Targets of Opportunity : Hypergraph Analysis of the Retrosynthetic Design Space

As biomanufacturing progresses towards the goal of producing a wide array of natural product compounds in vivo, many retrosynthetic design methods have emerged for finding and comparing pathways to a specific target compound.  Little attention, however, has been given analyzing the space of all potential natural product pathways accessible through retrosynthesis. Rather, studies focus on a specific target compound, selecting heterologous genes to express in a chassis organism in order to generate a synthetic natural product pathway.  This work examines the Retrosynthetic Design Space (RDS) containing all heterologous pathways which could potentially be added to a specific chassis organism.  By examining the entire RDS, we directly compare the costs and benefits of pathways across different target compounds rather than only between pathways for the same compound.  This analysis reveals which compounds are more easily accessible today and directs the efforts of the synthetic biology community to the most pressing challenges in biomanufacturing.

We present an analysis of the RDS for the chassis organism E. Coli constructed from publicly available databases of biochemical reactions and enzymes.  The RDS is represented as a hypergraph with metabolites as nodes and biochemical transformations as hyperedges. By traversing the hypergraph we determine the set of Pareto-optimal pathways which could be potentially added to the chassis organism and which target compounds these pathways would theoretically produce. By analyzing these pathways with graph-theoretic algorithms, we are able to compare targets with inter-pathway metrics, such as centrality and clustering, as well as intra-pathways metrics such as thermodynamics and toxicity. Specifically we: (1) Rank target compounds and compound classes by difficulty using multiple scoring objectives, including, pathway length, pathway thermodynamics, enzyme availability, intermediate toxicity, and RDS centrality. (2) Identify hub compounds in the RDS which are key intermediates for multiple target pathways, including key host metabolites which which serve as precursors. (3) Identify gaps in the RDS for which further investigation could unlock additional natural product targets. From this analysis we identify "targets of opportunity" which are not only easier to reach, but also serve as intermediates for producing more difficult compounds in the future.