Visualizing Deep Mutational Scan Data

Abstract

Site-directed and random mutagenesis are biochemical tools to obtain insights into the structure and function of proteins. Recent advances such as deep mutational scan have allowed a complete scan of all the amino acid positions in a protein with each of the 19 possible alternatives. Mapping out the phenotypic consequences of thousands of single point mutations in the same protein is now possible. Visualizing and analysing the rich data offers an opportunity to learn more about the effects of mutations, for a better understanding and engineering of proteins. This work focuses on such visualization analyses applied to the mutational data of TEM-1 {beta}-lactamase. The data is examined in the light of the expected biochemical effects of single point mutations, with the goal of reinforcing or retraining the intuitions. Individual attributes of the amino acid mutations such as the solvent accessible area, charge type change, and distance from the catalytic center capture most of the relevant functional effects. Visualizing the data suggests how combinations of these attributes can be used for a better classification of the effects of mutations, when independently they do not offer a high predictability.