Combining multiplexed assays of variant effect for enhanced BRCA2 variant classification

Determining the clinical relevance of BRCA2 variants of uncertain significance is critical for informed risk management. Recently, two saturation genome editing studies assessed the functional effects of all single nucleotide variants in the BRCA2 C-terminal DNA Binding Domain. To improve the accuracy of functional data used for ACMG/AMP variant classification, we combined results from these studies in four composite models and evaluated the performance of each model using variants with known classifications. Here, we show that an "Integrated VarCall Model", which combined raw functional data for 6383 variants from the original studies, yielded 98.8% accuracy and out-performed the original studies and other combined data models. Incorporation of the "Integrated VarCall Model" functional data with other sources of evidence according to ClinGen BRCA1/2 variant curation expert panel specifications resulted in classification of 5926 (92.8%) BRCA2 variants as pathogenic (n = 735) or benign (n = 5191) and provides valuable insights for individuals with BRCA2 variants.