Learning the pattern of epistasis linking genotype and phenotype in a protein

Abstract

Understanding the pattern of epistasis—the non-independence of mutations—is critical for relating genotype and phenotype. However, the combinatorial complexity of potential epistatic interactions has severely limited the analysis of this problem. Using new mutational approaches, we report a comprehensive experimental study of all 213 mutants that link two phenotypically distinct variants of the Entacmaea quadricolor fluorescent protein—an opportunity to examine epistasis up to the 13th order. The data show the existence of many high-order epistatic interactions between mutations, but also reveal extraordinary sparsity, enabling novel experimental and computational strategies for learning the relevant epistasis. We demonstrate that such information, in turn, can be used to accurately predict phenotypes in practical situations where the number of measurements is limited. Finally, we show how the observed epistasis shapes the solution space of single-mutation trajectories between the parental fluorescent proteins, informative about the protein’s evolutionary potential. This work provides conceptual and experimental strategies to profoundly characterize epistasis in a protein, relevant to both natural and laboratory evolution.

Known Evolutionary Paths Are Accessible to Engineered ß-Lactamases Having Altered Protein Motions at the Timescale of Catalytic Turnover

“…However, exhaustive investigation of combinatorial mutations is resource intensive and out of reach for the average-sized protein (a 50 amino acid protein requires a library of 20 50 = 1.1 × 10 65 variants). Therefore, focused combinatorial libraries and directed molecular evolution experiments are currently the most useful approach to uncover epistatic interactions (Acevedo-Rocha et al, 2014;Sato et al, 2016;Poelwijk et al, 2019).…”

Inferring a complete genotype-phenotype map from a small number of measured phenotypes

“…The additive model, classifier, and nonlinear correction left 19% of the quantitative variation unaccounted for (the scatter from the line in Fig 2C ). We next sought to account for the remaining scatter by incorporating epistatic interactions between specific mutations [ 45 , 47 – 50 ]. We added pairwise interactions to the model and then fit the model parameters using lasso regression [ 50 , 51 ].…”

“…We next sought to account for the remaining scatter by incorporating epistatic interactions between specific mutations [ 45 , 47 – 50 ]. We added pairwise interactions to the model and then fit the model parameters using lasso regression [ 50 , 51 ]. This method uses L1 regularization to penalize the addition of unneeded parameters, thereby minimizing the number of new parameters included.…”

DiMSum: an error model and pipeline for analyzing deep mutational scanning data and diagnosing common experimental pathologies

“…A key feature of a DMS experiment is that it preserves the link between quantitative phenotypic effects and their underlying causal genotypes measured for many variants simultaneously ( Figure 1a). The three main steps can be summarized as follows: (1) construction of a library of DNA variants corresponding to the assayed biomolecule (genotype), (2) selection (or separation) of variants according to a given molecular function (phenotype), and (3) quantification of the variant abundances before and after selection by DNA sequencing (measurement), which is either done by counting sequencing reads of variants directly or unique barcodes previously linked to them [29][30][31][32][33] . A fitness score for each variant is then calculated by comparing its relative abundance (with respect to a reference sequence e.g.…”