Specify priority/evidence level for candidate genes

[allasm]

...at least for candidate genes manually specified by name only; priority/evidence as in "5 star rating" or "not specified"/"little evidence"/"strong evidence" or [0..1] score

[buske]

This should also specify whether a candidate gene was manually curated or automatically prioritized.

[Relequestual]

Decipher stores two data points for this. Pathaginicty (class 1-5 with 4 including 4a and 4b), and Contribution (full, partial, uncertian, none). I'm not sure how this would fit in with our data, or what sort of data others have for this.

[buske]

@Relequestual, is that at the gene level, or at the variant level?

PhenomeCentral currently stores pathogenicity for variants (1-5), and evidence for candidate genes (3 binary items: related to mechanism, candidate pathogenic variant, known disease-associated).

[mellybelly]

@nlwashington, @AAMargolin, @wakibbe This is an example of the curation cats that I think we need to herd in GA4GH. @pgaudet can help with curation needs inventorying.

I think Matchmaker should consult the G2P group on this.

[pgaudet]

The pathogenicity score are interesting, but there is a lot more data in the literature we should make use of ! Admittedly it's more work.

[Relequestual]

@buske Decipher can do pathogenicity 1-5.

As for evidence for candidate gene, Decipher has morbid ids for genes, plus the DDD created DDG2P genes list (which contains Yes or Possible).

@mellybelly @pgaudet I'm not sure what you mean by this. From my understanding, we're talking about what our databases currently store. I've been told a 1-5 pathogenicity rating is pretty much the accepted standard. Is this not the case?

[Relequestual]

There has been a lot of discussion around this in the community! I think it best to remove it from milestone 1.1. Much more discussion is needed! For example how does each system currently record this property?

[buske]

+1

Just FYI, within PC, gene-level rationale/strategy can be listed as one or more of:

• Sequencing
• Del/dup
• Familial mutation/disease
• Common mutations

Variant-level details include:

• ACMG interpretation (1-5)
• Evidence, one or more of:
  • Rare
  • Predicted damaging by computational methods
  • Reported in other affected individuals

[Relequestual]

We have similar within DECIPHER. Regarding "chatter", specifically I mean the recent paper on new ACMG standards for interpretation of sequence variants.

https://www.acmg.net/docs/standards_guidelines_for_the_interpretation_of_sequence_variants.pdf

[jxchong]

We have

Benign Likely Benign VUS Suspected Pathogenic Likely Pathogenic Pathogenic