Fig. 3: Evaluation of 244 BRCA2 variants by applying Bayesian inference to the MANO-B method.
From: High-throughput functional evaluation of BRCA2 variants of unknown significance
a–c Functional classification by the MANO-B method with four drugs. A total of 244 BRCA2 variants were classified by Bayesian inference. The number of variants classified for each drug as fClass 1/2 (a), fClass 3 (b), and fClass 4/5 (c) is shown. No variant was classified as fClass 1/2 with one drug and fClass 4/5 with another drug. d Bar plots of functional variant effects and functional diagnosis of 244 BRCA2 variants mapped against the BRCA2 full-length sequence and domains. The key domains of BRCA2 consist of a PALB2 interaction domain encompassing amino acids (a.a.) 10–40, a transactivation domain (TAD) encompassing a.a. 18–105, a RAD51-binding domain including eight BRC repeats encompassing a.a. 1008–2082, an MEILB2-binding domain (MBD) encompassing a.a. 2117–2339, a DNA-binding domain encompassing a.a. 2402–3186 and containing a helical domain (HD) encompassing a.a. 2402–2669, oligonucleotide/oligosaccharide-binding domains (OBs) (OB) encompassing a.a. 2670–2803, 2809–3048, and 3056–3102, and a C-terminal RAD51-binding domain (CTD) encompassing a.a. 3270–3305. The function of each variant was estimated as a variant-specific effect value, ηv, based on a two-component model. Data of niraparib are shown as a representative case. The upper and lower components correspond to normal function and loss-of-function, respectively. The dotted lines indicate the median value of each component. The classification of each variant based on the Bayes factor (BF) is indicated by the color and shape of the lines and plots, as shown in the legend. Variant functions were classified into five classes by the MANO-B method: fClass 1 (normal) (BF ≤ 0.003), fClass 2 (likely normal) (0.003 < BF ≤ 0.053), fClass 3 (intermediate) (0.053 < BF < 18.7), fClass 4 (likely abnormal) (18.7 ≤ BF < 350), and fClass 5 (abnormal) (350 ≤ BF). N = 3 cells were examined over 2 or 3 independent experimental batches. Plots indicate mean values. Open error bars, 95% CIs. WT, wild-type; NS, nonsense variants.
