Automating bioengineering: First the hands, then the head
March 4-8, 2018
Amyris was founded over a decade ago on the premise that synthetic biology could help address some of the world’s most pressing challenges. Focusing on supply-limited molecules of societal and economic value such as medicines, nutrients, and commodity & specialty chemicals, Amyris has enabled their cost-effective and sustainable production by building a high-throughput genetic engineering platform coupled to industrial scale fermentation processes. Until 2015 this platform primarily aimed to replace the “hands” in the lab, using automation & computing to perform routine calculations and liquid manipulations for the reproducible construction of engineered microbes. Since then we have focused on replacing the “head” in the lab by implementing a suite of computational algorithms to design metabolic pathways, direct their synthesis and phenotyping, learn from the data, and iterate on the design for improved target molecule production. In this presentation I will describe our progress on joining the hands to the head by attempting to create discrete strains that produce 450 novel molecules in 24 months. This effort involved first creating an expert system that successfully directed the creation of >110,000 in silico strain designs, identification of >14,000 metabolic routes, ordering of >2,400 genes, assembly of >280 million bps of DNA, and generation of >32,000 distinct strains. We are now collecting >400,000 data points each week, measurements that are fed back into proprietary machine learning and optimization algorithms to drive strain re-design.
Ben Kaufmann-Malaga, Amoolya Singh, and Joel R. Cherry, "Automating bioengineering: First the hands, then the head" in "Microbial Engineering", Prof. Eli Keshavarz-Moore , University College London, England Dr. Barry Buckland, BiologicB, USA Eds, ECI Symposium Series, (2018). http://dc.engconfintl.org/microbial/45