From Feasible to Practical: Pareto-Optimal Synthesis Planning

Current computer-aided synthesis planning (CASP) methods often treat retrosynthesis as solved once a single feasible route is identified, focusing primarily on convergence or shortest-path metrics. This view is misaligned with real-world practice, where chemists must balance competing objectives such as cost, sustainability, toxicity, and overall yield. To address this, we formulate synthesis planning as a multi-objective search problem and introduce MORetro$^\ast$, an algorithm that generates a Pareto front of synthesis routes to explicitly capture trade-offs between user-defined criteria. MORetro$^\ast$ uses weighted scalarization and solution-informed sampling to efficiently navigate the combinatorial search space and prioritize promising trade-offs. Building on multi-objective A$^\ast$-search, we provide optimality guarantees showing that, for a fixed single-step model, MORetro$^\ast$ recovers the true Pareto front under admissibility. Across multiple retrosynthesis benchmarks, MORetro$^\ast$ produces diverse, high-quality Pareto fronts, uncovering solutions overlooked by single-objective approaches and better aligning CASP outputs with industrial decision-making.

When chemists want to make a new molecule such as a drug or a new material, they need a step-by-step plan to build it from ingredients they can buy. Finding even one feasible plan is hard, but in practice chemists also care about things like cost, environmental impact, and whether toxic chemicals are involved. Current algorithms typically stop once they find any valid plan, leaving chemists to figure out these trade-offs on their own. We developed a tool called MORetro* that instead presents chemists with a set of alternative plans, each representing a different compromise. One plan might be cheaper but use harsher chemicals; another might be greener but harder to scale up. Our tool maps out these trade-offs automatically, so chemists can make informed choices rather than settling for the first option a computer finds. We also show mathematically that our approach can find all the best compromises; it won't miss a promising option. Tested across multiple datasets, our tool consistently discovers better and more diverse plans than existing methods, helping bridge the gap between what computers suggest and what chemists actually need.