Genetic control and tissue specificity of transcriptomic rhythmicity
Circadian rhythms influence molecular traits in most tissues, yet genetic studies mostly treat molecular measurements as static. In particular, genetic effects on gene expression are typically investigated through expression quantitative trait locus (eQTL) studies, which ignore sampling time and therefore miss genetic effects on rhythm parameters such as amplitude and acrophase. In this study, we used a cosinor-based joint genotype-by-rhythm interaction framework applied to GTEx version 10 gene expression data, with donor inferred phase as a proxy for circadian time. We first identified rhythmically expressed genes in 49 of the 50 tissues profiled in GTEx. We then tested common variants within 50 kb of each rhythmic gene's transcription start site for genetic regulation of transcriptomic rhythmicity. We identified 167 circadian expression QTLs (cd-eQTLs), comprising 54 lead signals for 51 genes across 28 tissues. Most cd-eQTLs altered oscillation amplitude, while others affected acrophase or both. We then investigated the tissue specificity of cd-eQTLs and found that those acting on core circadian genes had consistent effects across tissues, whereas those acting on clock-regulated genes were largely tissue-specific. A phenome-wide association study in the Pan-UK Biobank linked 34 cd-eQTLs to 100 phenotypes, including self-reported chronotype, adiposity, blood cell counts and blood biochemistry. These results show that genetic variation shapes circadian gene expression in a tissue-dependent manner, and that accounting for time of day reveals regulatory effects that static eQTL analyses cannot detect.