From Pythagoras to Pinot: the Genetics of Wine Appreciation

By Brian Ring, PhD.

Alongside of the Olympics, gyros, and the concept of democracy, the ancient Greeks also gave us two other important contributions: our word for wine (from the Greek oinos), and the idea of consumer genetics. Pythagoras, the philosopher and mathematician who thought of the Pythagorean theorem also proclaimed, “kyamon apechehesthe”: avoid fava beans. We don’t have his reason for this injunction, but it may be due to a practical understanding of genetics. Eating fava beans can lead to a form of anemia in people who lack working copies of the enzyme glucose-6- phosphate dehydrogenase (G6PD). Deficiency in G6PD is relatively common in people from the Mediterranean region, and thus Pythagoras’s caution was a useful warning to his community.

Genetics can tell us a lot more about our food choices than just whether or not we can eat fava beans. Some of us are picky eaters while some will eat anything not nailed down (I’m in that latter class– and upon reflection, having enjoyed a Chinese tea called ‘bitter nail’, I may even be farther down the spectrum). Some of this is due to our exposure and familiarity with different foods and some is due to our genetics, which we are learning can be a key player in our eating habits. Using coffee as an example, the contribution our individual genetic variations play into determining our coffee consumption has been estimated to range from 35 to 60% (1) , and estimates for the heritability (the proportion of a trait’s variation due to genetics) for caffeine metabolism range from 73 to 89% (2,3) . Genetic variation related to our perception of bitterness, sweetness, and several odors have all been identified (4).

To study how our genetics might influence our enjoyment of wine, we surveyed 541 people at a series of wine tasting events, asking them about their taste preferences, eating and drinking habits, and relevant demographic questions. They were offered twelve wines encompassing a number of varietals, then questioned on their preferences among these wines and their ability to detect previously defined flavors in each. In addition, 41 genetic variants were assessed in all survey participants. Each of these variants have been shown to be associated with our ability to perceive certain tastes in food, and we deemed them solid candidates for genes which may also affect our wine preferences.

We used a variety of scientific tools, including clustering and regression, to explore the relationship between wine preferences and genetics in order to understand the complexity of the data. With the help of wine experts and this data exploration, we identified eight major taste categories, or ‘bins’, into which all of the wines fell. We then split the participants into two groups, one to use for training a model (aka, the control group), and another for testing. The survey responses and genetic variants having the greatest effect size and with the highest significance for their association with wine preference were singled out, and a tool called elastic net regression was used to remove variables strongly associated with other variables.Ultimately, thirteen survey responses and five genetic variants were used to model each of the wines. All the models showed strong and consistent relationships between similar wines and between the discovery and test groups.

Several of the genetic variants showed relationships with wine preference that were consistent with their previously reported taste associations. For example, a variant in the Ghrelin gene, a neuropeptide known to regulate appetite and our consumptions of sweets, showed a strong negative association with dry, unoaked white wine and a weak positive association with a sweet Riesling. Similarly, a variant in a taste receptor (TAS2R46) that has been associated with the ability to taste bitterness in coffee and other foods was negatively associated with liking the sweet Riesling and positively associated with the more bitter red wines. In addition to the five variants that were used in the models, five additional variants we found to have interesting correlations with wine taste preferences were included in the test. These variants will hopefully aid in the development of an even more robust tool for predicting wine preference.

These taste and genetic algorithms allow us to suggest wines that a person might enjoy by applying the models to their survey question and gene variants, and then offering a wine from their highest ranked bins. New wines, beyond the initial set of wines from the tasting events, are assigned to the bins based on a set of defined tasting profiles. The models will be further refined over time by feedback from participants as the panel of assessed wines grows.

There are, of course, several limitations to this method. For one, we limited our genetic variants for this first set of models to those with known associations with tasting preferences or tasting ability. While this means that the science behind these genes is strong, it also means that the variants have a hard time, mathematically, of standing out compared to the taste questions also
used in the model. For example, questions about whether one prefers one’s coffee black or with sweetener are strongly associated with preference for bold red wines. However this same question about coffee is also strongly correlated with a gene variant we tested with known associations with bitterness perception. The gene variants increase the statistical significance of the model compared to just using the survey questions, but a goal is to expand the model beyond the initial set of curated variants.

Secondly, our taste preferences are complicated and change over time, while our genetic variants are fixed at birth. We all have foods we once considered dreadful yet are now among our favorites. I still look askance at celery, but my younger self could not imagine my current acceptance of okra or stuffed peppers. Genes that affect our perception of bitterness also affect our alcohol consumption, those who are sensitive to bitterness are less likely to drink alcohol (5,6). However experience with alcohol seems to change how our bitter tasting ability affects our consumption behavior (7,8). For those who want to enjoy wine, a heightened ability to taste bitterness may lead to more consumption, possibly because these people have the ability to discern more flavors in wine’s complex taste profile.

Genetics is an important part of how we choose foods, but is only part of the story. Unlike medical tests, which must have ‘actionability’ (the ability to offer clear directions of an optimal and effective treatment plan), consumer genetics lets us be in the driver seat. Genetics can be a flexible and highly useful aid that gets us part of the way to making a decision, but isn’t a
complete answer. We’re use to this, most of our everyday problems, from what shirt to put on in the morning, which restaurant to choose in the evening and which wine to try, are made using a variety of sources of information. In contrast to Pythagoras’s simple declaration against beans, genetics is beginning to give us individualized and adaptable tools we can use as we see fit.

1.   Yang, A., Palmer, A. A. & de Wit, H. Genetics of caffeine consumption and responses to caffeine. Psychopharmacology 211, 245-257 (2010).
2.   Matthaei, J. et al. Heritability of Caffeine Metabolism: Environmental Effects Masking Genetic Effects on CYP1A2 Activity but Not on NAT2. Clinical pharmacology and therapeutics
100, 606-616 (2016).
3.   Rasmussen, B. B., Brix, T. H., Kyvik, K. O. & Brosen, K. The interindividual differences in the 3-demethylation of caffeine alias CYP1A2 is determined by both genetic and environmental factors. Pharmacogenetics 12, 473-478 (2002).
4.   Reed, D. R. & Knaapila, A. Genetics of taste and smell: poisons and pleasures. Progress in molecular biology and translational science 94, 213-240 (2012)
5.   Duffy, V. B. et al. Bitter receptor gene (TAS2R38), 6-n- propylthiouracil (PROP) bitterness and alcohol intake. Alcoholism, clinical and experimental research 28, 1629-1637 (2004).
6.   Lanier, S. A., Hayes, J. E. & Duffy, V. B. Sweet and bitter tastes of alcoholic beverages mediate alcohol intake in of-age undergraduates. Physiol Behav 83, 821-831 (2005).
7.   Hayes, J. E. & Pickering, G. J. Wine Expertise Predicts Taste Phenotype.  Am J Enol Vitic 63, 80-84 (2012).
8.   Pickering, G. J. & Hayes, J. E. Influence of biological, experiential and psychological factors in wine preference segmentation.  23, 154-161 (2017).