Oncogenic/Predisposing EID update with pos/neg axis for disease causality

[arpaddanos]

In this figure from Alex Wagner, note how Predictive and Prognostic evidence have axes with opposing concepts, and a neutral term in the center.

Note how in CIViC for predictive evidence, approaching the center from the left yields "does not support sensitivity" annotation.

Updating the data model in CIViC for Predictive and Oncogenic EIDs would look as in this mock up below.

Note that this image also contains the suggestion that a third Evidence Direction be added, but that has not yet been agreed on, while the adoption of Wagner's proposed model of Predisposition/Protectiveness and Oncogenicity/Protectiveness for Predisposing and Oncogenic EIDs was agreed on during a Friday meeting held in Winter 22.

[arpaddanos]

Should we go with the term Oncogenic, or find a different term for this EID type, like we moved away from using Pathogenic for Predisposing Evidence type?

[ahwagner]

"Protectiveness" as an alternate predicate to "predisposition" seems natural and it easy to envision how such evidence might be derived (e.g. healthy elderly cohort analysis).

I think that variants supporting oncogenicity are acquired and selected for driving cancer growth and development. However, another predicate for "protectiveness" on the "oncogenic" axis is difficult to imagine and probably doesn't belong here.

[arpaddanos]

CIViC built into its data model that evidence was supposed to meet a bar before conclusion. The argument for introducing Inconclusive is to capture studies where neither supports or does not support was observed. So the idea that we should be cutting across all EID with a fundamental new change of Inconclusive Evidence Direction to capture evidence that gives no direction seems against the spirit of CIViC which was to have a higher bar for evidence. If there is no statement to make on the variant then it should not be curated. Just like we do not curate the observation of a variant in a tumor unless it meets the bar of interest in an SC-VCEP

[jsaliba10]

I do not think inconclusive should be part of the data model. For evidence to be placed in CIViC an interpretation should be needed. The authors in most cases have a question or hypothesis they are addressing, does the data then support or not support that question. If the study is underpowered and cannot draw a final conclusion, we can note that in a comment, make the evidence low star, and also even really consider whether the evidence should even be in civic. If the term were included, I can foresee an issue with curators applying this designation inconsistently as well.

[arpaddanos]

Oncogenicity is the Clinical Significance, Oncogenic is the EID and AID name

[arpaddanos]

The source queue could be used as a mechanism to mark and comment on papers that hold no content, instead of using the inconclusive Evidence Direction

[obigriffith]

Just thinking about the potential for variants with evidence type of oncogenic and significance of "protectiveness". We don't have obvious examples of this yet. But, one could imagine a passenger somatic variant that modulates or mitigates in some way the oncogenic properties of a driver variant (e.g., in the same gene, pathway, etc). This is more easy to imagine with common polymorphism some protective effect but I don't see why it couldn't occur (by chance) with a somatic passenger variant.

[malachig]

https://lucid.app/lucidspark/23ae4dc2-a014-4e02-b259-366208a47407/edit?viewport_loc=-1426%2C-769%2C5376%2C2928%2C0_0&invitationId=inv_8a4a7a13-a05a-4608-a4ff-fef0890aa56f#

[arpaddanos]

There is a flaw in the overall logic of the new Predisposing EID type, which has been discussed before. The EID type describes predisposition on two axis:

1. predisposition for protectiveness
2. predisposition for pathogenicity

therefore the Clinical Significance for Predisposing EIDs should be Protectiveness and Pathogenicity