Add hgnc family info into MyGene.info

[kevinxin90]

hgnc contains gene group info:
https://www.genenames.org/data/genegroup/#!/group/567

[andrewsu]

This file has the gene-to-family links: http://ftp.ebi.ac.uk/pub/databases/genenames/hgnc/csv/genefamily_db_tables/gene_has_family.csv

hgnc_id	family_id
11148	3
3960	3
3961	3
3477	1963
4621	1963
4622	1963
9962	1963
16719	1963

This file has the name and metadata for each HGNC family: http://ftp.ebi.ac.uk/pub/databases/genenames/hgnc/csv/genefamily_db_tables/family.csv

id	abbreviation	name	external_note	pubmed_ids	desc_comment	desc_label	desc_source	desc_go	typical_gene
1296		TIR domain containing	NULL	NULL	NULL	NULL	NULL	TIRAP
75	ZDBF	Zinc fingers DBF-type	NULL	NULL	NULL	NULL	NULL	ZDBF2
302	CLCN	Chloride voltage-gated channels	NULL	NULL	NULL	NULL	NULL	CLCN1
228	HCRTR	Hypocretin receptors	NULL	NULL	NULL	NULL	NULL	HCRTR1

The combination of these two files should be what we initially add to mygene.info records for each human gene.

[colleenXu]

It looks like there's already some gene group info in MyGene.info that is shown as node attribute in BTE.

BTE brings in interpro info using this code.

You can see that some gene family info is included when you look at that field in mygene like this, as well as maybe some info that's for specific domains of the protein?: https://mygene.info/v3/query?q=CDK2&fields=interpro.desc%2C%20type_of_gene

[jal347]

I made the plugin for the hgnc_family. The main branch contains the manifest and parser. v2 branch contains the advanced plugin. If we use the advanced plugin can someone check if I did the mapping correctly? thanks. https://github.com/jal347/hgnc_family

[jal347]

This is a quick summary of the current hgnc mapping. The total number of hgnc_id data points is 29872. The total number of unique hgnc_ids is 24952. Out of the 24952 hgnc_ids 24895 were mapped while 57 could not be queried in mygene.info. The number of 1-1 hgnc_id to family_id is 21100 and 1-n mapping is 3852. 1-7 is the max hgnc_id to family_id mapping. An example is shown below and more detailed information of the 1-n mappings.

{
    "_id": "6624",
    "hgnc_genegroup": [
        {
            "id": "3",
            "abbr": "FSCN",
            "name": "Fascin family",
            "comments": "",
            "pubmed": [
                21618240
            ],
            "typical_gene": "FSCN1"
        }
    ]
}

[zcqian]

(I remember commenting on this yesterday where did it go ...)

added here: https://github.com/biothings/mygene.info/tree/add_hgnc_family/src/plugins/hgnc_family

@newgene should PubMed ID be of type long and not indexed? This is what we have in other sources in MyGene.

[newgene]

@zcqian (RE: pubmed) good catch. Let's keep this field the same as other sources then.

[zcqian]

@newgene should we index the PubMed ID field?

[newgene]

No for now, we can change if later we do need to query it.