Scout: OMIM-AUTO mising crucial genes

[ViktorHy]

I cannot make sense of how OMIM-AUTO is created.

With the latest OMIM update (https://www.omim.org/statistics/geneMap) There should be 4333 genes with matching phenotype and mutation. OMIM-AUTO gene panel created through scout update omim contain 4030 genes. Regardless of how I try to filter the OMIM-files from the api I keep coming to the same conclusion, there should be 4333 genes.

For instance, RAD51D is missing in OMIM-AUTO. It clearly has a matching phenotype on OMIM. It does also exist in the scout database, for both hg19 and hg38.

Please help a confused soul

[northwestwitch]

Hi! We haven't updated genes, diseases and OMIM-AUTO panel on our production server yet, but I'll give it a try on my local instance!

[northwestwitch]

Yes, I also end up with an OMIM panel of 4030 genes. I'll check what happens in the case of RAD51D!

[dnil]

:+1: Hm, RAD51D qualifies rather as a susceptibility gene, so I can imagine there may be some funky setting somewhere on it not clearly indicating it as a morbid one. At one point we didn't include them at all, and especially not the pharmacophenos, but now I think the filter is simply for "established" and "provisional". Let's see!

[northwestwitch]

Do you want to take over this one @dnil ?

[dnil]

Well, not really: I would rather you keep using my key - thats more profitable.. 😜

But, I think this may be exactly that issue: RAD51D is indeed classified "only" as a susceptibility gene, which we have chosen not to include in the OMIM-AUTO panel. Anyone is naturally welcome to include it on their own panels - it is loaded as a gene, and a phenotype.

The ones we load onto OMIM-AUTO are the established and provisional ("?"-entries), not the "{}" risk/susceptibility or "[]" non disease entires of OMIM.

[dnil]

We have this status map that effectively gets checked when genes are added to OMIM-AUTO:

OMIM_STATUS_MAP = {"[": "nondisease", "{": "susceptibility", "?": "provisional"}
...
phenotype_status = OMIM_STATUS_MAP.get(phenotype_info[0], "established")
...
def get_omim_panel_genes(genemap2_lines, mim2gene_lines, alias_genes):
...
status_to_add = set(["established", "provisional"])
...
if not phenotype_info.get("status") in status_to_add:

[dnil]

Might that explain your count difference @ViktorHy?

[dnil]

I think it just might:

While there is bound to be a lot of overlap due to N-to-1 1-to-M and N-to-M connections between disease and gene, the ballpark of some 4300 genes with established+susceptibility on the top and some 500 susceptibility genes may well fit, right?

[dnil]

In particular RAD51D does not seem to have an established disease causation logged, and will so not be included, whereas e.g. BRCA1 has one for Fanconi - while it is also "just" {susceptibility} with regard to BRC - and is so included. I hope this brings some balm to the wounds on the confused soul, and feel very free to suggest e.g. a toggle that includes susceptibility genes - or make a new OMIM-S-AUTO or such that include them, if you want them?

[ViktorHy]

Thanks! This is indeed what separates them. I need to check with our clinicians to see if OMIM-S-AUTO would be of interest.