Reading the genome

A genome is a string of letters about three billion long. Picture it printed out — one fat phone book per chromosome, a whole shelf of them, every page solid with four letters: A, C, G, T. To “read” your genome is to spell that whole string out, letter by letter, and write it down. The first time anyone did this for a human, it took years and cost hundreds of millions. Now the same job costs less than a nice dinner and finishes overnight. That collapse in price changed everything — and it created a new way to be fooled.

What a sequencer actually does

A DNA sequencing machine has one narrow job: take a DNA sample and tell you the order of its letters. Nothing more. It does not know what the letters mean. It does not know whose genome it is. It just spells.

The trick that made it cheap is reading in parallel. Early methods spelled out one stretch at a time, slowly. Modern machines chop the genome into millions of short fragments and read all of them at once, then a computer stitches the overlapping pieces back into the full string — the way you could rebuild a shredded book by matching torn edges. Doing a million small reads at the same time is what dropped the price by a factor of millions in about two decades.

So the output is a text file: your three billion letters, in order. That file is the input to everything else. The hard part — and the honest part of this lesson — is what you can and cannot conclude from it.

A letter changed is not a fate sealed

Here is the thing people get wrong. A genetic test finds places where your letters differ from the common version. Most differences mean nothing. A few sit inside a gene and change the protein it builds. Of those, some nudge your risk of a disease up or down. We call these risk variants — and the word “risk” is doing the heavy lifting.

A risk variant does not flip a switch from “well” to “ill”. It shifts the odds. Most people who carry it never get the disease. Some people without it get the disease anyway. The variant moved a probability; it did not write your future.

Why only the odds? Because most diseases are the work of many genes plus a lifetime of environment — what you eat, breathe, catch, and inherit beyond that one letter. A single variant is one voice in a large, noisy committee. It can lean the vote. It cannot decide it.

A worked example: 8 to 20

Make it concrete with round, illustrative numbers. Suppose a condition affects 8 in 100 people in the general population. That is the baseline — the chance for someone with the ordinary version of the gene. Now a test tells you that you carry a risk variant that, the studies say, more than doubles your chance — pushing it to about 20 in 100.

So the headline could read: “this gene more than doubles your risk.” True — going from 8 to 20 is about two and a half times. But walk it out in people.

Take 100 carriers of the variant. About 20 develop the condition over their lives. That leaves 80 of the 100 carriers who never get it — the large majority, despite the “scary” gene. Now take 100 non-carriers. About 8 develop it. So 8 of 100 people without the gene still get the condition.

Stack the two side by side. Carriers: 20 affected, 80 fine. Non-carriers: 8 affected, 92 fine. The gene is real — 20 is clearly more than 8. And it is nowhere near destiny — being a carrier, the single most likely outcome is still that nothing happens. A test that reports “you carry this variant” has told you the odds moved from 8 to 20. It has not told you which of the two groups you will land in.

That gap — between “the odds shifted” and “you are doomed” — is where most genetic-test fear lives, and where most genetic-test marketing earns its money.

Reading it honestly

So what is a genome read good for? Plenty. A handful of conditions really are caused by a single broken gene — there, a sequence can be close to a yes-or-no answer, and that is genuinely useful. For the common diseases, a read can flag that your odds run higher than average, which can be worth knowing. It is a weather forecast, not a calendar entry. “70% chance of rain” is real information and you might take an umbrella — but it is not a promise the sky will open.

Two cautions worth keeping. First, the studies behind a risk number were done on particular groups of people; a variant measured in one population may carry a different risk in another, or none. Second, a probability for a thousand people is not a verdict for you — the same number that means “20 in 100” cannot tell you whether you are one of the 20.

None of this is medical advice. The point is narrower and sturdier: a sequence is a string of odds, not a sentence handed down.

On the whole

We can now spell out the entire instruction set of a living thing for the price of a meal — a genuinely strange power, arrived at almost casually through machines reading a million fragments at once. But the letters are quieter than the headlines about them. They tilt probabilities inside a system far larger than any single gene: the rest of the genome, the body around it, the world the body moves through. You are not the readout of one variant. You are the running average of a vast committee of them, voting alongside everything that ever happened to you — which means the genome can describe the odds you were dealt without ever being able to name the hand you will play.