq",        CSubSource::eSubtype_insertion_seq_name    },
    { "isolation_source",     CSubSource::eSubtype_isolation_source      },
    { "lab_host",             CSubSource::eSubtype_lab_host              },
    { "map",                  CSubSource::eSubtype_map                   },
    { "plasmid",              CSubSource::eSubtype_plasmid_name          },
    { "plastid",              CSubSource::eSubtype_plastid_name          },
    { "pop_variant",          CSubSource::eSubtype_pop_variant           },
    { "rearranged",           CSubSource::eSubtype_rearranged            },
    { "segment",              CSubSource::eSubtype_segment               },
    { "sex",                  CSubSource::eSubtype_sex                   },
    { "subclone",             CSubSource::eSubtype_subclone              },
    { "tissue_lib ",          CSubSource::eSubtype_tissue_lib            },
    { "tissue_type",          CSubSource::eSubtype_tissue_type           },
    { "transgenic",           CSubSource::eSubtype_transgenic            },
    { "transposon",           CSubSource::eSubtype_transposon_name       }
};

typedef struct orgmodinit {
    const char *      key;
    COrgMod::ESubtype subtype;
} OrgModInit;

static OrgModInit orgmod_key_to_subtype [] = {
    { "acronym",          COrgMod::eSubtype_acronym          },
    { "anamorph",         COrgMod::eSubtype_anamorph         },
    { "authority",        COrgMod::eSubtype_authority        },
    { "biotype",          COrgMod::eSubtype_biotype          },
    { "biovar",           COrgMod::eSubtype_biovar           },
    { "breed",            COrgMod::eSubtype_breed            },
    { "chemovar",         COrgMod::eSubtype_chemovar         },
    { "common",           COrgMod::eSubtype_common           },
    { "cultivar",         COrgMod::eSubtype_cultivar         },
    { "dosage",           COrgMod::eSubtype_dosage           },
    { "ecotype",          COrgMod::eSubtype_ecotype          },
    { "forma_specialis",  COrgMod::eSubtype_forma_specialis  },
    { "forma",            COrgMod::eSubtype_forma            },
    { "gb_acronym",       COrgMod::eSubtype_gb_acronym       },
    { "gb_anamorph",      COrgMod::eSubtype_gb_anamorph      },
    { "gb_synonym",       COrgMod::eSubtype_gb_synonym       },
    { "group",            COrgMod::eSubtype_group            },
    { "isolate",          COrgMod::eSubtype_isolate          },
    { "nat_host",         COrgMod::eSubtype_nat_host         },
    { "pathovar",         COrgMod::eSubtype_pathovar         },
    { "serogroup",        COrgMod::eSubtype_serogroup        },
    { "serotype",         COrgMod::eSubtype_serotype         },
    { "serovar",          COrgMod::eSubtype_serovar          },
    { "specimen_voucher", COrgMod::eSubtype_specimen_voucher },
    { "strain",           COrgMod::eSubtype_strain           },
    { "sub_species",      COrgMod::eSubtype_sub_species      },
    { "subgroup",         COrgMod::eSubtype_subgroup         },
    { "substrain",        COrgMod::eSubtype_substrain        },
    { "subtype",          COrgMod::eSubtype_subtype          },
    { "synonym",          COrgMod::eSubtype_synonym          },
    { "teleomorph",       COrgMod::eSubtype_teleomorph       },
    { "type",             COrgMod::eSubtype_type             },
    { "variety",          COrgMod::eSubtype_variety          }
};

typedef struct bondinit {
    const char *        key;
    CSeqFeatData::EBond subtype;
} BondInit;

static BondInit bond_key_to_subtype [] = {
    { "disulfide",         CSeqFeatData::eBond_disulfide  },
    { "thiolester",        CSeqFeatData::eBond_thiolester },
    { "xlink",             CSeqFeatData::eBond_xlink      },
    { "thioether",         CSeqFeatData::eBond_thioether  },
    { "",                  CSeqFeatData::eBond_other      }
};

typedef struct siteinit {
    const char *        key;
    CSeqFeatData::ESite subtype;
} SiteInit;

static SiteInit site_key_to_subtype [] = {
    { "active",                      CSeqFeatData::eSite_active                      },
    { "binding",                     CSeqFeatData::eSite_binding                     },
    { "cleavage",                    CSeqFeatData::eSite_cleavage                    },
    { "inhibit",                     CSeqFeatData::eSite_inhibit                     },
    { "modified",                    CSeqFeatData::eSite_modified                    },
    { "glycosylation",               CSeqFeatData::eSite_glycosylation               },
    { "myristoylation",              CSeqFeatData::eSite_myristoylation              },
    { "mutagenized",                 CSeqFeatData::eSite_mutagenized                 },
    { "metal binding",               CSeqFeatData::eSite_metal_binding               },
    { "phosphorylation",             CSeqFeatData::eSite_phosphorylation             },
    { "acetylation",                 CSeqFeatData::eSite_acetylation                 },
    { "amidation",                   CSeqFeatData::eSite_amidation                   },
    { "methylation",                 CSeqFeatData::eSite_methylation                 },
    { "hydroxylation",               CSeqFeatData::eSite_hydroxylation               },
    { "sulfatation",                 CSeqFeatData::eSite_sulfatation                 },
    { "oxidative deamination",       CSeqFeatData::eSite_oxidative_deamination       },
    { "pyrrolidone carboxylic acid", CSeqFeatData::eSite_pyrrolidone_carboxylic_acid },
    { "gamma carboxyglutamic acid",  CSeqFeatData::eSite_gamma_carboxyglutamic_acid  },
    { "blocked",                     CSeqFeatData::eSite_blocked                     },
    { "lipid binding",               CSeqFeatData::eSite_lipid_binding               },
    { "np binding",                  CSeqFeatData::eSite_np_binding                  },
    { "DNA binding",                 CSeqFeatData::eSite_dna_binding                 },
    { "signal peptide",              CSeqFeatData::eSite_signal_peptide              },
    { "transit peptide",             CSeqFeatData::eSite_transit_peptide             },
    { "transmembrane region",        CSeqFeatData::eSite_transmembrane_region        },
    { "",                            CSeqFeatData::eSite_other                       }
};

typedef struct trnainit {
    const char * key;
    int          subtype;
} TrnaInit;

static TrnaInit trna_key_to_subtype [] = {
    { "Ala",   'A' },
    { "Asx",   'B' },
    { "Cys",   'C' },
    { "Asp",   'D' },
    { "Glu",   'E' },
    { "Phe",   'F' },
    { "Gly",   'G' },
    { "His",   'H' },
    { "Ile",   'I' },
    { "Lys",   'K' },
    { "Leu",   'L' },
    { "Met",   'M' },
    { "fMet",  'M' },
    { "Asn",   'N' },
    { "Pro",   'P' },
    { "Gln",   'Q' },
    { "Arg",   'R' },
    { "Ser",   'S' },
    { "Thr",   'T' },
    { "Val",   'V' },
    { "Trp",   'W' },
    { "Xxx",   'X' },
    { "OTHER", 'X' },
    { "Tyr",   'Y' },
    { "Glx",   'Z' },
    { "Sec",   'U' },
    { "Ter",   '*' },
    { "TERM",  '*' }
};

typedef struct singleinit {
    const char * key;
} SingleInit;

static SingleInit single_key_list [] = {
    { "pseudo"               },
    { "germline"             },
    { "rearranged"           },
    { "transgenic"           },
    { "environmental_sample" }
};

// constructor
CFeature_table_reader_imp::CFeature_table_reader_imp(void)
{
    // initialize common CFeature_table_reader_imp tables
    for (int i = 0; i < sizeof (feat_key_to_subtype) / sizeof (FeatInit); i++) {
        string str = string (feat_key_to_subtype [i].key);
        m_FeatKeys [string (feat_key_to_subtype [i].key)] = feat_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (qual_key_to_subtype) / sizeof (QualInit); i++) {
        string str = string (qual_key_to_subtype [i].key);
        m_QualKeys [string (qual_key_to_subtype [i].key)] = qual_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (orgref_key_to_subtype) / sizeof (OrgRefInit); i++) {
        string str = string (orgref_key_to_subtype [i].key);
        m_OrgRefKeys [string (orgref_key_to_subtype [i].key)] = orgref_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (genome_key_to_subtype) / sizeof (GenomeInit); i++) {
        string str = string (genome_key_to_subtype [i].key);
        m_GenomeKeys [string (genome_key_to_subtype [i].key)] = genome_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (subsrc_key_to_subtype) / sizeof (SubSrcInit); i++) {
        string str = string (subsrc_key_to_subtype [i].key);
        m_SubSrcKeys [string (subsrc_key_to_subtype [i].key)] = subsrc_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (orgmod_key_to_subtype) / sizeof (OrgModInit); i++) {
        string str = string (orgmod_key_to_subtype [i].key);
        m_OrgModKeys [string (orgmod_key_to_subtype [i].key)] = orgmod_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (bond_key_to_subtype) / sizeof (BondInit); i++) {
        string str = string (bond_key_to_subtype [i].key);
        m_BondKeys [string (bond_key_to_subtype [i].key)] = bond_key_to_subtype [i].subtype;
    }

    for (int i = 0; i < sizeof (site_key_to_subtype) / sizeof (SiteInit); i++) {