Decode the Genome: What Sequencing Can Tell Us—and What It Cannot

1990: an international consortium sets out to read the complete human instruction book. 2003: the Human Genome Project declares the job done—two years ahead of schedule, though roughly 8% of the sequence is still missing. 2022: the Telomere-to-Telomere consortium finally fills in the last gaps. Today: sequencing a human genome is a routine laboratory service rather than a moonshot.

Somewhere along that timeline, reading DNA became easy. What did not become easy is understanding what we read. That gap—between the letters and their meaning—is the most important thing to grasp before you order a genetic test, read a headline about a “gene for” something, or wonder what your own DNA might reveal.

What sequencing actually does

Your genome is the complete set of DNA in your cells: billions of chemical letters, or bases, written in a four-letter alphabet (A, C, G and T). A gene is a stretch of that text that carries instructions, usually for making a protein. Sequencing means determining the exact order of the letters.

Not every “DNA test” reads the same amount of text:

• Whole-genome sequencing attempts to read nearly all of it.
• Exome sequencing reads only the protein-coding portions of genes—a small fraction of the genome, but the part where many known disease-causing changes sit.
• Gene panels read a chosen set of genes linked to a particular condition, such as inherited cancer risk.
• Genotyping arrays, used by many consumer tests, do not read continuous text at all. They check a pre-selected list of known positions, like spot-checking a book instead of reading it.

That distinction matters: a test that never looked at a particular region cannot tell you anything about it. Try our Decode a Genome game to see how a sequence becomes a result.

From a $3 billion project to a routine test

The Human Genome Project cost about $3 billion over 13 years, according to the National Human Genome Research Institute (NHGRI). That figure covered far more than reading one genome—it built maps, tools and methods. NHGRI estimates that generating the draft sequence itself cost around $300 million, with more spent to finish it.

Then costs collapsed. By 2006, NHGRI data suggested a high-quality draft human genome cost roughly $14 million; by late 2015, the cost had fallen below $1,500. NHGRI is careful to note that these figures cover producing the raw sequence, and typically exclude the expensive, skilled work of quality control, assembly and—above all—interpreting what the variants mean.

The first reference genome was also a patchwork. NHGRI notes that about 70% of it came from a single anonymous donor, with the rest from 19 other people, mostly of European descent. The 2003 version left roughly 8% unresolved, largely in repetitive stretches that older technology could not untangle. The Telomere-to-Telomere consortium announced the first truly complete, gapless human sequence in March 2022.

Sequencing turns biology into text. Medicine is still learning how to read it.

What sequencing can tell us

In the right setting, sequencing already changes care. Some of its most established uses:

• Diagnosing inherited conditions. For many rare diseases caused by a change in a single gene, finding that change can end years of uncertainty and guide care.
• Inherited cancer risk. Germline testing looks at DNA you were born with, usually from blood or saliva. The National Cancer Institute (NCI) notes that about 5–10% of cancers are thought to be caused by harmful inherited changes—so most cancers are not primarily inherited.
• Tumour profiling. Somatic testing looks at genetic changes that arose in cancer cells during a person’s life. It can help match some patients to targeted treatments, but NCI stresses that it does not replace testing for inherited risk. Our guide to the new cancer toolkit explains how these results feed into precision oncology.
• Choosing and dosing medicines. Pharmacogenomics is the study of how genes affect drug response. The FDA maintains a table of drugs whose labels include genetic information—for example, the heart drug clopidogrel and the gene CYP2C19, and the HIV drug abacavir and the immune gene HLA-B.

What it cannot tell us—yet

A lot of variants are simply unexplained

Everyone carries many DNA differences, and most have no known effect on health. When a lab finds a change it cannot classify, it reports a variant of uncertain significance (VUS). MedlinePlus, the US National Library of Medicine’s health site, explains that a VUS cannot establish a diagnosis or a risk. NCI adds that, as evidence accumulates, a VUS is most often reclassified as benign—which is why it is not something to act on alone.

Negative is not the same as “no risk”

A negative result means the test did not find a relevant change in what it examined. MedlinePlus notes that a test can miss some alterations. And a consumer test that checks only a handful of known positions can miss many harmful variants in genes such as BRCA1 and BRCA2, a limitation NCI highlights.

Common diseases are mostly many-gene, many-factor problems

For heart disease, type 2 diabetes and many other common conditions, risk is spread across a large number of small genetic nudges, plus diet, activity, sleep, smoking, environment and chance. A polygenic risk score adds up those small nudges into a single estimate. NHGRI emphasises three caveats: such scores are probabilities, not certainties; most of the underlying research has been done in people of European ancestry, so scores may be less accurate for everyone else; and they are not yet routinely used by health professionals because practice guidelines are lacking.

DNA is a script, not the performance

Your genome is largely the same in every cell, yet a neuron and a skin cell behave very differently. Which genes are switched on, when and how strongly depends on layers of regulation that a DNA sequence alone does not reveal. That is one reason a genome can suggest risk but rarely deliver a forecast.

Reading your own results: questions worth asking

This is not medical advice, and results should always be interpreted with a clinician or genetic counsellor who knows your personal and family history. But these questions help frame the conversation:

1. Was this a clinical-grade test or a consumer product, and which kind of test was it—genome, exome, panel or genotyping?
2. Does the result concern inherited (germline) DNA or tumour (somatic) DNA?
3. Is a finding classified as pathogenic, benign or uncertain—and could that classification change?
4. Was the reference data drawn from people with ancestry similar to mine?
5. Does the result change anything I or my doctor would actually do?

Where genomics goes next

The next chapter is less about reading faster and more about reading wisely: building reference data that reflects the full diversity of humanity, sharpening how variants are classified, and connecting DNA to outcomes over years. Reading a gene is also increasingly the first step toward changing one—the subject of our feature on gene editing, gene therapy and RNA medicines.

For now, the honest summary is this: sequencing can answer some questions with remarkable precision, and many others not at all. Knowing which is which is the real literacy of the genomic age.