tooluniverse-cancer-variant-interpretation

What it does

Provides a comprehensive clinical interpretation of a somatic cancer mutation by transforming a gene symbol + variant (and optional cancer type) into an actionable, evidence-graded report. It prioritizes treatment options, checks for known resistance mechanisms, tailors to cancer type when provided, and cites each statement to the supporting tool/database.

How it works

Follows a multi-phase workflow:

• Phase 1: Gene disambiguation and ID resolution to Ensembl/Entrez/UniProt and OpenTargets descriptors; optionally resolves cancer type to an OpenTargets disease ID.
• Phase 2: CIViC-based evidence collection for the specific variant, including all gene variants, matching the input, and retrieving evidence items and variant details; if CIViC data is sparse, it falls back to literature or OpenTargets for supporting evidence.
• Phase 3: Mutation prevalence from cBioPortal across studies; computes distribution and frequency of the target variant and identifies co-mutants.
• Phase 4: Therapeutic associations from OpenTargets and ChEMBL (and FDA label data) to assemble approved and trial therapies, mechanisms, and combinations.
• Phase 5: Resistance mechanisms by checking known variants and pathway bypasses; includes secondary mutations.
• Phase 6: Clinical trials presence, phase, status, and eligibility cues.
• Phase 7: Prognostic impact and pathway context with literature and pathway data.
• Phase 8: Report synthesis with executive summary, clinical actionability score, prioritized treatment recommendations, and a completeness checklist.

Commands and file references are used exactly as described in the skill to query tools, resolve IDs, and fetch evidence.

When to use it

Apply when clinicians or researchers ask for treatment options for a specific cancer-variant pair (e.g., EGFR L858R in a given cancer type, BRAF V600E across cancers, or a resistance scenario like progression on osimertinib).

What it can touch

Interacts with tools such as MyGene_query_genes, UniProt_search, OpenTargets_get_target_id_description_by_name, OpenTargets_get_disease_id_description_by_name, UniProt_get_function_by_accession, civic_get_variants_by_gene, civic_get_variant, civic_get_molecular_profile, civic_tooling variants/examples, cBioPortal_get_cancer_studies, cBioPortal_get_mutations, OpenTargets_get_associated_drugs_by_target_ensemblID, and related drug/target queries. The exact tools and parameters are invoked per the Phase 0 parameter verifications and workflow.

Caveats

Results depend on external databases (CIViC, cBioPortal, OpenTargets, ChEMBL, FDA data) and their current entries; sparse CIViC data may require fallbacks to literature or OpenTargets for corroboration. The approach emphasizes an evidence-graded (T1-T4) output and explicit source citations for every statement.