Multi-target computational pipeline for discovery of pan-influenza neuraminidase inhibitors

The continuous evolution of influenza A and B viruses, coupled with the emergence of drug resistance, creates a pressing need for novel antiviral agents with broad-spectrum activity. The viral neuraminidase enzyme remains a prime target, but its structural variability across different strains complicates the discovery of universal inhibitors. To address this challenge, we developed and implemented a multi-target computational pipeline designed to identify pan-influenza neuraminidase inhibitors. Our strategy involved high-precision molecular docking of a curated library containing 499,721 compounds against three structurally distinct neuraminidase representatives from influenza A (H1N1, H2N2) and influenza B viruses. Hits were prioritized using a cascade of energetic and geometric filters, followed by a rigorous two-tiered validation using extensive molecular dynamics simulations. This validation not only confirmed binding stability on the primary target but also critically assessed whether candidates maintained stable interactions across the other neuraminidase subtypes. This cross-validation approach was essential for eliminating subtype-specific binders, ultimately identifying ten compounds with robust, pan-influenza binding profiles. Notably, the successful identification of a diastereomer of the established drug zanamivir among the top candidates provides strong validation for the pipeline’s ability to find biologically relevant scaffolds. Overall, this work demonstrates the integration of multi-target screening with cross-validated molecular dynamics (cross-MD) that overcame target variability and yielded ten promising hits candidates for next-generation anti-influenza therapeutics.