Developing a Design Guideline of Boronic Acid Derivatives to Scavenge Targeted Sugars in the Formose Reaction Products using DFT-based Machine Learning

Abstract Formose reaction facilitates the synthesis of sugars from formaldehyde, yet the valuable sugars constitute only a small portion of the total products. This necessitates the need for a chemical scavenger capable of selectively capturing only valuable sugars. With over 600,000 potential combinations of boronic acid-based scavengers available, pursuing a deductive search approach is unfeasible. This study aims to derive guidelines for designing scavengers that readily bind with target sugars while avoiding nontarget ones via machine learning informed by density functional theory calculations.