(592e) Optimization of Gene Copy Number Combination in Metabolic Pathway By Machine Learning

Producing biochemicals by microorganisms is a green and economical approach. Improving gene expression level is a commonly method to efficiently produce single gene products. However, for multi-gene pathway, it is a time-consuming and laborious work to explore the optimal multi-gene copy number combination to maximize metabolic flux. To overcome these challenges, we introduced machine learning method to rationally optimize the copy number combination of different genes in multi-gene pathway. As a proof of concept, we obtained a high-performing engineered Y. lipolytica strain for eicosapentaenoic acid (EPA) biosynthesis in 6 months, producing the highest titer of 27.5 g/L in a 50 L bioreactor. Moreover, the lycopene production in E. coli was also greatly improved, yeilding in an increased lycopene titer of 9.5-fold. The combination of machine learning and synthetic biology has implications for the production of chemicals by microorganisms.