Notes from the interpretation layer.
Clinical genomics, variant interpretation, pharmacogenomics and polygenic risk — written for the people who read the reports and sign them.
Beyond CYP2D6: SMN1, GBA and PMS2, where short reads fail
SMN1, GBA and PMS2 sit beside near-identical pseudogenes. Three genes, three different ways short reads fail — and what has to be declared.
Read articleWhy a normal result is not a negative: copy number, pseudogenes and residual risk
A normal result only rules out what was assessed. Why a report has to state what was evaluated, what was filtered and what was not looked at.
Read articleCYP2D6: the false normal that arrives looking like a result
A pipeline that does not treat the CYP2D6 locus specifically does not return an error. It returns *1/*1 — a poor metaboliser reported as normal, with no alarm of any kind.
Read articleSilent false normals: the error nobody sees
Why standard NGS pipelines can return normal results for regions they never actually resolved — and why an unresolved region must be declared, not omitted.
Read articleACMG classification shouldn’t be a black box
How a Bayesian ACMG interpretation model with an auditable ledger lets a geneticist review the reasoning, not just the verdict.
Read articleA polygenic score is only as good as its ancestry
Why a PRS without explicit ancestry inference is a number without units — and how calibration changes what a percentile means.
Read articleWant the underlying documents?
The technical notes and validation dossiers behind these articles are available for your quality system.