In our preprint released last Friday, we show that evolutionary sequence information learned by our genomic language model, gLM2, can be used to design complex, multi-domain enzymes and expand their chemistry beyond their natural repertoire. Practically, this could enable: New therapeutics. More than half of FDA-approved small-molecule drugs over the past four decades are natural products, their derivatives, or synthetic mimics. Redirecting biosynthetic assembly lines toward new molecular targets could substantially expand the chemical space accessible through biological synthesis. Sustainable chemistry. Redesigning biosynthetic assembly lines to produce materials and commodity chemicals from biology rather than petrochemical feedstocks could make chemical manufacturing more sustainable while also enabling molecules and materials with properties that are difficult to access through conventional synthesis. Preprint: biorxiv.org/content/10.64898…

We're excited to share our new preprint! In collaboration with the @Keasling_Lab at UC Berkeley, we made gLM2 (our genomic language model) generative and used it to redesign a chimeric Type I polyketide synthase in the context of the full biosynthetic assembly line. The PKS is a 2,488-amino-acid molecular assembly line composed of eight protein domains that must work together to carry out coordinated, multi-step catalysis. Instead of optimizing one component at a time, gLM2 designs sequences in the context of the entire biosynthetic system. Experimentally, gLM2-designed variants increased production of δ-valerolactam, a molecule not naturally synthesized by PKSs, by up to 9.4× over the starting enzyme. We think this is an exciting step for genomic language modeling: from learning the organization of natural biological systems to controllably designing biosynthetic machinery that can access chemical space beyond nature’s native repertoire. This work was an incredible collaborative effort with @jaykeasling and the @jbei team, the @KortemmeLab, and an amazing group of coauthors. Huge thanks to everyone who made it possible. Preprint: biorxiv.org/content/10.64898… Authors: @nathanlanclos_, @ibrahimovicious, @ancornman1, Marco Huang, Vikram Gill, Aalini Jiang, Jonathan Abraham, Jennifer Gin, Yan Chen, Christopher Petzold, Justin Baerwald, Tanja Kortemme (@KortemmeLab), @jaykeasling, @Micro_Yunha This work was funded by @DARPA, @SchmidtFutures, and @NSF.