Diversity in cellular metabolism offers vast potential to produce large numbers of chemicals. Yet, conventional metabolic engineering requires multiple design-build-test-learn (DBTL) cycles; hence, it is costly and laborious to create optimal strains that can explore such a large chemical space and achieve high product titers, rates, and yields. To address this challenge, a modular cell design (ModCell) concept has been proposed where strains can be designed systematically that require a minimal number of DBTL cycles using a multi-objective optimization framework derived from the mass balance principle, Pareto front optimization theory, and omics data. The design takes advantage of the modularity that exists in biological systems and utilizes different pathways that can be divided into tractable and exchangeable modules to be compatible with a modular (chassis) cell for optimal biosynthesis of a large library of products. To implement this design efficiently, we developed ModCell3, a software package driven by evolutionary algorithms and multi-objective optimization. Utilization of omics data enables us to create enzyme-constrained genome-scale models for ModCell3 which further improves the model constraints, and accounts for resource allocation (e.g., proteome allocation) and metabolic cost (e.g., ATP cost) for different production phenotypes. Different from conventional single-product strain designs, ModCell3 can systematically identify genetic modifications to design a modular cell(s) that can be coupled with a variety of production modules, exhibit a minimal tradeoff among modularity, performance, and robustness, predict resource reallocation, and alleviate metabolic cost associated with various production phenotypes. We vision ModCell3 to be a useful tool to fundamentally study modular designs in natural and synthetic biological systems and harness it for biomanufacturing.
