Accuracy of genome-enabled polygenic risk score prediction of cruciate ligament rupture risk in the Labrador Retriever

Cruciate ligament rupture (CR) is a prevalent and heritable orthopedic disease in dogs, especially in Labrador Retrievers. This study evaluates the accuracy of polygenic risk score (PRS) prediction for CR using genome-wide SNP data in a well-phenotyped Labrador Retriever population. A training set of 1,006 dogs and a validation set of 52 dogs were genotyped using the Illumina CanineHD array. Eight statistical models—including Bayesian regression and machine learning methods—were assessed for their ability to predict CR status based on genetic markers. Model performance was evaluated using AUC, accuracy, and R² metrics. The study further analyzed SNPs located in genic regions associated with CR and incorporated relevant covariates. Our results support the potential clinical utility of PRS in identifying dogs at high genetic risk for CR, enabling earlier intervention and improved breeding strategies.