Quality control on gene-disease assocs

[cmungall]

Many of these are wrong
https://monarchinitiative.org/disease/MONDO:0007947#genes

@kshefchek to fill in details

[cmungall]

One of these are due to monarch-initiative/mondo#560, but not all

[kshefchek]

Which genes? LTBP2, CEP152, FBN2 are not associated to Marfan in the latest ingest.

[kshefchek]

I'm going to assume the only gene we want to display FBN1.

For ClinVar:
We updated the ingest over the summer to that removes some gene-disease associations
See https://beta.monarchinitiative.org/disease/MONDO:0007947#genes and #593
This will be fixed in the next data release.

For CTD:
I'll propose that CTD and is_marker_for associations should be removed from the app until we can better display these. We can add a filter for a relation so these will still be retrievable via biolink and our tsv downloads. This can be updated on production.

[pnrobinson]

?There are multiple problems and this should be discussed on zoom asap

[kshefchek]

Theres not too much to do until we can push the latest data and version of mondo. In the meantime I've removed CTD is_marker_for associations on production.

[cmungall]

I still see the is_marker_fors?

When will the next data release be? Should we patch in the interim?

[kshefchek]

I realized that filtering on relation is not always reliable, SciGraph/golr-loader#35

I think the better solution is to require one more sources for each association, eg is_defined_by(omim OR omia ...), our sources and counts are here. Happy to specify omim as a requirement for all associations.