TY  - JOUR
T1  - Reaction Prediction Can Change How Natural Product Scaffolds Are Optimized by Connecting Biosynthetic Logic, Semisynthesis, and Medicinal Chemistry Search
A1  - Andrei Popescu
A1  - Mihai Ionescu
A1  - Elena Stan
A1  - Cristina Radu
JF  - International Journal of Pharmaceutical And Phytopharmacological Research
JO  - Int J Pharm Phytopharmacol Res
SN  - 2250-1029
Y1  - 2026
VL  - 16
IS  - 3
DO  - 10.51847/fkzzC6Pd5l
SP  - 147
EP  - 155
N2  - Natural products offer chemically differentiated scaffolds, but their optimization is often framed as a problem of choosing desirable analogues rather than identifying the reactions that make those analogues reachable. This distinction matters because biosynthetic history, pre-existing stereochemical complexity, functional-group density, site selectivity, protecting-group requirements, and route context can sharply restrict which structural changes are experimentally plausible. This article develops an original methodological framework in which reaction prediction is positioned between natural-product scaffold representation and analogue prioritization. The framework integrates three otherwise partly separated sources of chemical knowledge: biosynthetic transformations as context-bearing priors, semisynthetic reactions that exploit accumulated scaffold complexity, and medicinal-chemistry reaction data that provide broader precedent for synthetic modification. It argues that predicted reactions should be evaluated not only by model likelihood but also by stereochemical and chemoselective compatibility, condition plausibility, route burden, analogue diversity, and downstream pharmacological value. Multistep routes, model failure, negative reaction evidence, and property prediction are treated as distinct analytical layers rather than collapsed into a single synthetic-accessibility score.

The principal contribution is a proposed reaction-aware search architecture for moving from an available natural-product scaffold to evidence-qualified analogue families. The framework is not presented as prospectively validated, and its usefulness will depend on scaffold-specific data coverage, representation quality, uncertainty calibration, route-level assessment, and experimental testing. Its purpose is to redirect optimization from asking which structures are attractive to asking which chemically meaningful transformations can make valuable chemical space reachable.
UR  - https://eijppr.com/article/reaction-prediction-can-change-how-natural-product-scaffolds-are-optimized-by-connecting-biosyntheti-qouc92f6hclsycq
ER  -