Targeted sequencing identifies case-only variants that alter constrained elements and are more common in high OCD risk breeds. (a) We performed targeted sequencing of a small number of cases and controls from four breeds (row 1) and subsequently genotyped the top candidate variants in a larger panel of dogs from those four breeds as well as two more high ‘OCD-risk’ breeds and two low risk ‘control breeds’ (row 2). (b) Across all variants identified in the sequencing data, the number of case-only (orange box) and control-only (purple box) variants is similar, but constrained elements are enriched for case-only variants. Boxes mark the 25th to 75th percentile across dogs, with the median shown as a thick line, and whiskers extending to values within 1.5 times the difference between the 25th to 75th percentiles. Outliers are marked with circles. (c) Case-only variants have higher frequency in OCD-risk breeds and lower frequency across all genotyped breeds. The x-axis represents allele frequencies across all genotyped dogs. The y-axis represents normalized allele frequency (AF) differences between OCD-risk and control breeds ([AFOCD-risk - AFcontrol]/[AFOCD-risk + AFcontrol]). The straight downward line represents the linear model for the data points. The blue shade shows the 95% confidence interval for this model. Area under the curve (shaded in grey) is notably larger in AFOCD-risk > AFcontrol than in AFOCD-risk < AFcontrol, showing that case-only variants are more common in OCD-risk breeds than in control breeds.