Genome-wide analysis of social behaviour across social domains, reporters and developmental stages: a meta-regression approach

Abstract

Social behaviour is a heritable trait linked to well-being and mental health, yet it varies across social context, including social domains, informants and development, which may shape its underlying genomic architecture. Here, we conducted a genome-wide meta-regression meta-analysis of social behaviour from infancy to early adulthood, studying ~500,000 repeated measures in European-ancestry cohorts (N = 73,321). We modelled heterogeneity in single-variant associations across social domains (low prosocial behaviour and peer/social difficulties), multiple informants (parents, teachers and self-reports) and ages (2–29 years), identifying 6 associated loci, including variation within CADM2 (P = 2.51 × 10−9). Single nucleotide polymorphism-based heritability ranged from 2% to 7%, and the genetic architecture comprised 4 factors reflecting differences in either social domains or informants. Polygenic scores (N = 16,305) accurately predicted contextual variation in independent cohorts, with some cross-ancestry transferability. Context-specific patterns of genetic correlations, particularly when estimating the developmental onset of associations, offer opportunities to disentangle shared trajectories linking social behaviour with later-life and mental health outcomes.