What is Binding Affinity?

Binding affinity is governed by the thermodynamics of the molecular interaction. When a ligand binds to a protein, the binding free energy (delta-G) determines the equilibrium between bound and unbound states. This free energy is the sum of favorable contributions (hydrogen bonds, electrostatic interactions, van der Waals contacts, hydrophobic effects) minus unfavorable contributions (conformational entropy loss, desolvation penalties). For a drug to be effective, it must bind its target with sufficient affinity to elicit a biological response at achievable plasma concentrations. Most approved drugs have Kd values between 1 nM and 1 microM for their primary target.

Peptide therapeutics achieve binding affinity through different mechanisms than small molecules. Small molecules typically bind in deep, well-defined pockets on protein surfaces. Peptides, being larger and more flexible, can engage extended surface areas including protein-protein interaction interfaces that are flat and featureless — the so-called 'undruggable' targets. A peptide can make simultaneous contacts across a large binding surface, distributing the binding energy across many weak interactions that collectively produce strong affinity. Techniques like cyclization (constraining the peptide into a ring) and stapling (adding chemical cross-links) can pre-organize the peptide's conformation to match the target surface, significantly improving binding affinity by reducing the entropic cost of binding.

In PepFold's 10-dimensional scoring system, binding affinity is a core evaluation dimension. The system uses computational docking and energy scoring to estimate the predicted binding strength between each peptide candidate and the target protein. The binding site geometry is informed by the patient's specific genetic variant — a missense SNP that changes an amino acid at or near the binding site alters the electrostatic and steric landscape that the peptide must complement. By designing peptides that match the variant-specific binding surface, PepFold generates candidates with higher predicted affinity for the patient's actual protein rather than the wild-type reference.