Functional Evidence curation guidance is poor

[ahwagner]

Our Functional Curation Practices section has a subsection entitled "Functional Evidence from Clinical Trials". This sub-section needs to be moved or entirely reworked, as it seems to be describing the curation of Predictive evidence, instead of Functional. As this section is being reviewed, some additional attention should be paid to resolve the typos throughout.

In addition, we should provide guidance on how this type of functional classification differs from the common "Variant X is gain-of-function / loss-of-function for gene Y" type of functional statement. CIViC functional evidence (and later, functional assertions?) includes the notion of a disease type and variant origin, which is atypical for this type of statement in other resources. I believe the argument was that this is because the notion of functional variation is with respect to phenotypic function, where cells gain or lose clinically relevant (disease-linked) characteristics due to the variant. If so, we should state as much in our curation guidelines; and if not, we should consider revisiting the data model for Functional evidence types to omit variant origin and disease attributes.

[ahwagner]

Functional evidence is intended to capture whether or not a variant causes a change of biological function of a given gene.

When applying this evidence to broader assertions, care should be taken to consider if the associated gene would likely be expressed in a wild type state, before applying the predicted effect from functional evidence.

We have agreed that functional evidence is intended to be evaluated in contexts beyond the experimental conditions where it was originally observed, but also that CIViC is a cancer-focused resource and should contain functional characterizations that are of relevance to cancer genes. To accommodate this, we will not remove the notion of "disease" from the functional evidence objects, but we will constrain it to the notion of "cancer", indicating that this functional evidence must be considered in the context of relevance to this class of disease.

We have agreed to drop the notion of variant origin, and default to N/A, similar to the currently disallowed field of drug.

We also discussed the possibility of a future evidence type, oncogenic, describing the role of the variant in similarly broad (disease-agnostic) terms leading to the origination or growth of cancers. These two evidence statements would likely be combined in CIViC Assertions for oncogenicity, but allow for functional evidence to also be used in other Assertion types (predictive, diagnostic, prognostic).

Action items:

• Change the functional evidence entry form to match these changes (specific issue to be created in civic-client)
• Editorially review existing functional evidence with non-cancer type and ensure that this content is captured in evidence descriptions
• Update civic-docs to capture the above decisions and intent of these records

[kkrysiak]

Related pull request. #55

[kkrysiak]

Related to griffithlab/civic-client#1455