Epigenetic context predicts gene expression variation and reproductive traits across genetically identical individuals.

In recent decades, genome-wide association studies (GWAS) have been the major approach to understand the biological basis of individual differences in traits and diseases. However, GWAS approaches have proven to have limited predictive power to explain individual differences, particularly for complex traits and diseases in which environmental factors play a substantial role in their etiology. Indeed, individual differences persist even in genetically identical individuals, although fully separating genetic and environmental causation is difficult or impossible in most organisms. To understand the basis of individual differences in the absence of genetic differences, we measured two quantitative reproductive traits in 180 genetically identical young adult Caenorhabditis elegans roundworms in a shared environment and performed single-individual transcriptomics on each worm. We identified hundreds of genes for which expression variation was strongly associated with reproductive traits, some of which depended on prior environmental experience and some of which was random. Multiple small sets of genes together were highly predictive of reproductive traits across individuals, explaining on average over half and over a quarter of variation in the two traits. We manipulated mRNA levels of predictive genes using RNA interference to identify a set of causal genes, demonstrating the utility of this approach for both prediction and understanding underlying biology. Finally, we found that the chromatin environment of predictive genes was enriched for H3K27 trimethylation, suggesting that individual gene expression differences underlying critical traits may be driven in part by chromatin structure. Together, this work shows that individual differences in gene expression that arise independently of underlying genetic differences are both predictive and causal in shaping reproductive traits at levels that equal or exceed genetic variation.