Represent CancerHotspots using CohortAlleleFrequencyStudyResult

[korikuzma]

Submitter Name

Katherine, Mrinal, + Kori

Submitter Affiliation

No response

Project Details

We'd like to represent Cancer Hotspots data using the latest VA-Spec release, in particular the CohortAlleleFrequencyStudyResult model.

Required Skills

Python
VA-Spec

[Mrinal-Thomas-Epic]

Goal

Represent Cancer Hotspots data using the latest VA-Spec release, in particular the CohortAlleleFrequencyStudyResult model

Plan

Produce a nested structure of CohortAlleleFrequencyStudyResult. The highest level will have a focus of a categorial variant at a particular amino acid position. This will have subCohortFrequencies for each observed amino acid change at that position. This will have subCohortFrequencies for each cancer type that amino acid change has been observed in. We will integrate this into the evidence-normalizer.

Tasks

• Update VA-spec to support categorical variation as the focus of CohortAlleleFrequencyStudyResult @korikuzma
• Format CancerHotspots to GKS models
  • Group rows within CSV that apply to the same amino acid position @Mrinal-Thomas-Epic
  • Produce categorical variation for amino acid change location
  • Produce ProteinSequenceConsequence for specific amino acid changes
  • Create a DataSet objects @kattaylor511
    • Investigate if we need to subdivide datasets
  • Produce cohorts @kattaylor511
    • Represent cancer type with MappableConcept
  • Produce StudyResult with counts
  • Produce NDJSON file with formatted StudyResult lines

[korikuzma]

For now, we're not going to annotate with VRS IDs. That is a future task that will be done in the future

[korikuzma]

Evidence Normalizer updates by @Mrinal-Thomas-Epic and @kattaylor511 : cancervariants/evidence-normalization#71