Computational Investigations to the Physical Features Contributing to Structural Perturbation in Experimental Protein Structures

Variations to the amino acid sequences of proteins are known to have potential to alter the native fold of the protein and result in deleterious effects and observable disease phenotypes. While there has been much success in computational prediction of experimental protein structure with the advent of AlphaFold it has been shown that these prediction methods are inaccurate in predicting the structure and properties of mutant proteins. Using a curated dataset of experimental single amino acid variants and duplicate protein structures single mutations capable of causing significant structural perturbation have been identified in experimental data. Structural perturbation is calculated using two measures: Superimposition of a wild type and mutant structure to calculate the position deviation of all residues and the difference in solvent accessible surface area for all residues between the wild type and mutant structure. Mutations causing significant structural perturbation are found to contain a set of residues that are also significantly perturbed around the mutation site in 3 - dimensional space. This set of significantly perturbed residues define the neighborhood of residues for specific mutations and the physical properties that can be used to describe the structural perturbation of the mutation.