Not all interface residues contribute equally to binding affinity. A small subset, the binding "hotspots", accounts for the majority of the binding free energy. Identifying these residues is the first step in designing a binder that works.
Alanine scanning mutagenesis, systematically mutating each interface residue to alanine and measuring the change in binding free energy (ΔΔG), is the gold standard for hotspot identification. Computational alanine scanning accelerates this process by orders of magnitude:
- Rosetta-based scanning: the
ddg_monomer and flex_ddG protocols estimate ΔΔG values by sampling side-chain rotamers and backbone relaxation around each mutation.
- FoldX force field: an empirical energy function calibrated against experimental ΔΔG databases, providing rapid and reasonably accurate predictions.
- Consensus approaches: combining predictions from multiple methods improves accuracy and reduces false positives through predictor agreement.
Static structures miss the dynamic nature of protein interfaces. Molecular dynamics (MD) simulations reveal:
- Conformational heterogeneity at the interface: residues that appear well-packed in a crystal structure may be mobile in solution.
- Water-mediated contacts: bridging water molecules that contribute to binding but are invisible to static scoring functions.
- Transient pockets: cryptic binding sites that open during MD trajectories but are absent in the starting structure.
Once hotspots are identified, they become design constraints for the binder generation pipeline:
- Each hotspot residue defines a spatial constraint: the binder must present a complementary side chain within a specified distance and angle tolerance.
- Hotspot geometry, including side-chain orientation and local backbone conformation, constrains the binder scaffold selection.
- Multi-hotspot designs require simultaneous satisfaction of all constraints, which exponentially increases the design complexity but also dramatically improves specificity.
WeaveSeq's hotspot identification pipeline integrates computational alanine scanning, short MD refinement, and consensus scoring to produce a ranked hotspot map for each target. This map feeds directly into the binder generation engine, ensuring that every design targets the energetically most important interface residues.
Explore our open design campaigns to see hotspot analysis applied to real targets. View open targets
Related reading: our binding-site fingerprinting benchmark, geometric complementarity in interface engineering, and the architecture of de novo protein binders.
