Pre-Conference Workshop Day

Structural biology is generating multiple layers of information across cryo-EM, X-ray crystallography, HDX, SAXS, SPR and biochemical assays, but these datasets are often handled separately, making it difficult to build a complete picture of protein-ligand interactions and confidently guide drug discovery decisions. This technical workshop will explore how to bring together complementary structural and experimental datasets to generate more actionable insights by:

• Integrating complementary structural biology techniques to understand how cryo-EM, X-ray crystallography, HDX and SAXS can be combined to capture protein plasticity, conformational dynamics and protein-ligand interactions
• Connecting structural, biophysical and biochemical readouts to reconcile discrepancies between indirect binding assays and direct structural observations, and improve confidence when prioritizing binders and validating mechanisms
• Building multi-modal data workflows for SBDD to break down data silos, connect experimental and computational insights, and translate complex datasets into more confident decisions across hit identification, validation and molecular optimization

Accurately predicting how molecules bind to proteins remains one of the major unsolved challenges in structure-based drug design, with current approaches struggling to rank binding affinities, predict accurate poses and design proteins with sufficiently high success rates. This technical workshop will explore how AI, physics-based modeling and experimental validation can be combined to improve predictive performance by:

• Improving small-molecule docking and binding pose prediction by examining the limitations of current approaches, including challenges with homologous versus novel structures and the difficulty of accurately modeling protein-ligand interactions
• Advancing affinity prediction and free-energy calculations to better rank compounds by exploring how AI and physics-based approaches can address complex parameters such as protein flexibility, solvent and water interactions that limit current predictive accuracy
• Strengthening de novo protein design and computationally designed binders by examining how prediction models can be improved, experimentally validated and coupled with high-throughput screening to increase the success rate of designed proteins
• Benchmarking predictive models against experimental data to identify which AI/ML approaches genuinely improve SBDD outcomes and where expert oversight and wet-lab validation remain essential