Elucidating Which Pairwise Mutations Affect Protein Stability: An Exhaustive Big Data Approach

The specific sequence of amino acids in a polypeptide chain dictates the three dimensional structure, and hence function, of a protein. Mutagenesis experiments on physical proteins involving amino acid substitutions provide insights enabling pharmaceutical companies to design medicines to combat a variety of debilitating diseases. However such wet lab work is prohibitive, because even studying the effects of a single mutation may require weeks of work. Computational approaches for performing exhaustive screens of the effects of single mutations have been developed, but methods for conducting a systematic, exhaustive screen of the effects of all multiple mutations are not available due to the large number of mutant protein structures that would need to be analyzed. In this work we motivate and demonstrate a proof of concept approach for conducting in silico experiments in which we generate all possible mutant structures with 2 amino acid substitutions for three proteins with 46, 67, and 99 residues; for the largest protein we in silico generate 1,751,211 mutants. We leverage an efficient combinatorial algorithm to assess the effects of the mutations among the mutant protein structures. We also produce heat maps for several mutation metrics to facilitate identifying which pairs of amino acid in a protein have the greatest impact on protein stability based on how those amino acid substitutions affect the protein's flexibility.