Skin, Decoded
← Contents

Lesson 06 · Use it on the next one

How to read a genetics headline

4 minute read

01 WHICH CELL TYPE? 02 HOW BIG IS THE EFFECT? 03 CODING OR REGULATORY? 04 REPLICATED? 05 PREPRINT? “Scientists find the gene for ageing skin” EFFECT SIZE NOT REPORTED CELL TYPE NOT NAMED ONE COHORT, NO REPLICATION
Anatomy of a claimFive questions, applied to the headline

“Scientists find the gene for...” is almost always a compression of something more careful. These five questions recover the missing detail.

01

Which cell type?

A variant’s effect can exist in one cell type and nowhere else — the ETS2 story is exactly this. If the study does not name a cell type, its claim is probably about an average across tissue, which can hide the effect that matters.

02

How big is the effect?

An odds ratio of 1.03 across a population is a real statistical signal and a meaningless personal prediction. Ask for the effect size, not the p-value.

03

Coding or regulatory?

Coding variants change a protein and tend to have large effects. Regulatory variants change how much of a protein is made, and usually have small ones. Both are real; they license very different claims.

04

Has it been replicated?

A single cohort finding is a hypothesis with good PR. Look for independent replication, and for a mechanism — not just a correlation in one dataset.

05

Is it a preprint or peer-reviewed?

Preprints are how science moves fast now, and they are not wrong by default. But they have not been through review, and the wording of a press release rarely distinguishes the two. AlphaGenome 2’s own paper is a preprint; the original model’s paper is in Nature.

The one distinction worth keeping

Heritability is not determinism. A trait being 60% heritable means that, across a population, 60% of the variation between people is associated with genetic differences. It says nothing about how much of one person’s outcome is fixed, and nothing about whether anything can be changed.

And a boundary

A model that predicts molecular activity is not a diagnosis, a treatment plan, or a product recommendation. The moment a claim jumps from ’this variant changes expression in this cell type’ to ’therefore you should buy this’, the evidence has been left behind.

Check yourself

A study reports a variant with an odds ratio of 1.04 for a skin condition. What does that mean for one person?

“Heritability of 60%” means: