Integrative genetic analyses of circadian chronotype and autism highlight regional co-association at 17q21.31.

Abstract

Chronotype and sleep problems are common in autism spectrum disorder (ASD), but their genetic relationship remains uncertain. We integrated published cross-trait LD score regression (LDSC), brain eQTL-based summary-data-based Mendelian randomization (SMR), SNP-based two-sample MR, two-step mediation MR, secondary parallel SMR, and Bayesian colocalization using UK Biobank chronotype GWASs, the iPSYCH/Psychiatric Genomics Consortium ASD GWAS, and GTEx v8 brain eQTLs. The published LDSC estimate was negative for morningness-coded chronotype and ASD (rG = -0.199). Primary ASD-outcome SMR identified 50 gene-tissue pairs representing 24 genes at the exploratory multi-SNP PSMR < 0.001 threshold; none met Bonferroni significance. Fifteen candidates occurred in the LD-rich 17q21.31 inversion and are not independent gene discoveries. Secondary eveningness-outcome SMR tested all 24 candidates in matched tissues; 22 had PSMR < 0.05, but GTEx reuse made this descriptive. At 17q21.31, the prespecified p12 = 1 × 10⁻⁵ favored distinct variants (PP.H3 = 0.785) over one shared variant (PP.H4 = 0.126). H3 dominance strengthened at p12 = 1 × 10⁻⁶ and reversed only at the permissive p12 = 1 × 10⁻⁴, equal to p1 and p2. Eveningness-oriented IVW MR was positive (β = 0.270, P = 0.008), whereas robust estimators were attenuated; no mediation was detected through sleep duration or insomnia complaints. These analyses prioritize exploratory regional associations without proving shared causality or mechanism.