Natural Product-likeness Score and Its Application for Prioritization of Compound Libraries

TLDR

Natural products, refined by long evolutionary selection, provide validated substructures that, when analyzed with cheminformatics, aid drug discovery. The study proposes a Bayesian natural product‑likeness score to quantify how closely molecules resemble the natural product chemical space. The score is computed by a Bayesian model that evaluates structural similarity to known natural products. Cross‑validation shows the score distinguishes natural products from synthetic compounds, and it can be applied to virtual screening, library prioritization, and building‑block design.

Abstract

Natural products (NPs) have been optimized in a very long natural selection process for optimal interactions with biological macromolecules. NPs are therefore an excellent source of validated substructures for the design of novel bioactive molecules. Various cheminformatics techniques can provide useful help in analyzing NPs, and the results of such studies may be used with advantage in the drug discovery process. In the present study we describe a method to calculate the natural product-likeness scorea Bayesian measure which allows for the determination of how molecules are similar to the structural space covered by natural products. This score is shown to efficiently separate NPs from synthetic molecules in a cross-validation experiment. Possible applications of the NP-likeness score are discussed and illustrated on several examples including virtual screening, prioritization of compound libraries toward NP-likeness, and design of building blocks for the synthesis of NP-like libraries.