What your DNA can — and can't — tell you about your diet

"A DNA test told me I should avoid saturated fat. My colleague's said she needs more omega-3. My other colleague's said she has a 'caffeine sensitivity gene'. All three of us eat the same lunch every day. So — how much of this is actually real?" — The question I get asked more than any other

Consumer genomics has exploded over the last decade. Millions of people have taken SNP-based tests and received personalised diet recommendations based on their genetic variants. Some of these recommendations are grounded in solid science. Others are not. The problem is that the packaging rarely tells you which is which.

I have worked in nutrigenomics for over twenty years — first in population genetics and molecular evolution, now in clinical genomics pipelines and personalised health through CESGEN. What follows is an honest account of what your DNA can tell you, what it genuinely cannot, and how to think about the gap.

What is nutrigenomics, precisely?

Nutrigenomics is the study of how individual genetic variation influences the body's response to nutrients — and, conversely, how nutrients influence gene expression. The field rests on a well-established biological fact: two people eating the same meal will not have identical metabolic responses. Some of that difference is explained by their gut microbiome, their hormonal state, their activity levels. A meaningful fraction of it is explained by genetics.

The genetic signals relevant to nutrition come mostly from single nucleotide polymorphisms (SNPs) — positions in the genome where the DNA letter varies between individuals. A SNP in the gene that codes for lactase, for example, strongly predicts whether an adult can digest lactose. That is one of the cleanest gene-diet interactions in human biology, and it illustrates what nutrigenomics looks like at its best: a specific gene, a specific enzyme, a specific dietary substrate, a clearly measurable outcome.

Where the evidence is solid

A small number of gene-diet interactions are supported by strong, replicated evidence and have genuine clinical utility:

Lactase persistence (LCT gene)

The rs4988235 variant in the regulatory region of the LCT gene determines whether lactase enzyme expression persists into adulthood. Carriers of the derived allele can digest milk; non-carriers increasingly cannot. This is not a risk — it is a functional difference with a clear dietary implication.

Folate metabolism (MTHFR gene)

Variants in MTHFR — particularly rs1801133 (C677T) — reduce the efficiency of folate conversion and are associated with elevated homocysteine. The evidence for supplementation in carriers is reasonably robust, particularly in the context of pregnancy. This is one of the most clinically useful findings in the field.

Saturated fat and cardiovascular risk (APOE gene)

APOE ε4 carriers show a stronger LDL cholesterol response to dietary saturated fat than ε2 or ε3 carriers. The effect is real and has been replicated, though the magnitude varies between studies. APOE ε4 is also the strongest genetic risk factor for late-onset Alzheimer's — information that carries its own weight when delivered in a clinical context.

Caffeine metabolism (CYP1A2 gene)

The rs762551 variant in CYP1A2 distinguishes fast from slow caffeine metabolisers. Slow metabolisers show a greater blood pressure response to caffeine and, in some studies, a higher risk of myocardial infarction with high coffee intake. The association is credible, though the absolute risk differences are modest.

Where the evidence is weak — or absent

Beyond these well-characterised interactions, the picture becomes considerably murkier. Most complex traits — body weight, metabolic rate, appetite regulation, insulin sensitivity — are influenced by hundreds or thousands of variants, each with tiny individual effects. No single SNP tells you whether you will gain weight on a high-fat diet, or whether intermittent fasting will work for you. Polygenic scores for these traits exist, but their predictive power at the individual level remains limited.

Many of the recommendations issued by commercial tests stretch well beyond what the underlying evidence supports. "You have a gene variant associated with vitamin D metabolism, so you should supplement with vitamin D" — this sounds specific, but the variant in question may explain a small fraction of your vitamin D status, and the recommendation would apply to a large proportion of the population regardless. When a recommendation is indistinguishable from general public health advice, the genetic framing adds little value.

What to do with this information

The most useful nutrigenomic tests are those that identify genuine functional variants with clear dietary implications — not those that generate the longest list of personalised recommendations. When evaluating a test or interpreting a result, it is worth asking: is this variant on a gene with a well-understood biological function? Is the gene-diet interaction specific and mechanistically plausible? Has it been replicated in independent cohorts? What is the effect size?

Done well, nutrigenomics is not about replacing general dietary advice. It is about identifying the subset of individuals for whom specific modifications are likely to make a meaningful difference — and distinguishing them from the majority for whom standard recommendations are appropriate.