P568: Population frequency versus in silico prediction accuracy for variant pathogenicity criteria: Evaluation in a cohort of 4000 exomes*

The classification of genetic variants under the ACMG/AMP/ACGS framework has served as a standard for clinical molecular diagnostics for nearly a decade. As genomic testing expands in scale and resolution, the strengths and limitations of the 2015 guidelines have become increasingly evident. Current efforts to update the criteria include proposals to reduce the contribution of population-frequency data while increasing the evidentiary weight assigned to computational prediction tools. Although in silico methods are increasingly sophisticated, they remain indirect proxies for biological effect and show considerable variability in specificity, concordance, and clinical performance.