Genetics of Homosexuality: Twin Studies, part I

By Mister Curie

Please let me know if anything in this post is confusing or too technical.  I'm being trained as a geneticist and am not always sure how to write for a non-technical audience.

I found the results of my poll on whether scientists should study the genetics of homosexuality to be quite interesting.  While I received several comments on the post cautioning against the study of homosexuality genetics and describing the ways in which such research could be misused, when forced to choose one or the other, 90% (28 votes) thought that scientists should study the genetics of homosexuality, while only 10% (3 votes) believed that scientists should not.  This tells me that even though people are nervous about the implications of such research, most of my readers that voted are interested in learning more about the genetics of homosexuality, and that many would also be interested in a discussion of the ethical concerns raised by such research.  I am looking forward to working on several series of future posts to address these different interests.

Before delving into the details, a bit of background is probably required.  Homosexuality is complex and influenced by both genes and the environment.  When scientists first want to understand the relative contributions of genes and environment to the outcome of interest, they often turn to families and in particular, families with twins.  The first thing to investigate is whether the outcome of interest (such as homosexuality) is more likely to occur in families than in the general population.  If it is, that may come from shared genetics or shared environment.  Twins are useful in teasing these two apart.  Siblings share the same environment (except perhaps the uterine environment) but only half of the same DNA.  Identical twins share the same environment (including uterine environment) and have the exact same DNA.  Fraternal twins share the same environment (including uterine environment) but only half of the same DNA, they are no more genetically alike than siblings.  Really fancy twin studies will further tease the effects apart by studying adopted siblings (who share the environment but none of the same genes) and identical twins separated at birth (who share the same genes, but not the same environment - except the uterine environment).

 

Scientists then look at the different types of twin and sibling pairs and calculate how often the pairs have the same outcome of interest (here homosexuality).  Scientists can then estimate how much of the outcome comes from shared genes, shared environment, and unique environment.  If the identical twins have the same outcome of interest more often than fraternal twins, that is evidence of a genetic effect because the twins all shared the same common environment.  If fraternal twins have the same outcome of interest more often than siblings, that is evidence that the uterine environment may be important.  If related siblings have the same outcome more often than adopted siblings that is evidence of a genetic effect. If adopted siblings have the same outcome more often than population rates of the outcome, that is evidence of shared environment effects.  If identical twins that grow up together have the same outcome more often than identical twins separated at birth, that is evidence of a shared environment effect. Non-shared outcomes are attributed to unique environmental effects.

A few words additional words, identical twins do not need to always share the same outcome in order for it to be genetic, they only need to share the same outcome more often than fraternal twins (who share the same environment but not the same genes).  Twin studies are typically small and the results are typically presented as point estimates (given the data, what is the best estimate for the genetic contribution, shared environment contribution, and unique environment contribution) and confidence intervals (given the size of the sample, how much might the point estimate vary if we re-did the study with different individuals).  The point estimate is considered significant if the confidence interval does not include zero.  However, just because a confidence interval includes zero, that does not mean that there is no effect.  Larger studies are more likely to have confidence intervals that do not include zero if the point estimate is not zero.

Tomorrow we will review the latest and most informative twin studies for homosexuality.