sabrinatoro
commented
1 year ago This is very confusing because 4 out of 5 of the "POLYGENIC" examples were associated with only one gene. That being said, using the predicate "biolink:gene_associated_with_condition" is broad enough. It means there is some association between the gene and the disease. Therefore, I think we can use it in all cases.
We could probably get more specific for the "MENDELIAN" one (using RO:0003303, causes condition). But I don't see an existing biolink predicate that represents it (it doesn't mean it doesn't exist)
I hope that helps!
kevinschaper
We have HPO's g2d file sitting in our data-cache now (thanks @iimpulse!) and we can swap it in for our OMIM g2d ingest
The file looks like (taken from a few places to get examples of all 3 association types)
ncbi_gene_id
No change necessary, just pass through as-is (and it will be mapped to HGNC later)
gene_symbol
Not used
association_type
values: 6573 MENDELIAN 621 POLYGENIC 8158 UNKNOWN
We should check with @sabrinatoro on predicate mappings.
my partial guess: MENDELIAN:
biolink:affects_risk_forUNKNOWN:biolink:gene_associated_with_conditionPOLYGENIC: ???disease_id
prefixes: 7194 OMIM 8158 ORPHA
Need to replace("ORPHA:", "Orphanet:") to match the curie in MONDO sssom
source
The source column values are
We should map these to infores as the primary_knowledge_source (maybe
infores:omimfor the medgen, then medgen as an aggregator? - plusinfores:orphanet)Also add aggregating_knowledge_source that includes
infores:monarchinitiative,infores:hpo-annotations, and probablyinfores:medgendownload & file path
The file is available in the
data/hpoa/genes_to_disease.txtin gs://monarch-ingest-data-cache already, it won't be available from the HPO site until the next release, but we should still add a commented out entry in download.yaml and make a second issue to enable the download later.