Comprehensive in Silico Reclassification of MECP2 Variants of Uncertain Significance in Rett Syndrome: Performance Evaluation and Structural Analysis

Abstract

Rett syndrome (RTT) is a severe neurodevelopmental disorder primarily caused by missense variants in the MECP2 gene. However, the presence of variants of uncertain significance (VUS) poses major challenges for clinical diagnosis and genetic counseling. In this study, we systematically evaluated the performance of 33 in silico prediction tools using a curated ClinVar dataset of MECP2 missense variants. Performance metrics included accuracy, sensitivity, specificity, area under the curve (AUC), and Matthews correlation coefficient (MCC), incorporating gene-specific pathogenicity thresholds to enhance predictive precision. Evolutionary conservation was assessed using ConSurf, while structural consequences were examined using UniProt, HOPE, DUET, PyMOL, and RING. Nine top-performing tools—MutPred, MetaRNN, REVEL, MutScore, SNPred, BayesDel, ClinPred, AlphaMissense, and DeepSAV—achieved accuracies exceeding 91% and correctly classified all 19 functionally validated pathogenic variants. These tools consistently predicted 15 VUS as pathogenic, with 14 located within the methyl-CpG-binding domain (MBD) and one within the NCOR2/SMRT interaction region; all occurred at highly conserved residues (ConSurf score: 9). Structural analyses revealed destabilizing effects through altered hydrophobicity, electrostatic charge, and residue interactions, implicating impaired DNA binding or disrupted co-repressor interactions. This integrative framework, combining high-performance computational prediction with structural modeling, offers a robust approach to reclassifying MECP2 VUS and supports improved diagnostic accuracy and personalized care in RTT.