--$Revision: 1.31 $ 
--********************************************************************** 
-- 
--  NCBI Entrezgene 
--  by James Ostell, 2001 
--   
--  Generic "Gene" object for Entrez Genes 
--    This object is designed to incorporate a subset of information from 
--    LocusLink and from records in Entrez Genomes to provide indexing, 
--    linkage, and a useful summary report in Entrez for "Genes" 
-- 
--********************************************************************** 
 
NCBI-Entrezgene DEFINITIONS ::= 
BEGIN 
 
EXPORTS Entrezgene, Entrezgene-Set, Gene-track, Gene-commentary;
 
IMPORTS Gene-ref FROM NCBI-Gene 
    Prot-ref FROM NCBI-Protein 
    BioSource FROM NCBI-BioSource 
    RNA-ref FROM NCBI-RNA 
    Dbtag, Date FROM NCBI-General 
    Seq-loc FROM NCBI-Seqloc 
    Pub FROM NCBI-Pub; 
 
--******************************************** 
-- Entrezgene is the "document" indexed in Entrez 
--  and presented in the full display 
-- It also contains the Entrez ID and date information 
--******************************************* 
Entrezgene ::= SEQUENCE { 
    track-info Gene-track OPTIONAL , -- not in submission, but in retrieval 
    type INTEGER {                   -- type of Gene
        unknown (0) ,
        tRNA    (1) ,
        rRNA    (2) ,
        snRNA   (3) ,
        scRNA   (4) ,
        snoRNA  (5) ,
        protein-coding (6) ,
        pseudo  (7) ,
        transposon  (8) ,
        miscRNA  (9) ,
        ncRNA (10) ,
        other (255) } ,
    source BioSource , 
    gene Gene-ref ,                     -- for locus-tag see note 3
    prot Prot-ref OPTIONAL , 
    rna RNA-ref OPTIONAL , 
    summary VisibleString OPTIONAL ,    -- short summary 
    location SEQUENCE OF Maps OPTIONAL,
    gene-source Gene-source OPTIONAL ,             -- NCBI source to Entrez 
    locus SEQUENCE OF Gene-commentary OPTIONAL ,   -- location of gene on chromosome (if known)
                                                   -- and all information about products
						   -- (mRNA, proteins and so on)
    properties SEQUENCE OF Gene-commentary OPTIONAL , 
    refgene SEQUENCE OF Gene-commentary OPTIONAL , -- NG for this? 
    homology SEQUENCE OF Gene-commentary OPTIONAL , 
    comments SEQUENCE OF Gene-commentary OPTIONAL ,
    unique-keys SEQUENCE OF Dbtag OPTIONAL ,              -- see note 3
    xtra-index-terms SEQUENCE OF VisibleString OPTIONAL , -- see note 2
    xtra-properties SEQUENCE OF Xtra-Terms OPTIONAL ,     -- see note 2
    xtra-iq SEQUENCE OF Xtra-Terms OPTIONAL,              -- see note 2
    non-unique-keys SEQUENCE OF Dbtag OPTIONAL }

Entrezgene-Set ::= SET OF Entrezgene

Gene-track ::= SEQUENCE { 
    geneid INTEGER ,     -- required unique document id 
    status INTEGER {
        live (0) ,
        secondary (1) ,   -- synonym with merged
        discontinued (2), -- 'deleted', still index and display to public
        newentry (3)      --  for GeneRif submission
    } DEFAULT live ,
    current-id SEQUENCE OF Dbtag OPTIONAL , -- see note 1 below
    create-date Date ,   -- date created in Entrez 
    update-date Date ,   -- last date updated in Entrez 
    discontinue-date Date OPTIONAL } --
 
Gene-source ::= SEQUENCE { 
    src VisibleString ,                -- key to the source within NCBI locuslink, Ecoli, etc 
    src-int INTEGER OPTIONAL ,         -- eg. locuslink id 
    src-str1 VisibleString OPTIONAL ,  -- eg. chromosome1 
    src-str2 VisibleString OPTIONAL ,  -- see note 3
    gene-display BOOLEAN DEFAULT FALSE ,  -- do we have a URL for gene display? 
    locus-display BOOLEAN DEFAULT FALSE , -- do we have a URL for map/locus display? 
    extra-terms BOOLEAN DEFAULT FALSE }   -- do we have a URL for extra indexing terms? 
 
Gene-commentary ::= SEQUENCE { 
    type INTEGER {            -- type of Gene Commentary
        genomic (1) ,
        pre-RNA (2) ,
        mRNA (3) ,
        rRNA (4) ,
        tRNA (5) ,
        snRNA (6) ,
        scRNA (7) ,
        peptide (8) ,
        other-genetic (9) ,
        genomic-mRNA (10) ,
        cRNA (11) ,
        mature-peptide (12) ,
        pre-protein (13) ,
        miscRNA  (14) ,
        snoRNA  (15) ,
        property  (16) , -- used to display tag/value pair
	                 -- for this type label is used as property tag, text is used as property value, 
	                 -- other fields are not used.
        reference (17), -- currently not used             
        generif (18), -- to include generif in the main blob             
        phenotype(19), -- to display phenotype information
        complex (20), -- used (but not limited) to identify resulting 
                      -- interaction complexes
        compound (21), -- pubchem entities

        ncRNA (22), 
        gene-group (23), -- for relationship sets (such as pseudogene / parent gene)
        comment (254) ,
        other (255) } ,
    heading VisibleString OPTIONAL ,      -- appears above text 
    label VisibleString OPTIONAL ,        -- occurs to left of text
                                          -- for protein and RNA types it is a name
					  -- for property type it is a property tag  
    text VisibleString OPTIONAL ,         -- block of text 
					  -- for property type it is a property value  
    accession VisibleString OPTIONAL ,    -- accession for the gi in the seqloc, see note 3
    version INTEGER OPTIONAL ,    -- version for the accession above
    xtra-properties SEQUENCE OF Xtra-Terms OPTIONAL , -- see note 2
    refs SEQUENCE OF Pub OPTIONAL ,       -- refs for this 
    source SEQUENCE OF Other-source OPTIONAL ,    -- links and refs 
    genomic-coords SEQUENCE OF Seq-loc OPTIONAL , -- referenced sequences in genomic coords
    seqs SEQUENCE OF Seq-loc OPTIONAL ,           -- referenced sequences in non-genomic coords
    products SEQUENCE OF Gene-commentary OPTIONAL ,
    properties SEQUENCE OF Gene-commentary OPTIONAL ,
    comment SEQUENCE OF Gene-commentary OPTIONAL ,
    create-date Date OPTIONAL ,   
    update-date Date OPTIONAL } 
 
Other-source ::= SEQUENCE { 
    src Dbtag OPTIONAL ,                -- key to non-ncbi source 
    pre-text VisibleString OPTIONAL ,   -- text before anchor 
    anchor VisibleString OPTIONAL ,     -- text to show as highlight 
    url VisibleString OPTIONAL ,        -- if present, use this URL not Dbtag and datbase 
    post-text VisibleString OPTIONAL }  -- text after anchor 


Maps::= SEQUENCE {
        display-str VisibleString ,
        method CHOICE {
            proxy VisibleString ,  --url to non mapviewer mapviewing resource
            map-type ENUMERATED {  -- units used in display-str to query mapviewer 
                    cyto (0) ,
                    bp (1) ,
                    cM (2) ,
                    cR (3) ,
                    min (4)}}}
                        
Xtra-Terms ::= SEQUENCE {  -- see note 2
    tag VisibleString ,
    value VisibleString }

END 

--********************************************************************** 
-- 
--  Comments, notes, etc.
--   
--  1)  Ignored unless status = secondary.  This is where gene_ids (db = "GeneID")
--      are placed toward which the interface will direct users.  It is also
--      available for placing other source-db specific tags (i.e., db = "LocusID").
--
--  2)  These 'xtra' objects are for submitting data for Entrez indexing
--      that might not fit anywhere in the Entrezgene specification but
--      are considered by the data source submittor to be important.
--          xtra-index-terms is any string.
--          xtra-properties are tag/value pairs of properties/feilds as
--              defined in the Entrez database (i.e.: UNIGENE/Hs.74561)
--          xtra-iq are tag/value pairs of Entrez database/UID as defined
--              in the Entrezgene indexing code (i.e.: NUCLEOTIDE/20270626)
--
--  3)  Locus-tag and src-str2 are expected to be unique per organism (tax_id).
--      Protein accessions and the tag-value pairs in unique-keys
--      are expected to be unique over all organisms.
--**********************************************************************