Identification of potent inhibitors targeting the pre-fusion DENV envelope protein: A consensus physics-and AI-based in silico multi-tier screening approach.

Dengue virus, a well-known arbovirus, causes devastating effects in various parts of the globe, with an infection rate of 300-400 million annually, and nearly half of the world's population remains at risk of acquiring dengue infection. Currently, there are no specific approved drugs against the dengue virus, and existing vaccines have significant limitations and usage restrictions. Therefore, the search for potent drugs against the dengue virus remains a vital approach to control disease spread. Hence, in this study, we focused on developing a potent inhibitor of the dengue envelope protein using a synergistic scoring approach that integrates conventional physics-based (Glide docking) and AI-based (GNINA) scoring functions to select high-affinity compounds rigorously. This work uses a diverse set of natural product databases (TCM, COCONUT, IMPPAT, CMNPD, and SWMD) for a virtual screening campaign targeting the β-OG pocket of the dengue virus prefusion envelope protein. A consensus scoring approach was employed to rank the top lead candidates that consistently ranked favourably across three distinct scoring methods: the Glide XP score, AutoDock Vina score, and the GNINA CNN-based score. The compounds TCM44233, CNPD7484, IMPPAT5321435, CMNPD28734, and RL148 demonstrated an optimal balance of strong binding affinity and favourable ADMET profiles. Among the top candidates, CNPD7484 demonstrated superior stability within the binding pocket, as evidenced by extended microsecond simulations, MM/PBSA, and essential dynamics analyses. Based on this rigorous in silico evaluation, CNPD7484 is identified as a potent inhibitor of the dengue virus envelope protein and warrants further experimental validation.