Virtual screening is a powerful component of CrystalsFirst’s discovery engine, allowing us to explore vast chemical space in silico and pinpoint promising candidates efficiently. Our computational chemistry employs advanced molecular modeling, docking simulations, and molecular dynamic simulations to evaluate large libraries of compounds against high-value targets. By leveraging high-quality protein structures (from crystallography or CryoEM), our virtual screens include experimental details for the set up of virtual screening protocols.
We incorporate into our virtual screening pipeline the DSX scoring function. DSX is a knowledge-based scoring method that refines protein–ligand docking results using high-resolution structural data from crystallographic complexes. Unlike physics-based scoring, DSX integrates statistical potentials derived from experimental binding geometries, offering superior performance in predicting binding affinity and pose accuracy.
We also incorporate FragAI generative insights to prioritize molecules with optimal binding profiles and novelty, going beyond traditional docking to suggest truly innovative chemotypes. This in silico screening process is fully integrated with our experimental workflows. Hits identified computationally are rapidly tested and validated via SmartSoak® crystallographic screening or biophysical assays, creating a seamless feedback loop between virtual and physical screening. Such coupling of virtual and real data accelerates hit discovery and reduces false positives. Additionally, we can perform virtual pre-screening of multiple targets to triage the best candidates for lab experiments, optimizing resource use. CrystalsFirst’s AI-enhanced virtual screening ultimately broadens the search for viable drug molecules while saving significant time and cost in early discovery.