Jonathan Yaacov Weinstein

1 INTRODUCTION Over the past decade, high-throughput screening of protein variant libraries and deep sequencing have revolutionized protein characterization, engineering, and design (Fowler et al., 2010; Markel et al., 2020; Wójcik et al., 2015). Common library-synthesis approaches introduce random or systematic single-point mutations (Alejaldre et al., 2021).

Abstract Golden Gate assembly (GGA) can seamlessly generate full-length genes from DNA fragments. In principle, GGA could be used to design combinatorial mutation libraries for protein engineering, but creating accurate, complex, and cost-effective libraries has been challenging. We present GGAssembler, a graph-theoretical method to design DNA fragments for the most economical way to assemble a combinatorial library that encodes any desired diversity.

Library design Phylogenetic analysis A phylogenetic analysis was conducted as previously24,55,61 using all sequences in the lineage of avGFP according to FPBase35. A total of 153 sequences were retrieved from FPBase, all synthetic variants of avGFP. Briefly, sequences were clustered by cd-hit62 and aligned using MUSCLE63. The resulting multiple sequence alignment (MSA) was segmented by secondary structure elements.