In Silico Design and Analysis of Cyanobacterial Pseudo Natural Products

Abstract

Marine cyanobacteria produce natural products (NPs) with potent and selective bioactivity against a broad range of diseases. However, like many NPs, most exhibit poor drug-like physicochemical properties, and the discovery of structurally novel NPs is declining. To address these challenges, we generated an in silico library of 2,415 cyanobacterial pseudo-NPs by tethering cyanobacterial NP fragments with privileged scaffolds from noncyanobacterial NPs via hypothetical amide bond formation. This library was analyzed using computational platforms to assess predicted physicochemical and ADME/Tox properties, lead-likeness penalties, NP-likeness scores, Tanimoto similarity coefficients, and Synthetic Accessibility Scores. Comparisons to public compound libraries showed that most cyanobacterial pseudo-NPs possess favorable drug- and lead-like characteristics; occupy low-density chemical space; and display unique, synthetically accessible scaffolds. Our results suggest that these pseudo-NPs are promising synthetic targets for drug development. Moreover, this platform can be expanded by using artificial intelligence (AI)-based fragment harvesting tools to create larger libraries of NP-inspired compounds. By integrating cyanobacterial fragments with known bioactive motifs, we aim to bridge the gap between natural diversity and drug-like properties, providing a novel and tractable chemical space for drug discovery efforts.