When using the 3D Ligand Based Score in a generator, the generated molecules are optimized with respect to a crystallized reference molecule using two 3D scores: the 3D Shape Similarity and 3D Pharmacophore Similarity.

The model generates initial poses using our Flow Matching algorithm. These poses are then refined and optimized under shape, pharmacophore, and internal-energy constraints. 

3D Shape Score

The 3D Shape Similarity corresponds to the Shape Tanimoto Distance, which is a ratio of volume overlap between the reference compound and the generated molecule. It is calculated as an "Intersection over Union":

where O represents the volume of a molecule, A and B are two different molecules, and AB is the overlap between both molecules.

References

For more information, please check the following NeurIPS paper:

Template-Guided 3D Molecular Pose Generation via Flow Matching and Differentiable Optimization

Also available as a preprint: https://arxiv.org/abs/2506.06305