Evaluating Genetic Variants of Unknown Significance Using Protein Thermodynamic Simulations in the Context of Non-Syndromic Hearing Loss

Mallory R. Tollefson, University of Iowa


Accurate classification of missense variants as pathogenic or benign is a crucial step in translational genetic research. The most widely used prediction algorithms are based on sequence conservation, however, they often yield ambiguous or even contradictory results. Computational molecular biophysics can provide a complementary approach by incorporating three-dimensional protein structural data and ab initio thermodynamic principles into variant effect prediction. In this work, we apply Free Energy Perturbation (FEP) to calculate the mutational folding free energy (ΔΔGWT>Variant) of 28 variants across three genes that are associated with hearing loss. We hypothesized that a significant (>1 kcal/mol) free energy change, indicative of significant whole-protein destabilization and misfolding, correlates with pathogenicity. Assessment of this hypothesis was tested using thermodynamic predictions on eleven positive and negative control variants of known status, i.e. those known to be benign or pathogenic. FEP calculations on positive and negative controls showed agreement. We then evaluated 17 additional variants of uncertain significance, i.e. variants classified specifically as VUS, likely pathogenic, likely benign. We conclude from this study that further investigation of protein thermodynamics as a tool for clarification of variant pathogenicity in hearing loss is warranted. We expect computational protein thermodynamics to prove advantageous in the analysis of genetic data associated with other disease states as well.