S1: Understanding and Engineering Enzymes via Recombination

When facing an enzyme engineering challenge suitable for directed evolution, one must select a method to diversify the target enzyme. Relative to mutagenesis, recombination results in a very different family of synthetic sequences: a family of chimeras. These libraries can be highly diverse, yet contain a high fraction of functional enzymes. Site-specific recombination results in a library of proteins comprised of distinct sequence blocks. As illustrated by several examples (cytochrome P450, cellbiohydrolase, and endoglucanase), this type of library offers unique advantages in interpretability. Given a small training set (tens of proteins) it is possible to create predictive high-resolution models for protein structure as well as functional properties such as thermostability. While directed evolution is often successful in the absence of predictive models, such models can be extremely useful (e.g. given a limited screening capacity).