Understanding Genomes in Social Contexts

The debate over the importance of social forces relative to human nature in determining aspects of
human life, like health, is one of the oldest in modern academia. Recent evidence indicates that
the traditional binary dichotomy is wrong - outcomes are dependent on both. Specifically, there is
evidence that our response to socio-environmental risks is moderated by genetic
predispositions. Less, however, is known about how social contexts affect genomes.
A well replicated finding in social sciences is that friends are similar to each other. Research by
Fowler and Christakis shows that this association is not purely phenotypic, but that there is an
association between non-related friends at the level of genes. For example, on average, friends'
genotypes are as similar as fourth cousins'. Indeed, they used the association between two nonrelated
genotypes to predict friendships. They stipulate that these findings might explain the
clustering and spread of non-communicable diseases in social networks. However, the way in
which friend's genotypes and phenotypes become similar remains unclear. Fowler and Christakis
propose several explanations for their finding: 1) the finding could be due to population
stratification, 2) active gene-environment correlation, 3) people choosing friends who have similar
phenotypes to them, or 4) environments which select people with certain genetic traits, e.g. a
professional sports team. There is evidence to support the occurrence of all four. For example,
Boardman found that the socio economic status of a school explained a large part of the variation in
genetic similarity between school friends. Recent animal research by Baud et al (2017) provides
experimental evidence that phenotypic similarities in social networks might also be directly caused
by the genotypes of those around them. She found that 'social genetic effects' could explain up
to 29% of the phenotypic variation in pairs of genetically identical or non-identical cage mice, forced
to live with each other. There is thus a growing body of evidence which suggests that social contexts
effect gene expression. However, the exact mechanism by which this occurs is unclear.
My proposed project will examine how genomes relate to social contexts. Specifically, it will:
1. Seek to replicate Fowler and Christakis' findings, by estimating the genome wide association
between friend pairs, and, subject to a literature review, explore how the association is
dependent on putative social factors, like SES, and/or specifically genetic factors.
2. To test if Baud et al (2017)'s laboratory findings generalise to humans. I will do this by testing the
hypothesis that an individual's friend's polygenic risk score for various outcomes, like
depression, should predict their own risk for these outcomes. I hope to then be able to use more
sophisticated epidemiological methods to assess the causal effect of any found association. For
example, attempting to use statistical modelling, or even Mendelian randomisation, to test a
causal relationship.
3. Finally, social media has become an increasingly important part of how social networks form and
evolve. Using new Twitter data collected in the Avon Longitudinal Study of Parents and Children,
I will explore whether the results from 'off-line' social networks generalise to 'on-line' social
networks.