Abstract
The human genome encodes a limited number of genes yet contributes to individual differences in a vast array of heritable traits. A possible explanation for the capacity our genome to generate this virtually unlimited range of phenotypic variation in complex traits is to assume functional interactions between genes. Therefore we searched two mammalian genomes to identify potential epistatic interactions by looking for co-adapted genes marked by excess two-locus genetic differentiation between populations/lineages using publicly available SNP genotype data. The practical motivation for this effort is to reduce the number of pair-wise tests that need to be performed in genome-wide association studies aimed at detecting G×G interactions, by focusing on pairs predicted to be more likely to jointly affect variation in complex traits. Hence, this approach generates a list of candidate interactions that can be empirically tested. In both the mouse and human data we observed two-locus genetic differentiation in excess of what can be expected from chance alone based on simulations. In an attempt to validate our hypothesis that pairs of genes showing excess genetic divergence represent potential functional interactions, we selected a small set of gene combinations postulated to be interacting based on our analyses and looked for a combined effect of the selected genes on variation in complex traits in both mice and man. In both cases the individual effect of the genes were not significant, instead we observed marginally significant interaction effects. These results show that genome wide searches for gene-gene interactions based on population genetic data are feasible and can generate interesting candidate gene pairs to be further tested for their contribution to phenotypic variation in complex traits.
Highlights
The presence of epistatic interactions between genes has fundamental consequences for the course and outcome of evolution by natural selection [1,2,3,4] and results in the emergence of co-adapted gene complexes, i.e. combinations of variants at different genes that give a selective advantage only when both are present in the same individual
As Recombinant inbred lines (RILs) are derived from a known breeding scheme, any significant result should not be caused by demography
The simulations showed that very high levels of Linkage Disequilibrium (LD) may be present in RILs just by chance, starting from r2.0.04, the real data consistently exhibits a significant excess of non-independent combinations (Figure 1), when compared to the average and standard deviation observed in the 1000 replicate simulations
Summary
The presence of epistatic interactions between genes has fundamental consequences for the course and outcome of evolution by natural selection [1,2,3,4] and results in the emergence of co-adapted gene complexes, i.e. combinations of variants at different genes that give a selective advantage only when both are present in the same individual. It should be possible to detect genes involved in epistatic interactions from a population genetic approach, without prior knowledge of the specific phenotypes that are affected. Between (sub-) populations the same process will generate genetic differentiation because different combinations of alleles may perform well, resulting in phenotypically equivalent but genetically different populations [2]. This theoretical prediction of Wright’s Shifting Balance Theory of evolution has empirical support [4]. By looking for deviations from equilibrium two-locus genotype frequencies, we aim to detect the signature of natural selection [6,7,8,9] on pairs of genes from population genomic data
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
