Scope more granular clinical significance for somatic variants in ClinVar

[ireneisdoomed]

As the data team we want to investigate the new annotation about the clinical impact somatic variants in ClinVar because this information is key to interpret ClinVar t/d associations.

Background

@apriltuesday has summarised new changes in the ClinVar data download, that affect both the structure of the XML and its content. ClinVar will support the existing format until March 2024.

We want to scope how we want to accommodate the more granular somatic clinical impact terms:

Somatic clinical impact terms: AMP/ASCO/CAP Tiers (PMID 27993330)
- Tier I - Strong
- Tier II - Potential
- Tier III - Unknown
- Tier IV - Benign/Likely benign

Assertion types for clinical impact:

- For therapeutic:
    - sensitivity/response
    - resistance
    - reduced sensitivity

- For diagnostic:
    - supports diagnosis
    - excludes diagnosis

- For prognostic:
    - better outcome
    - poor outcome

Oncogenicity terms: ClinGen/CGC/VICC recommendation (PMID 36063163)
- Oncogenic
- Likely oncogenic
- Uncertain significance
- Likely benign
- Benign

(from https://github.com/EBIvariation/CMAT/issues/396).

This is a mock from ClinVar of how significance could be reported after the changes, with multiple possible assertions per variant/disease:

Tasks

The clinical impact is how we score evidence from ClinVar. For the next meeting with EVA we need to determine

• [ ] new structure to report clinicalSignificances - should we nest all 3 assertions? We need to unify this field to report impact for germline and somatic variants
• [ ] revisit scoring system - should we consider all assertion types?
• [ ] evaluate how we can use the role of the variant in cancer

It'd be good to gather ideas for the next meeting on February 26