Dynamic Network Context Predicts Mutation-Induced Reshaping of Protein Conformational Free-Energy Landscapes

Missense substitutions can reshape protein conformational free-energy landscapes, but predicting such effects at scale remains challenging. We develop a mutation-impact framework based on the internal-coordinate, essential-dynamics-refined elastic network model (ICed-ENM). Changes in vibrational entropy quantify substitution-level dynamic perturbation; aggregation across substitutions and neighboring positions yields a position-level dynamic sensitivity score. Across three conformationally diverse proteins, alanine substitutions at high-sensitivity sites change molecular-dynamics-derived conformational landscapes more than low-sensitivity controls selected in broadly comparable structural contexts. Analysis of more than 560,000 modeled substitutions reveals systematic relationships with side-chain volume, amino-acid identity, solvent accessibility, and secondary structure. In a PTEN functional benchmark, ICed-ENM position-level scores outperform those of two comparator methods in distinguishing activity-selective loss (reduced phosphatase activity with preserved intracellular abundance) from abundance-related loss, achieving an area under the receiver operating characteristic curve of 0.745. Here, we show that structure-derived dynamic sensitivity provides a functionally informative basis for prioritizing mutations.