Does DisGeNet propagate associations via disease ontology?

Question

[Yep]

Hi,

I was just wondering if DisGeNet included indirect associations for diseases that are ancestors of other diseases in the disease ontology (like what they call in Open Targets indirect evidence). I tried to look for some descriptions but it looks like there isn't any. Does anyone happen to know this? If they indeed did not incorporate this functionality, are there any pointers for code snippets that might be useful to propagate this manually (Python preferrably)? Thanks!

Answer 1

[Yep]

Copied from the response from the disgenet team:

To answer your question, no, we have not propagated the associations in DisGENET using an ontology. Curated associations come from annotations in data sources such as UniProt, Orphanet, etc, and we perform mappings of vocabulary and data integration. See more details here.

Nevertheless, we do have an "INFERRED" dataset that includes GDAs from the Human Phenotype Ontology, and GDAs inferred from VDAs reported by Clinvar, the GWAS catalog and GWAS db. We include GDAs from HPO in this "inferred" dataset because these gene-disease associations are not direct. Read more about it here. We called inferred GDAs coming from Clinvar because the "actual" annotation in Clinvar is for the variant - disease, but I do believe that these GDAs derived from Clinvar VDAs are "strong" GDAs, or valid. Finally, we called inferred GDAs derived from GWAS data because again, the original information from these two resources are variant-disease relationships, and we only map the variants to the closest gene, and then assign the gene to the corresponding phenotype/disease. And this might not be the best practice for the case of GWAs data.