Optimisation strategies for directed evolution without sequencing

Directed evolution can enable engineering of biological systems with minimal knowledge of their underlying sequence-to-function relationships. A typical directed evolution process consists of iterative rounds of mutagenesis and selection that are designed to steer changes in a biological system (e.g. a protein) towards some functional goal. Much work has been done, particularly leveraging advancements in machine learning, to optimise the process of directed evolution. Many of these methods, however, require DNA sequencing and synthesis, making them resource-intensive and incompatible with developments in targeted in vivo mutagenesis. Operating within the experimental constraints of established sorting-based directed evolution techniques (e.g. Fluorescence-Activated Cell Sorting, FACS), we explore approaches for optimisation of directed evolution that could in future be implemented without sequencing information. We then expand our methods to the context of emerging experimental techniques in directed evolution, which allow for single-cell selection based on fitness objectives defined from any combination of measurable traits. Finally, we explore these alternative strategies on the GB1 and TrpB empirical landscapes, demonstrating that they could lead to up to 19-fold and 7-fold increases respectively in the probability of attaining the global fitness peak.Author summary: The standard approach to sorting-based selection in directed evolution is to take forward only the top-performing variants from each generation of a single population. There are, however, many possible approaches to exploring non-convex evolutionary fitness landscapes, and choosing this strategy as default may not always be the strongest approach. In this work, we begin to explore alternative selection strategies within a simulated directed evolution framework. We propose “selection functions”, which allow us to tune the balance of exploration and exploitation of a fitness landscape, and we demonstrate that splitting a population into sub-populations can improve both the likelihood and magnitude of a successful outcome. We also propose strategies to leverage emerging selection methods that can implement single-cell selection based on any combination of measurable traits. We finally assess the space of alternative directed evolution strategies on the empirical fitness landscapes of the GB1 immunoglobulin protein and of TrpB tryptophan synthase. Our resulting proposal is that researchers should consider moving away from the standard approach, which we find to be generally sub-optimal, and implement population splitting to improve experiments. With improved knowledge from fitness landscape inference, directed evolution strategies could be further tailored using the tools proposed here.