Fig. 2

NCBRS-SMARCA2 DNAm signature classifies variants of uncertain significance. a The heatmap shows the hierarchical clustering of NCBRS cases (n = 8) and age- and sex-matched neurotypical controls (n = 23) using 429 differentially methylated CpG sites specific to SMARCA2 pathogenic variants. The color gradient indicates the β (DNAm) value ranging from 0.0 (blue) to 1.0 (yellow). DNAm profiles fall into two separate clusters corresponding to NCBRS cases (orange) and controls (cyan). Euclidean distance metric is used in the clustering dendrogram. b Classification model based on DNAm signature. The median-methylation profile for signature-derivation NCBRS cases (n = 8) and controls (n = 23) were calculated at the CpG sites comprising the NCBRS-SMARCA2 DNAm signature. The Pearson correlation of each sample with the median profile of controls and that of NCBRS cases are plotted on the x- and y-axes respectively. The difference of these correlations constitute the NCBRS-SMARCA2 score. Positive NCBRS-SMARCA2 scores (pathogenic) fall above the decision boundary (red line) and negative (benign) fall below it. Additional neurotypical control whole-blood samples (n = 94; Control Validation) all classified as benign. Additional NCBRS cases with pathogenic SMARCA2 variants (n = 8; NCBRS Validation; GSE116992) classified as pathogenic. SMARCA2 variant test cases (n = 9; grey squares with SMARCA2_IDs denoted) were scored: three were classified as pathogenic (SMARCA2_4, SMARCA2_10, and SMARCA2_14), five were classified as benign (SMARCA2_15–19, IDs not shown), and one was classified as benign but its score was near 0, falling close to the decision boundary (SMARCA2_12). c Schematic of the SMARCA2 amino acid sequence with NCBRS signature cases and SMARCA2 test variants indicated. Numeric labels indicate sample IDs corresponding to those in (b)