Whole genome sequencing vs a DNA array file — what actually differs
Almost every raw DNA file in circulation came from a genotyping array. Sequencing is increasingly offered alongside it, at several times the price, and the marketing rarely explains what the extra money buys.
The difference is not “more markers”. It is a different kind of measurement, and that changes which questions can be asked at all.
What is the difference between a DNA array and sequencing?
An array reads a pre-selected list of positions — typically 500,000 to a million of them — chosen before anyone looked at your sample. Sequencing reads the genome as it comes, letter by letter, and works out afterwards what is there. One asks prepared questions; the other takes dictation.
That distinction explains almost everything downstream. An array is fast and cheap because the questions were decided in advance and the same chip is used for everyone. Sequencing costs more because it has to read everything and then make sense of it.
An array file lists roughly 0.02% of your genome. Whole genome sequencing covers close to all of it, and whole exome sequencing covers the protein-coding portion, which is about 1–2%.
What can sequencing find that an array cannot?
Rare variants, and anything the chip designer did not anticipate. An array can only report positions on its list, so a rare or novel variant is simply absent from the file. Sequencing also resolves insertions and deletions, and with good coverage it supports structural findings that arrays cannot address at all.
This matters most in exactly the situations people care about. Clinically significant variants in genes like BRCA1 and BRCA2 are frequently rare and family-specific. Arrays typically carry only a handful of founder variants in those genes, so a clear array reading is not an all-clear.
It matters least for common traits. If a trait is driven by common variants — most of nutrition, fitness, traits and ancestry — the array already carries them, and sequencing adds cost without adding much answer.
Does more coverage always mean better results?
No, and coverage depth is where sequencing marketing gets loosest. Depth describes how many times each position was read, so higher depth means fewer errors at any given position. But depth cannot compensate for a weak variant-calling pipeline, and it does not improve the interpretation applied afterwards.
Low-pass sequencing, sometimes sold at around one-fold coverage, is a different product again. It reads each position roughly once and then uses statistical imputation against reference panels to fill the gaps. That works reasonably for common variants and poorly for the rare ones that were the reason to sequence in the first place.
So “30x” or “100x” on a product page is one input to quality among several. The pipeline, the reference build, the curation and the reporting all sit downstream of it, and none of them is described by a coverage number.
Should you pay to upgrade from an array file to sequencing?
It depends entirely on what you want answered. For common traits, pharmacogenomics and ancestry, the array file you already have carries the relevant markers and an upgrade buys little. For a specific clinical question — a family history, a known variant in the family — sequencing through a clinic is the right route, not a consumer upgrade.
The case that does not work well is upgrading in the hope of a more interesting report. More data does not produce more certainty about common traits, because the uncertainty there comes from the research rather than from the file.
Anything with a real clinical dimension belongs with a clinician who can order the right thing and interpret it in context. That is a different service from any consumer product, ours included.
Which file formats can an interpretation service actually read?
Array exports arrive as .txt or .csv, a few hundred kilobytes to a few megabytes, and almost every upload tool reads them. Sequencing produces VCF files of variant calls, or much larger BAM, CRAM and FASTQ files of raw reads, and support for those varies considerably between services.
If you have sequencing data and want it interpreted, check the accepted formats before paying. A service built around array files may accept a VCF and then use only the fraction of it that overlaps its existing marker list — which is a real outcome, and not always disclosed.
We currently read array exports from the major consumer services. That is a stated limit rather than a claim about what matters, and it is the honest thing to say before you upload anything.
What we look at in a raw DNA file → · What the major services’ files contain →