Three coding changes in one bitter receptor gene decide whether compounds like PTC and PROP taste intensely bitter or of almost nothing. It is the strongest genetic association on this site, and one of the oldest known human polymorphisms.
Some people put a strip of paper treated with a compound called PTC on their tongue and recoil from it. Others taste nothing at all and assume the whole thing is a trick. Both are telling the truth. The difference is a single receptor gene, TAS2R38, and it is one of the clearest demonstrations in all of human genetics that two people can inhabit measurably different sensory worlds.
The receptor detects a particular chemical shape — the thiourea group — found in PTC and PROP, the two compounds used to test for this in the laboratory, and in a family of natural compounds called glucosinolates that give brassica vegetables their bite: broccoli, brussels sprouts, kale, cabbage, turnip.
This is not a single-switch trait. Three coding changes in TAS2R38 travel together and are read as one unit, called a haplotype:
Two combinations account for the great majority of people. PAV (proline, alanine, valine) builds a receptor that responds strongly — the taster form. AVI (alanine, valine, isoleucine) builds one that barely responds — the non-taster form. Everyone carries two copies, so the usual outcomes are PAV/PAV (strong taster), PAV/AVI (intermediate), and AVI/AVI (non-taster). Looking at any one of the three positions on its own gives a partial answer at best.
Unusually, this is a page where the honest answer is "extremely". Most pages on this site have to explain that a genetic effect is smaller than the internet suggests. Here the association between this haplotype and measured bitterness is among the strongest ever reported for any common human trait — the statistical strength at position 296 reaches p = 3 x 10-199, which is many orders of magnitude beyond the strongest association anywhere else on this site.
It is also old. The taster/non-taster split was described in the early 1930s, after a chemist in a laboratory noticed that a compound blowing about in the air tasted bitter to one colleague and of nothing to another. It has been studied ever since.
It reliably predicts how bitter PTC and PROP taste. It is a weaker guide to anything else. The obvious next question — do non-tasters eat more vegetables? — has been asked many times, and the results genuinely conflict: some studies find bitter-sensitive people eat fewer brassicas, others find no difference once age, culture and what is actually available to eat are accounted for. Food preference is shaped by far more than one receptor, and this page will not pretend otherwise.
What it does explain, quite precisely, is a common domestic argument. If coffee, tonic water, dark chocolate, grapefruit or brussels sprouts taste sharply bitter to you and unremarkable to someone at the same table, this is a real physiological difference, not fussiness.
Gene and haplotype. TAS2R38 (7q34) encodes a type 2 taste receptor, a G-protein-coupled receptor expressed on type II taste cells, responsive to thiourea-containing compounds including phenylthiocarbamide (PTC) and 6-n-propylthiouracil (PROP). Three non-synonymous variants define the common haplotypes: rs713598 (p.Ala49Pro), rs1726866 (p.Ala262Val) and rs10246939 (p.Ile296Val). PAV is the taster haplotype and AVI the non-taster; rarer recombinant haplotypes (AAI, AAV, PVI) exist at low frequency and are functionally intermediate.
A strand caution. TAS2R38 is transcribed from the minus strand, so the genomic alleles reported by variant databases are the complement of those used in protein nomenclature. Genomic G at rs713598 gives Pro49, genomic G at rs1726866 gives Ala262, and genomic C at rs10246939 gives Val296 — that is, the taster residue in each case. Reports that state this the other way round are a recurring source of error.
Effect sizes. In a bivariate genome-wide analysis of bitter taste perception in 1,999 individuals of European ancestry from 929 twin families (BMC Genomics 2018; PMID 30223776), rs10246939 reached p = 3 x 10-199 and rs1726866 p = 3 x 10-59. An earlier study of quinine taste intensity in 1,457 individuals of European ancestry (Hum Mol Genet 2010; PMID 20675712) reported rs713598 at p = 2 x 10-104. rs10246939 was independently associated in a Brazilian cohort (Hum Mol Genet 2013; PMID 23966204). Effect sizes of this magnitude are exceptional for a common variant and reflect that the variant is the causal change in the receptor protein itself, not a marker in linkage with something else.
Population frequency. Both haplotypes are common worldwide, and the proportion of non-tasters differs substantially between populations; commonly cited estimates place it near a third in European-ancestry groups, with lower figures reported in several African and East Asian populations. Precise, broadly representative diplotype frequencies are not consistently reported across studies and are not given here at a false level of precision.
Reported associations beyond bitterness. Correlations with brassica vegetable intake, alcohol consumption, smoking behaviour and body weight have all been reported and none replicate consistently; effect sizes are small and confounded by cultural and availability factors that dominate food choice. TAS2R38 genotype has also been examined in relation to upper respiratory innate immunity, where the receptor is expressed in sinonasal epithelium and responds to bacterial quorum-sensing molecules — a mechanistically interesting line of work that has not produced clinical application. None of these should be presented to a reader as consequences of their genotype.
Clinical status. None. TAS2R38 genotyping has no diagnostic, prognostic or therapeutic role, is not used in clinical practice, and predicts a laboratory taste phenotype rather than a health outcome.
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Bitter Taste Perception (TAS2R38) comes down to these specific, well-studied positions — not a diagnosis.
Because the three coding changes travel together as a haplotype and the receptor is built from all of them. PAV (proline-alanine-valine) is the taster form and AVI (alanine-valine-isoleucine) the non-taster form. Reading any one position alone gives a partial answer.
No. This receptor detects one particular chemical shape. Sweet, salty, sour and umami are handled by entirely different receptors, and most other bitter compounds are detected by other TAS2R genes. A non-taster tastes normally; they simply do not register these specific compounds strongly.
Not reliably. The link between this genotype and how many vegetables people actually eat has been studied repeatedly with conflicting results, because food choice is shaped far more by culture, habit and what is available than by one receptor. What the genotype does explain is why the same vegetable can taste sharply bitter to one person and unremarkable to another.
Yes. The association at position 296 reaches p = 3 x 10^-199, many orders of magnitude beyond anything else on this site. That is possible because the variants change the receptor protein directly rather than sitting near a gene and influencing it indirectly.
No. It has no clinical use and is not connected to any health outcome. It is an explanation, not a finding — which is a category this site has rather few of and is glad to publish.
Free to reuse. This page's text is original writing from freely-available research, licensed CC BY 4.0 — reuse it, including commercially, with attribution to MyGeneLog. It's general research-derived information, not medical advice or a diagnosis — see Terms of Use.