Optimal compound downselection to promote diversity and parallel chemistry

Abstract: Early stage drug discovery and molecular design projects often follow iterative design-make-test cycles. The selection of which compounds to synthesize from all possible candidate compounds is a complex decision inherent to these design cycles that must weigh multiple factors. We build upon the algorithmic downselection framework SPARROW that considers synthetic cost, synthetic feasibility, and compound utility, extending it to address additional critical factors related to the risk of synthesis failure, molecular diversity, and parallel chemistry capabilities. These design considerations further align algorithmic compound selection with the true complexity of this decision-making process, allowing SPARROW to capture a broader set of principles typically reliant on expert chemist intuition. The application of these formulations to an exemplary case study highlights SPARROW's ability to promote the selection of diverse batches of compounds whose syntheses are amenable to parallel chemistry.