How it is done

From your saliva to the letters of your DNA

Between a saliva sample and a list of letters there are four stages. These are them.

ATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGCATGC

or step through with the ‹ › arrows

Stage 1

The full tour, in text

1 · From saliva to DNA ready to read

The sample is saliva or buccal mucosa. Saliva in a tube with stabilizing solution, or a swab of the buccal mucosa. It requires no blood draw and no fasting: collection at the lab takes a few minutes, and you can also pick up the kit and do it later.
Buccal epithelial cells provide the DNA. The mucosal epithelium renews itself continuously and its shed cells remain in the saliva, together with the leukocytes present in it. Both populations retain the nucleus, and with it the complete genome.
Genomic DNA is isolated. The cells are lysed —the membrane is broken and the contents released— and the DNA is separated from the remaining macromolecules: proteins, lipids, RNA and the bacterial flora of the mouth. What is left is a solution of purified genomic DNA.
The material is amplified. The amount of DNA obtained is not enough for the assay, so it is amplified by isothermal replication. The sequence is not modified: the process copies; it does not edit or correct.
It is fragmented enzymatically. An intact chromosome cannot hybridize on the surface of the chip. Enzymatic fragmentation reduces it to segments that diffuse freely until they find their complementary probe.

2 · A chip the size of a microscope slide

A piece of glass with 24 sections. The size of a microscope slide. Each run takes 24 samples; yours occupies one section.
Millions of wells. The surface is not smooth: it is etched with wells spaced less than six microns apart. About 30,000 fit into one square millimeter.
A silica bead in each well. Each bead carries hundreds of thousands of copies of the same oligonucleotide. Redundancy is what makes the signal measurable: a single molecule does not give detectable fluorescence.
Each bead type interrogates one position. The oligonucleotide coating each bead type is complementary to a single position in the genome. Everything else follows from that: the chip measures only what its probes interrogate.
Assembly is random, and that is why it has to be decoded. They self-assemble through Van der Waals forces, so neither the order nor the quantity is defined beforehand. That is why every chip is decoded AT THE FACTORY, hybridizing against a 22-to-24-nucleotide tag that identifies each bead: by the time your sample arrives, the map already exists.
Each position, more than thirty times. Because the distribution is random, each probe ends up repeated in more than thirty different wells: that is 19,620,810 measurements for 654,027 positions. The value that reaches your report is the average, and measurements that depart from the rest are discarded.

3 · From a loose fragment to a colored dot

The letters always pair the same way. A with T, C with G. It is not a convention: it is how the molecules fit together.
Each fragment finds its place on its own. Nothing guides them. They are poured onto the chip, move at random and collide with probes until they meet the complementary one; if it does not match, they let go and carry on. That is why the well where one ended up bound already tells which position of the genome it came from.
A single base, and it comes labeled. The probe ends just before the position of interest, and stops there. An enzyme extends it by exactly one base: the complement of the one you have, which acts as the template. That base arrives labeled.
Two labels, not four. A and T share one label, C and G the other. The slide is then stained: each label takes on a different color, and the well ends up lit with the color of the version you have.

4 · From colors to genotypes

The scanner head photographs the slide. A laser illuminates the chip section by section and a camera records the fluorescence of each well. The instrument does not return letters: it returns an image with one colored dot per well, and turning that into genotypes is a separate job.
Each dot is two numbers. How brightly one version glows and how brightly the other does. With those two values, the position is placed on a plane.
Three clusters form on their own. Nobody draws them. With two possible versions there are exactly three combinations, and they separate by themselves.
Your letter is the cluster you fall into. Twice the same version, twice the other, or one of each. That is a genotype.
And if it falls between two clusters. It is discarded rather than guessed. Declaring that it could not be read is preferable to putting down a letter that may not be yours.
And from there, to your report. Every letter read is matched against the clinical catalogues and comes out as a result with its level of evidence stated. What could not be read is named rather than omitted: an honest report also says what it does not know.

The limits

What this technology cannot do

Understanding how the chip works explains where its limits come from. They are not fine print: they follow directly from the test asking about a list of positions instead of reading everything.

It finds, but it does not rule out

If a known variant has its probe on the chip, the test will see it. If the variant is rare, or private to your family, there is no probe asking about it and the result comes out negative all the same. That is why a negative lowers the probability, but does not bring it to zero.

It reads letters, it does not count copies

The chip says which letter is at each position, not how many copies of a gene you have. The conditions that depend on that —and some important ones do— fall outside its scope, and the report names them instead of omitting them.

It is not sequencing

Sequencing means reading the genome letter by letter, without deciding beforehand what to look at. A microarray asks about a fixed list of positions. It is far cheaper and faster, and that is the reason it exists; it is also the reason for its limits.

Genetics from scratch

What DNA is, what you inherit, what a risk really means and what ancestry can say. Nine chapters with figures you can touch.

See a sample report

The complete report with fictional data: exactly what you will receive, section by section.

Book the test — ARS 159,000

You pay online with Mercado Pago. The sample takes five minutes.