Your genome, from scratch

Your DNA is a sequence of four letters —A, C, G and T— repeated three billion two hundred million times.

Between you and any other person, more than 99% of those letters are identical. Everything that sets you apart lives in the rest. Each position where people differ is called a variant.

Those letters are not loose: they come packed into 23 pairs of chromosomes —46 in all— and of each pair you inherit one from your mother and one from your father. That is why at every position you have two letters, not one.

When the two copies are the same it is called homozygous; when they are different, heterozygous. All the genetics that comes afterwards rests on that difference.

You do not inherit loose letters: you inherit whole stretches of chromosome. Before passing you a copy, each parent crosses their own two copies and assembles a new mixture. That process is called homologous recombination, and it happens in every generation.

Because the cut points fall in different places in each child, you share 50% with a sibling on average —the pairs measured range from 37% to 61%— and a relative's test does not replace your own.

And that is why it is possible to read where each stretch comes from: ancestry compares those blocks against reference populations. In Argentina the panels of Indigenous populations are smaller, and that component is estimated with less precision. It is not a defect in your sample: it is a limit of the science available today.

Most variants are the trace of the histories of entire populations, not individual errors.

Three things make them common, and only the first has to do with them being useful for anything. Look at the shape of each column: one rises and stays, another zigzags with no direction until it disappears, and the third jumps the day a group splits off.

The clearest case is lactose: tolerating it as an adult is the new variant, and it became common only in the populations that raised livestock.

A gene is a stretch of those letters with an instruction inside: almost always, how to build a protein. In a Mendelian condition the result depends on a single gene. They are uncommon, but when the variant is present the effect is decisive and the arithmetic is exact.

The most common form is recessive: both copies must carry the variant, which means both members of the couple have to be carriers. With a single copy you are a healthy carrier —no symptoms and no consequences— and that is what the Carriers section reports.

The enzyme that digests lactose stops being produced after childhood in almost all mammals, and in a good share of people. Continuing to produce it as an adult is what is new, and it depends on a single position.

It is exactly the Mendelian mechanism from the previous page, with one difference: here a single copy with the variant is enough to keep producing the enzyme. That is why there are three results and not two.

It is a way of seeing the rule without the weight of a disease: one gene, three possible genotypes, three results that can be told apart.

Diabetes, heart attack, hypertension. Here no single gene decides: thousands of variants each contribute a minimal push, on top of everything that is not genetic.

Many small, independent effects always add up into the same shape, a bell, so here there is no exact calculation like the 25% in the cross square: there is a position. The score adds the weight of each variant you carry, is compared against a reference population, and that gives a percentile. Which population matters: the vast majority of these scores were fitted on cohorts of European ancestry, and they transfer worse to admixed populations such as South American ones. Your report adjusts the percentile for your measured ancestry and states, score by score, how much of it is represented.

Look at the width of the band: the percentile comes with an interval, and the report always shows it. In the general population the lifetime risk of this condition is 10.5% — which is why the report speaks of how many people out of a hundred and not of how many times more: “double” can be 2 to 4 out of 100, or 30 to 60.

Every finding in the report comes with a letter beside it. It is not an ornament: it is how much support that statement has, and it completely changes what can be done with it.

At A there is a professional guideline that says what to do, and that is why pharmacogenomics and carriers can change clinical management. At B the association is firm but adds little to what clinical assessment already tells us.

At C the finding is replicated and even so no one has shown that knowing it improves an outcome. And at D it is exploratory: interesting, with no clinical implication.

Muscle fiber type, the skin's response to the sun, perception of bitter taste. They are in the report because they are interesting, and because many people are curious to see what their genome says about everyday things.

They are also stated at grade D, which is where they belong: the effect of each variant is minute and the studies backing them are small or were not replicated. A result from here does not enable any decision, and the report says so before showing it.

Both things are true at once, and that is why the section exists and comes closed: you can enjoy something while knowing that it does not mean much. What you cannot do is present it with the same weight as an interaction with a medication.