Single or romantically involved…don’t blame the “happy gene”

As we approach the end of November we might start thinking about the best and the worst of 2014. This week, a study was mentioned “as a strong candidate for worst genetic study to make the headlines in 2014” by Professor Tim Frayling, a geneticist at the University of Exeter.

The study, which was led by Jinting Liu at the Peking University, appeared in Scientific Reports with the title: “The association between romantic relationship status and 5-HT1A gene in young adults”.

This was a hypothesis driven study… though perhaps driven by quite a peculiar hypothesis (more on that below). The field of human genetics has now almost universally accepted the idea that most studies that are designed to test single genes being associated with one particular phenotype (known as candidate gene studies) have led to many false positives that are rarely replicated. Genetic variants can come with two flavours, or alleles (e.g. 1 and 2), in three possible combinations: 1/1, 1/2 or 2/2. The basic principle of genetic association studies is to measure the frequency of allele 1 and allele 2  in two groups of people, typically cases and controls (e.g. high or low blood pressure). A frequency higher than expected for allele 1 of a particular marker in the high blood pressure group would be indicative that a specific marker is associated “with blood pressure”.

One complication is that allele frequencies of any selected markers (and millions have now been catalogued across the human genome) are variable across populations, even on a small geographical scale. If the group of high blood pressure cases were selected from a particular hospital and the low blood pressure controls were collected across a much wider geographical area, we would see different allelic frequencies, maybe statistically significant, but simply due to sampling effects rather than being genuinely associated with blood pressure. This and other artifacts explain why candidate genes studies have led to so many false positive results. A way to control for the validity of genetic association is to run the analysis in a completely independent sample and see if the results can be replicated. More reproducible results have been obtained through a hypothesis-free approach which tests the entire genome, using so-called genome-wide association studies, or GWAS.

The current study looked at romantic relationship status, and a single genetic variant was tested for the tendency to be in a romantic relationship on the basis of a somewhat arbitrary hypothesis. The 5-HT1A gene is a receptor for serotonin, which is a neurotransmitter linked to the development of depression. The chosen genetic variant has  previously been shown to be associated with different psychiatric phenotypes and regulation of mood. Relationship status is likely to be influenced by lots of environmental and social factors, but there is no evidence to believe that the ability of an individual to hold a relationship could be determined by a specific genetic factor. The study reported a very weak association, not supported by an independent replication, which for the reasons I explained above, cannot be distinguished from a false positive at present. This is one of the reasons Professor Frayling did not like the study. Paul de Bakker, Professor of Genetic Epidemiology and Bioinformatics at University Medical Center Utrecht and a member of the editorial board of Scientific Reports was not impressed either, and openly questioned the reviewing process of the journal.

The authors admit that one should “be cautious when interpreting the current finding”, yet they also state that “these findings provide, for the first time, direct evidence for the genetic contribution to romantic relationship formation.” This is a strong statement, not really supported by the findings themselves, which are at best marginally suggestive of an association that needs to be further investigated. It is easy to see how the media could make a story around these results and their interpretation, however.

Indeed a variety of colourful headlines followed: “A gene that might play a role in keeping you single”; “Single again? Maybe you should blame your genetics”, “Were you BORN to be single? Scientists discover a gene that makes certain people bad at relationships”, “Only the lonely: study identifies gene for ‘romantic relationship failure‘” or “‘Happy gene’ may increase chances of romantic relationships”. The “happy gene” expression is particularly bad as it implies people may or may not have a gene which can lead to happiness and, as a consequence, to establish relationships.

Recently, the media seemed to have become a bit better at reporting genetic studies, trying to move away from the “gene for” concept, the idea that individuals carrying one flavour of one gene determines the most diverse and complex phenotypes. It has become more common to see caution and we have started seeing discussions around the idea that multiple biological and genetic factors contribute to complex phenotypes, as in the case of common disorders like diabetes or schizophrenia or personality traits.

Caution in interpreting the findings was brought in when asking the opinion of researchers that were not involved in the study. The Daily Mail, in particular, reports skepticism, even if not with a direct quote, from ProfessorTim Spector, Professor of Genetic Epidemiology at Kings College. Indeed, both the paper and most of the media report that the effect of the 5-HT1A variant is very small (1.4%) and certainly not enough to explain why some people are single and other are romantically invovled. However, the Guardian, for example, chooses to end the report with the remark that 50% of people in a relationship are carriers of the specific 5-HT1A variant, compared to only 40% of those who are not, concluding that this “is quite a gap”.

This is a typical example of an inconclusive study which makes it into the media because a catchy story can be told around it, rather than because it holds a particular scientific merit.

Liu, J. et al. (2014). The association between romantic relationship status and 5-HT1A gene in young adults. Scientific Reports. DOI:10.1038/srep07049