Deep Learning-Based Virtual Screening Identifies Potential Small-Molecule Inhibitors of ASK1: Natural Product Lead Discovery for MASH.

Apoptosis signal-regulating kinase 1 (ASK1) represents a critical therapeutic target for metabolic dysfunction-associated steatohepatitis (MASH). Natural products, owing to their unique chemical diversity, constitute a rich reservoir for discovering novel ASK1 inhibitors. The emergence of artificial intelligence-assisted drug discovery (AIDD) has opened new avenues for exploring small-molecule inhibitors. Through virtual screening, molecular docking, interaction profiling, molecular dynamics simulations, and MM-GBSA binding free energy calculations, we systematically evaluated the binding mode, stability, and key residue contributions of the CMNPD10921-ASK1 complex. CMNPD10921 stably occupied the ASK1 active pocket, forming hydrophobic interactions and hydrogen bonds with multiple key amino acid residues. MM-GBSA analysis yielded a total computed binding free energy of -31.15 kcal/mol, suggesting a computationally favorable interaction, with van der Waals forces serving as the dominant energetic driver of complex stabilization. Residue energy decomposition further identified ILE324, THR288, and THR639 as major contributors to ligand binding. Integrating deep learning, molecular simulation, and quantum chemical calculations, this study successfully identified CMNPD10921 from a vast natural product library as a putative lead compound candidate targeting the ASK1 central regulatory region, offering a novel candidate molecule for anti-MASH drug development.