Pharmacogenomics (PGx)

Laboratories that test drug-metabolism genes validate their star-allele calling against the GeT-RM reference materials — public Coriell cell lines whose consensus diplotypes are published by the CDC's Genetic Testing Reference Materials program. Aver compares your calls against that published consensus, gene by gene, and records exactly what was compared.

The consensus table (consolidated V4, 363 samples) is fetched from the CDC and verified against a recorded SHA-256 before use. Consensus values are reproduced verbatim — including complex CYP2D6 structural calls like *3/(*68)+*4 — and the comparison does one thing only: unordered allele-set equality (*2/*1 equals *1/*2; nothing else is interpreted). Every row shows both strings, so the judgement is checkable by eye.

Usage

Your calls as JSON ({"CYP2D6": "*1/*4", ...}) or TSV (gene<TAB>diplotype):

aver pgx compare --sample NA12878 --calls my_calls.json \
  -o pgx-concordance.json

Each gene is reported as one of:

Status	Meaning
match	your diplotype equals the consensus (unordered alleles)
mismatch	it does not — both strings are shown
not_characterized	you called a gene the consensus does not cover for this sample
not_called	the consensus covers a gene you did not call

--fail-on mismatch exits non-zero for automation. The JSON output records the consensus set's version, URL and SHA-256 alongside every row, so the document is self-describing evidence.

Sample names must be GeT-RM-characterised Coriell IDs; anything else aborts, and the error never echoes the name. Aver only processes public reference materials.