Episode Transcript
[00:00:03] Here is one result from today's study. Near a gene called ZPR1, there is a single DNA letter that can be a G or a C.
[00:00:13] Among people with two copies of G, those who took fish oil had blood DHA a long chain omega 3 0.57 standard deviations higher than those who did not. Among people with two copies of C, the gap was 0.46. Same supplement, different genetic background, a different association with DHA. How do you find a letter like that in a whole genome?
[00:00:40] There is a catch. Finding that kind of variant is hard. A gene that matters only in combination with a habit like taking a supplement leaves a weak trace. In ordinary genetic studies you need enormous samples, and even then the signals are small.
[00:00:57] So the team tried a different route. Instead of looking first at the average level of each fatty acid in the blood, they looked at something most studies ignore how much that level varies from person to person.
[00:01:14] This is base by base. Every episode is built with AI tools from a real Open Access paper curated by a human, and every claim is checked against the paper before we we publish. You'll find this episode's fact check score on our website.
[00:01:28] Now back to the story.
[00:01:35] Today we celebrate the work of Susan Adana Iheijirika Eunice Steven and Kai Xiong Yi from the University of Georgia, who have advanced our understanding of how genes and fish oil supplements together shape the fatty acids in our blood. This base by base episode is based on their Open Access article A Variance QTL Approach to Uncover Gene Fish Oil Supplement interaction loci for 14 circulating unsaturated fatty Acid Traits, published in Human Genetics and Genomics advances in 2026.
[00:02:10] Here is the intuition Imagine a variant that raises blood DHA only in people who take fish oil. Among carriers, some take the supplement and some do not, so their DHA values fan out, some high, some low. The carrier's DHA varies more from person to person. That extra variation can be a fingerprint of an interaction with the environment.
[00:02:34] So can a genome scan for variation in fatty acid levels. Point us toward the genes that change how fish oil works.
[00:02:44] Start with the molecules.
[00:02:46] Unsaturated fatty acids are fats with at least one carbon double bond. Monounsaturated fatty acids, or MUFAs, have one. Polyunsaturated fatty acids, or PUFAs, have two or more. PUFAs come in two big families, Omega 3 and Omega 6, named by where the first double bond sits counting from the end of the chain.
[00:03:09] Dietary or circulating PUFAs and MUFAs have been associated with better outcomes in diseases and such as cardiovascular disease, diabetes and obesity.
[00:03:21] Two of these fats are essential, which means the body cannot make them. The omega 3 alpha linolenic acid and the omega 6 linoleic acid called LA. They must come from food. The body can then convert them step by step through desaturation and elongation reactions into long chain fatty acids on the omega 3 side, that includes EPA and DHA on the omega 6 side, arachidonic acid. In the modern western diet, linoleic acid alone accounts for about 85% of all dietary PUFA intake.
[00:03:58] Fish oil supplements are an accessible source of the long chain omega 3s, especially EPA and DHA. And genetics clearly matters for these fats.
[00:04:08] The best known example is a cluster of two genes, FADS1 and FADS2. They encode the desaturase enzymes that add double bonds as precursors are converted into long chain PUFAs. Variants in this cluster affect how efficiently the body makes them, and earlier work showed that diet can modify those genetic effects.
[00:04:30] That is what geneticists call a gene environment interaction.
[00:04:34] The effect of a variant on a measurement such as blood DHA differs depending on the environmental context.
[00:04:41] These interactions are recognized as a key mechanism behind differences between people in many measurable human characteristics, but they are notoriously hard to detect. Interaction studies have low statistical power. They need larger samples than ordinary association studies. The effects are modest and exposures are often measured imprecisely.
[00:05:05] This team had already tried the direct route in a large genome wide interaction study of fish oil supplements and the same 14 blood fatty acid measures. They highlighted the FADS cluster and a gene called GPR12.
[00:05:21] They estimated that genome wide gene fish oil interactions explained between 0.53 and 1.51% of the variation in some of these measures, such as the share of DHA among all fatty acids, measurable but small.
[00:05:38] And the strict correction for testing millions of variants may have hidden more modest interactions.
[00:05:49] So the authors split the problem into two stages.
[00:05:52] Stage 1 Scan the genome for variants quantitative trait loci or VQTLs. These are variants associated with how much a fatty acid level, such as DHA, varies from person to person among people who share the same genotype rather than with its average.
[00:06:09] Stage 2 Take only those VQTLs and test each one explicitly for an interaction with fish oil.
[00:06:16] Put another way, the variant scale works as a filter. Instead of testing millions of variants for interaction, you test a short prioritized list.
[00:06:27] The data came from the UK Biobank, a prospective cohort of about 500,000 volunteers in the United Kingdom recruited between 2006 and 2010, aged 40 to 69. At recruitment, the team kept unrelated participants of European ancestry. All of them passed a series of genetic quality checks. After quality control, 7,713,873 genetic variants entered the variant scan.
[00:06:57] The fatty acids were measured by NMR Metabolomics in a randomly selected subset of about 280,000 participants. The team studied 14 fatty acid measures. Six were absolute levels, total omega 3, DHA, total omega 6 Louisiana total PUFAs and total MUFAs.
[00:07:18] Six were the same fats as percentages of all fatty acids and two were ratios PUFAs to MUFAs and Omega 6 to Omega 3. Up to 207,736 people had complete data across these measures.
[00:07:35] How was fox fish oil measured? With a single question at recruitment on a touchscreen do you regularly take any of the following? Fish oil, including cod liver oil was one of the options. Anyone who ticked it was counted as a user, so the exposure was a simple yes or no without dose frequency or composition. The interaction analysis included 200,060 participants. About 31.8% of them took 15 oil regularly.
[00:08:06] For the variance scan, the team used the median based Levene's test which compares how far values fall from the group median across the three genotype groups of each variant. Each fatty acid measure was adjusted for age and genetic ancestry within each sex.
[00:08:22] Extreme outliers were removed. Values were standardized. Genome wide significance was set at the usual threshold, a p value below 5 times 10 to the minus 8 for stage 2. Each fatty acid measure got its own Bonferroni corrected threshold based on how many VQTLs it had.
[00:08:46] So what did the variant scan find? A lot.
[00:08:50] 172 pairs of a variant and a fatty acid measure across all 14 measures merged across measures. They pointed to 46 non overlapping regions of the genome. The fatty acid measures differed widely. Omega 6 percentage and Proofa percentage had the most 24. Each DHA percentage had just 1 and Omega 3 percentage 2.
[00:09:15] Most of these variants sat outside genes or inside introns in non coding DNA, which points to gene regulation.
[00:09:25] Some regions influence the variation of many fatty acid measures at once. A locus mapped to a long non coding rna near the Trib1 gene affected the variation of 12 of the 14 measures. A missense variant in the GCKR gene was linked to the variation of 11 measures. That variant has been associated with with over 100 human traits and diseases. So GCKR is highly pleiotropic not only for averages but also for variants. A colocalization analysis linked the GCKR locus to 13 of the 14 measures.
[00:10:05] Which tissues and pathways do these variation variants point to? Candid gens were enriched for expression in the liver, brain, whole blood, lymphocytes and some gastrointestinal and endocrine tissues. Pathways such as cholesterol homeostasis and MTORC1 signaling were shared across omega 3 and omega 6 measures. The authors stress that these patterns come from bulk tissue expression and gene set enrichment, not from causal assignments.
[00:10:36] How do variants link to variation relate to ordinary variants that shift the average fatty acid level? The team compared them with a recent large genome wide association study of the same kinds of fatty acid measures which reported 1,164 loci.
[00:10:53] 90.7% of the VQTLs 156 of 172 also had an effect on the mean and effect sizes correlated moderately with an R of 0.5.
[00:11:06] Variants that move the average level tend to move its variation too, but 16 VQTLs had no detectable effect on the mean at all.
[00:11:17] Those 16 are the most intriguing because a variant that changes the variation without changing the average is exactly what a hidden interaction could produce.
[00:11:27] One of them sits in an intron of the KSRT2 gene known to regulate the Ras pathway and energy balance and associated with obesity. Its C allele was linked to less person to person variation in the Omega 6 to Omega 3 ratio with a nominal interaction with fish oil a p value of 0.01.
[00:11:49] When the team split people by fish oil use, the picture became clearer. In non users. The seal allele was associated with a lower omega 6 to omega 3 ratio in fish oil users. It showed no significant effect on the mean ratio mix the two groups and the effect on the average may be attenuated while the variation still differs. The authors offer a speculative explanation that without fish oil this allele may favor Omega 3 production or retention and they stress that the mechanism remains to be elucidated.
[00:12:24] Now. Stage 2 each of the 172 variant and fatty acid pairs was tested for an interaction with fish oil.
[00:12:32] Six passed the threshold set for each fatty acid measure. They were in DHA, DHA percentage, Omega 3, Omega 3 percentage LA and the Omega 6 to Omega 3 ratio and they mapped to three the FADS cluster, the ZPR1 gene and a region spanning SUGP1 and TM6 SF2 that is 3.49% of the pairs tested. The last two regions had not reached significance in the team's earlier genome wide interaction scan take the ZPR1 variant, its G allele was associated with higher DHA and and that effect was larger in people taking fish oil than in people who did not. When the team split people by genotype, fish oil was associated with dha in all three groups, but most strongly in people with two copies of G at 0.57 standard deviation units compared with 0.47 and 0.46 in the other two groups. This variant sits near the Apoa 5 apolipoprotein cluster, which is involved in lipid metabolism and triglycerides.
[00:13:43] The SUGP1 and TM6 SF2 region told a different story. There a variant was associated with higher LA and omega 6, but less so in fish oil users. In people with two copies of the A allele, fish oil showed no effect on linoleic acid, while in the other genotype groups it was positively associated with it. The authors interpret this signal through TM6SF2, a gene involved in exporting fat from the liver. Because the variant is in complete linkage with a well known TM6 SF2 variant at the FADS cluster, the top variant had a smaller effect on omega 3 in fish oil users than in non users. The authors read this as consistent with a simple idea. Fish oil supplies these fatty acids from outside, which may reduce the need for the body's own synthesis and with it the weight of variants in that pathway. And all six interactions stayed significant after separate adjustments for oily fish intake and for the Townsend Deprivation Index, a measure of socioeconomic deprivation.
[00:14:52] Did the variance filter really help? The team tested the weight 1164 ordinary loci the same way. Only five significant interactions emerged there 0.43% of the pairs compared with that interactions were enriched among the variants linked to variation by a factor of 8.12, a factor of 8.12.
[00:15:16] The confidence interval is wide from 2.26 to 27.07, likely because the counts are small but its lower bound stays above 1 and the enrichment held at several P value cutoffs.
[00:15:31] A second check pointed the same way. 24 of the 172 variants linked to variation showed some sign of interaction with fish oil that used a looser cutoff, a p value below 0.05 that is 14%.
[00:15:47] Under pure chance you would expect about 5%.
[00:15:51] A one sided binomial test put the probability of seeing that many by chance at 5.8 times 10 to the minus 6 and the Eightfold enrichment sits close to what recent large analyses reported for other measurements, a tenfold and a five fold enrichment of interactions among variants linked to variation.
[00:16:14] What does this change?
[00:16:16] Mostly, it supports a strategy Variant screens can prioritize where to look for gene environment interactions and here they turned up two additional fish oil interaction loci that a direct genome wide scan had missed.
[00:16:30] The authors also publish the 14 sets of variants results as a public resource so other teams can test the same variants against other exposures, such as other dietary sources of these fats.
[00:16:44] There is a tempting conclusion here, and the authors are careful with it. Carriers of the ZPR1 gene allele may get more benefit from fish oil in terms of raising dha, they write, but these are associations in an observational cohort with differences of a fraction of a standard deviation. The mechanisms are hypotheses for ZPR1. The authors propose that genetic differences in lipid pathways affect how efficiently dietary omega 3s are incorporated into plasma lipids, and they say future studies are needed to test it.
[00:17:21] The limitations matter. The study included only participants of genetic European ancestry because other ancestries had small samples in the UK biobank and allele frequencies at these loci vary across populations. Fish oil use was a yes or no answer without frequency or supplement composition, and the primary analyses did not adjust for every factor linked to taking fish oil. Some confounding may remain still the six signals held after adjusting for oily fish intake and deprivation.
[00:17:58] So here is where this leaves us by looking first at how much fatty acid levels such as DHA varied from person to person, not just at their averages, the team narrowed the search for genes that change how fish oil works in about 200,000 people. Six interactions emerged at three loci, the fads cluster ZPR1 and SUGP1 with TM6SF2.
[00:18:23] The last two were newly found by this approach.
[00:18:25] Small effects one population and a long way from a personalized prescription, but the idea is powerful. The same capsule may not mean the same thing for everybody, and the variation in a blood measurement can tell you where to look.
[00:18:43] What does this mean for the future of personalized nutrition when genes and how habits finally get studied together?
[00:18:56] This episode was based on an Open Access article under the CCBY 4.0 license. You can find a direct link to the paper and the license in our episode description. If you enjoyed this, follow or subscribe in your podcast app and leave a five star rating. If you'd like to support our work, use the donation link in the description.
[00:19:14] Thanks for listening and join us next time as we explore more science. Base by base.