Ensemble-Conditioned Molecular Design

Molecular design is typically approached as a problem of finding molecules which can adopt a single bioactive conformation. In reality, molecules occupy a distribution over conformations, and many important properties depend on an ensemble rather than on any single conformer. We reframe molecular design as an optimisation of both the modes and properties of molecules' conformational ensembles, where modes can be represented as shapes, pharmacophore profiles or protein pockets, and properties are aggregate scalars computed over the whole distribution. To realise this we introduce ensemble-conditioned guidance, which conditions 3D molecular generative models on both axes by composing the vector fields produced under each condition at inference. Conditions may be targeted or avoided, mixed across modalities and combined in arbitrary numbers, so a wide range of design tasks can be expressed with a single trained model and no matched multi-condition training data. We evaluate on new benchmarks for multi-mode conditioning and ensemble property optimisation, and apply the framework to active-state-selective agonist design, where conditioning against the inactive receptor state more than doubles the active-state preference of generated molecules.