W55. REGIONAL BURDEN ANALYSIS OF FUNCTIONAL VARIANTS REVEALS GENOMIC HOTSPOTS ENRICHED FOR REGULATORY ELEMENTS AND OVERLAPPING WITH RECURRENT PATHOGENIC CNVS

Background We recently developed the regional burden (RB) approach, a novel analytic method that examines the regional burden of low-frequency functional genetic variants. Using this approach, we showed that a regional excess of these genetic variants is associated with increased schizophrenia (SCZ) risk as well as with decreased educational attainment. In this study, we implemented the RB analysis in autism spectrum disorder (ASD). We subsequently compared the identified RB hotspots in ASD to those found in SCZ and examined overlap with recurrent pathogenic CNVs. Lastly, we explored the biological function of genomic regions most vulnerable to excess regional burden. Methods We combined the two largest ASD datasets (SSC and SPARK-WES-1) comparing probands (8,842) with unaffected siblings (3,968) with available sequence data. We used 22q11.21 as template and applied our sliding-window approach across the genome, using logistic regression with ASD as outcome, regional SNV excess as predictor, while accounting for sex and 20 population stratification components. We determined RB hotspots indicated by consecutive associated windows. We subsequently compared the top 20% of ASD-related RB hotspots with those for SCZ, and with regions where recurrent pathogenic CNVs occur, using permutation to test for significance. We used enrichment analyses to explore the function of the overlapping hotspots for ASD and SCZ (10% of the genome). Results We found similar results for ASD as for SCZ reported in our initial study for 22q11.21 (OR=2, p=0.005). The RB-hotspots for ASD overlapped with hotspots for SCZ (p < 1e-19) and pathogenic CNVs (p < 1e-09). The overlapping hotspots for ASD and SCZ were highly enriched for evolutionary constrained transcriptional factors (p < 1e-99), suggesting implication of these regions in genome regulation. Discussion Our findings show that the RB analysis generates meaningful results when applied to ASD, adding to our previous observations in SCZ. We show that genomic regions most vulnerable for the cumulative regional impact multiple sequence level variants strongly overlap with regions of recurrent pathogenic CNVs. We also show significant overlap with previously identified RB-hotspots for SCZ. The genomic regions where the RB has the strongest impact on ASD and SCZ risk are strongly enriched for transcription factors suggestion a role in genome regulation. The question raised by these results is how evolutionary strongly constrained non-coding DNA sequences are implicated in the risk for two uniquely human conditions, ASD and SCZ. Our findings suggest that RB analysis is a novel tool that may contribute to the elucidation of genomic regions influencing vulnerability for ASD and SCZ.