Add a benchmark computing bond dissociation energies

[gelzinyte]

Summary

Performance in predicting C-H bond dissociation energies (BDEs) for 60 CYP3A4 drug-like substrates (CHO elements only), comprising 1117 sp3 C-H bonds across all molecules. The BDEs are computed as E_radical + E_isolated_H - E_molecule. The first metric computes the BDE error directly. The second metric compares in models' ability to predict relative BDEs. This is evaluated by comparing predicted and reference BDE ranks (lowest-highest order assignments) for each molecule and averaged across all molecules in the dataset. Direct BDE and BDE rank prediction is computed on DFT-optimised geometries and on MLFF-optimised geometries.

Interactive features

Table with (Direct BDE, BDE rank) x (DFT geometry, MLFF geometry) columns, clicking on datapoints bringing up the geometry of the radical.

Category

molecular

Data availability

The procedure is described in

Gelzinyte, E. et al. Transferable Machine Learning Interatomic Potential for Bond Dissociation Energy Estimation. J. Chem. Theory Comput. 20, 164-177 (2024).
DOI: 10.1021/acs.jctc.3c00710

which also contains the SMILES strings for the tested compounds.

Computational cost

Medium: the DFT geometry tests are fast, but the MLFF geometry optimisation tests may take several minutes per model on CPU.

Additional context

No response