Are Deep Learning Structural Models Sufficiently Accurate for Free- Energy Calculations? Application of FEP plus to AlphaFold2-Predicted Structures

The availability of AlphaFold2 has led to great excitement in the scientific community -particularly among drug hunters -due to the ability of the algorithm to predict protein structures with high accuracy. However, beyond globally accurate protein structure prediction, it remains to be determined whether ligand binding sites are predicted with sufficient accuracy in these structures to be useful in supporting computationally driven drug discovery programs. We explored this question by performing free energy perturbation (FEP) calculations on a set of well-studied protein-ligand complexes, where AlphaFold2 predictions were performed by removing all templates with > 30% identity to the target protein from the training set. We observed that in most cases, the delta delta G values for ligand transformations calculated with FEP, using these prospective AlphaFold2 structures, were comparable in accuracy to the corresponding calculations previously carried out using crystal structures. We conclude that under the right circumstances, AlphaFold2-modeled structures are accurate enough to be used by physics-based methods such as FEP in typical lead optimization stages of a drug discovery program.

Automated summary

The availability of AlphaFold2 has generated excitement in the scientific community, especially among drug researchers, due to its high accuracy in predicting protein structures. This study explores whether the predicted structures by AlphaFold2 are accurate enough to be useful in computationally driven drug discovery programs, specifically in predicting ligand binding sites. The findings suggest that, under certain circumstances, AlphaFold2-modeled structures can be used with physics-based methods, such as free energy perturbation, in the lead optimization stages of drug discovery programs.