Cellular and spatial context of genetic convergence between insomnia liability and DTI ALPS perivascular diffusivity

Sleep has been linked to brain waste clearance and glymphatic/perivascular transport, yet whether insomnia shares genetic architecture with perivascular diffusivity remains unclear. Here we show that insomnia liability and diffusion tensor imaging measures along the perivascular space have little genome-wide genetic correlation but converge locally at chromosome 17q21.31. Polygenic enrichment, conjunctional false discovery rate, and Bayesian colocalization analyses support shared statistical signals in this region, which contains the MAPT inversion locus. Transcriptome-wide, gene-level fine-mapping, and summary-data Mendelian randomization analyses prioritize HEXIM1, ACBD4, EFTUD2, and MAPT. Tissue, cell-type, spatial transcriptomic, and sleep-perturbation analyses place these candidates in neuronal, glial, immune, vascular, and perivascular contexts, with effects varying by brain region and imaging phenotype. Together, these findings support localized genetic convergence rather than a genome-wide shared architecture and provide biological context for future studies of perivascular imaging phenotypes and neurodegeneration-related processes. Insomnia and MRI-derived perivascular diffusivity show little genome-wide genetic correlation but converge locally at 17q21.31, highlighting HEXIM1, ACBD4, EFTUD2 and MAPT.