In Silico Design and Analysis of Cyanobacterial Pseudo-Natural Products

Marine cyanobacteria give rise to natural products (NPs) that exhibit potent and selective bioactivity towards a broad spectrum of diseases. Nonetheless, like many other NPs, most of these compounds suffer from poor drug-like physicochemical properties while the percentage of structurally novel NPs steadily declines. To address this, we have created a library in silico of 2,415 cyanobacterial pseudo-NPs, where cyanobacterial NP fragments were tethered with privileged scaffolds from a variety of non-cyanobacterial NPs via a hypothetical peptide bond formation. Our pseudo-NP library was analyzed using various computational platforms to provide predicted physicochemical properties, lead-likeness penalties, NP-likeness scores, and Tanimoto coefficients when compared to publicly available compound libraries. Upon comparative analysis, our results demonstrate that most of the cyanobacterial pseudo-NPs created are predicted to possess drug- and lead-like properties, populate low density chemical space in terms of their NP-likeness, and provide structurally unique scaffolds when compared to publicly available small-molecule databases. We predict that our cyanobacterial pseudo-NP library will provide suitable synthetic targets for drug discovery and development, and a platform to be expanded upon using AI fragment harvesting tools for the creation of larger pseudo-NP libraries, creating synthetically tractable NP-like compound for drug discovery.