Computational design of enzymatic pathways using the Act Synthesizer

Microbial chemical production provides the promise of biosynthesis of valuable small molecules with therapeutic, fuel, and material applications. We present a computational design tool, called the Act Synthesizer, that infers enzymatic pathways towards natural and unnatural small molecules. The tool takes as input repositories of known enzymatic data (BRENDA and KEGG) in addition to mining text from published literature for additional enzymes not covered by databases. Given a specific target molecule, it can then predict pathways--both natural and those made of enzymes from multiple organisms--to the target. We have found that these pathways contain previously unknown biosynthetic routes to unnatural molecules. We have assembled such a path, to N-acetyl-p-aminophenol (paracetamol), in E. coli and validated through LCMS that the predictions are indeed valid.

In addition to the design of pathways, the tool can also identify the risk and opportunity associated with the biosynthesizable chemicals. For instance, it can enumerate chemicals whose microbial production poses biosafety concerns, or those that are commercially expensive, or those that have known therapeutic applications.

We believe computational design tools like Act will be invaluable in the toolkit of a metabolic engineer looking to design microbes for novel biosynthesis applications.