Round-up July 23rd – Sept 2nd

The polygenics of sexuality (see below) has been in the limelight. I appreciated that the dialogue around this study got into some of the “so what” of this type of work. In a post on Medium I reflect on the fact that we seem to be a bit picky and choosy about which traits we are willing to embrace as having a genetic etiology. A recently published study reveals some other biases: Antisocial behavior is judged as less genetically influenced than prosocial behavior. Probably because people wanted to hold others to be blameworthy for their actions.

I continue to think that we’re racing into a future where we have some handle on biological differences between individuals (via probabilistically predictive scores of various types, genetic and otherwise), and that we lack the necessary frameworks for knowing what to do with this information. Putting my time where my worries are, starting today I’m officially employed to think about exactly these questions. 



  • The “no single gay gene” study in Science, covered by numerous news outliets, found that the trait ever versus never had same sex sex is ~32% (95% confidence interval 11-54%) heritable, and is polygenic. From a scientific point of view, that result is completely unsurprising, just shoeing that this trait is like all others studied. One thing that was different about this trait: male-female genetic correlations were considerably less than the usual one. Secondary results from the study included that personality traits, mental health disorders, and risky behaviors were genetically correlated with the trait, but not physical traits. They also looked at the proportion of same-sex to total partners, and found that this trait was not that highly correlated with the ever/never had same sex sex (0.73 for men, 0.52 for women). Their conclusion? Sexuality is complex, and can’t be captured by a single dimension, as scales like Kinsey’s try and do. The study is an interesting case example of how ethics was engaged with. 
  • Polygenic scores cannot be interpreted as a “genetic endowment” because the predictive power can come from a number of other sources: population stratification, assortative mating, and genotype-environment correlation (rGE). They found that the predictive power of polygenic scores for cognitive traits was higher (by 60%) between families than when looking at non-identical twins. But if socioeconomic status was controlled for, much of the difference disappeared. They conclude that socioeconomic status contributes to the between family predictive power of cognitive polygenic scores, acting as an “environmentally mediated parental genetic effect”, a type of genotype-environment correlation. 
  • A preprint that covers the potential effects of selecting embryos on the basis of polygenic scores. When selecting between ten embryos, with 20% of the variance captured by the polygenic score, the gain (selecting for highest score embryo versus average embryo) would be ~0.5 of a standard deviation, with a 95% confidence interval of -1.2 to 2.3. Compare to the standard deviation of 6cm for height and 15 IQ points. 
  • A preprint of a large (135,000) Finnish study on polygenic scores for a few major diseases. Comparing the top 2.5% to the middle 20-80%, hazard ratios of 2.0-4.3 (depending on disease). They conclude that the performance of the polygenic scores was similar to established clinical risk scores.
  • An example of sociogenomic research: finding that some polygenic scores are associated with exposure to bullying.
  • A preprint that points to a way polygenic scores could be made clinically relevant: by focusing on the variants that contribute to polygenic scores that also fall within gene sets tied to particular drugs. 
  • Polygenic score for Alcohol abuse, 
  • More evidence for the common genetic underpinnings of various psychiatric disorders.
  • Some genes in humans are up-regulated after death



  • Erik Parens and Paul Appelbaum consider the evidence for the widely predicted negative psychosocial impacts of genetic testing. Years of study have shown that individuals do not end up more depressed or anxious or stressed (though there may be transitory effects. Note the recent Hastings Center Special Report containing many papers on this topic). They see a path forward “Using more subtle measures than merely asking people if they feel depressed or anxious” to show that “genetic results can affect people in real ways.” Here are the examples of this newer generation of work that they point to:
    • A study that shows those were APOE e4+ and knew rated their memory worse and performed worse on a memory test than those who were positive and didn’t know. 
    • A study where individuals were told whether they had a genetic variant associated with decreased exercise capacity independently of their genetic results found that those told they had the variant performed worse.
    • A study that shows the impact of perceived genetic cause of obesity, with subjects more likely to eat more if they thought it was genetic.