How can a gene point at the cause of a disease?

You have been told that correlation is not causation. It is true, and it is usually where the conversation stops, which is a shame — because the interesting question is what you can actually do about it.

Here is the problem in its everyday form. People with a certain disease also tend to have some other trait. Does the trait cause the disease? Does the disease cause the trait? Or is there a third thing causing both?

Normally the only way to settle it is an experiment: change one thing deliberately, hold everything else still, see what happens. In humans, for most interesting questions, that experiment would be unethical, impossible, or both.

Genetics offers a partial way around this. Not a perfect one. But a real one.

Why genes are different

Your genetic variants were set before you were born, and they were dealt out essentially at random when your parents' DNA was shuffled.

That last part is what makes them useful. Your genes were not caused by your diet, your income, your neighbourhood, or your disease. The arrow can only point one way: genes can affect your health, but your health cannot reach back and change the gene you were born with.

So when a specific genetic variant is consistently more common in people with a disease, the usual tangle gets much simpler. The disease did not cause the variant. The variant came first.

That still is not proof. But it is a great deal better than noticing that two things happen together.

How it is done at scale

Researchers run something called a genome-wide association study, usually shortened to GWAS.

Take a large group of people with a disease and a large group without. Read variants across the whole genome in everyone. Then, for every position, ask: is this variant more common in the disease group?

Almost all of them are not. A handful are, by much more than chance would explain. Those are your leads.

Numbers matter enormously here. With a few hundred people you will find patterns that are pure noise. These studies now run to hundreds of thousands of participants, precisely because the real signals are small and the false ones are everywhere.

The honest limits

Three things a genetic association does not tell you, and it is worth being strict about them:

It does not tell you the mechanism. You learn that a region of DNA matters, not what it does or how.

It does not tell you it is the gene you think it is. Variants sit near genes, and "near" is not "controls." Pinning a signal to the responsible gene is separate work, and it is often wrong on the first attempt.

It does not tell you the effect is large. A variant can be real, replicated, and still shift your risk by a few percent.

Genetic evidence is a strong hint about where to look. It is not a conclusion, and anybody who presents it as one is overselling.

Try this

1. Go to Open Targets: https://platform.opentargets.org
2. Search for any disease that interests you.
3. Look at the genetic association score for the top targets — that column is mostly built from GWAS results.
4. Now compare two targets: one with high genetic evidence, one with high literature evidence (lots of papers written about it).

They are often different genes. Which one would you bet a decade of work on — the one the data points at, or the one everyone is already writing about?

One question to take away

Genetics tells you the arrow points from gene to disease. It does not tell you how long the arrow is. Why does that distinction matter before anyone spends money on a drug?