POLYGENIC AGGREGATION OF EQTL EFFECTS REVEALS CELL-TYPE AND TRAIT-SPECIFIC LONG-RANGE REGULATORY PATTERNS

Gene regulation varies across genomic regions and cellular contexts, shaped by both linear distance and 3D genome architecture. Recent work on chromosomes 22q and 16p revealed that common polygenic risk for neuropsychiatric disorders - particularly schizophrenia (SCZ), autism, ADHD, and low IQ - is associated with widespread decreases in gene expression across these chromosome arms in neuronal cell types. These findings motivated our development of a scalable method to investigate the extent to which long-range regulatory patterns exist genome-wide, and whether these patterns are specific to neuropsychiatric disease or neuronal cellular contexts.

To systematically assess how regulatory effects vary with SNP-gene distance, we developed a polygenic risk score (PRS)-inspired framework that aggregates eQTL effect sizes weighted by GWAS effect sizes across different trait and cell-type contexts.

Our initial analysis successfully replicated the previous 22q findings across four neuronal cell types: glutamatergic and GABAergic neurons, astrocytes, and oligodendrocytes. We observed that SCZ-weighted eQTL effects were consistently negative across distances up to 16 Mb, particularly in glutamatergic neurons and astrocytes, where height-weighted eQTL effects were null. Oligodendrocytes exhibited null effects in both trait contexts. These results confirmed the existence of trait- and cell-type-specific signatures of long-range gene regulation. We subsequently expanded the analysis genomewide, identifying many regions of the genome in which long range eQTL effects could not be identified in any trait context (e.g. chr4: 118Mb - 151Mb), and a small number of regions of the genome in which common variant risk for SCZ, but not somatic traits, was similarly downregulating genes across large genomic territories (e.g. chr1: 1Mb - 33Mb).

Ongoing work will incorporate additional brain and non-brain cell types, and integrate chromatin contact data to connect spatial genome organization with the observed regulatory patterns. These analyses will provide additional insight into how transcriptome-wide gene dysregulation creates risk for SCZ and other neuropsychiatric disorders, and the assign biological consequence to polygenic influences on psychiatric disease. More broadly, this PRS-style eQTL aggregation framework offers a scalable approach for identifying how genetic risk influences gene regulation across distance, cellular identity, and disease relevance.