A recently just-released publication from Ngyuen et al. ing JCIM pointed out that while AutoDock Vina is faster, AutoDock 4 tends to have better correlation with experimental binding affinity.1
[This post has been edited to provide more information about the cited paper, as well as providing additional citations.]
Ngyuyen et al. selected 800 protein-ligand complexes for 47 protein targets that had both experimental PDB structures complexed with a ligand, as well as their associated binding affinity values.
They performed re-docking of the 800 ligands to the 47 proteins, using both AutoDock4 and AutoDock Vina, with different options that controlled how far the docking engines searched, which they dubbed “short“, “medium“, and “long“. These protocols corresponded to 250k evals, 2.5m evals and 25m evals for AutoDock 4; and exhaustiveness values of 8, 56, and 400 for AutoDock Vina.
No surprises, they reported that AutoDock Vina was faster than AutoDock 4. But AutoDock 4 showed better performance, in terms of Pearson correlation coefficient of the predicted binding affinity with the experimental value, as well as better precision, & success rate, for 21 of the 47 targets. They found Vina was better than AD4 for 10 targets; while both AutoDock 4 and AutoDock Vina had poor affinity correlation for 16 targets. The RMSE values in kcal/mol they found for AutoDock 4 were very similar to the originally reported standard error of the scoring functions (although they did not say whether they used the bound, unbound, or extended assumption to model the ligand before binding).
Nguyen et al. concluded that the best docking protocol for their set of ligands with AutoDock 4 was long, while the short option was best for Vina. The correspondance between predicted binding free energy and the experimental value was given in their Table 1:
They also cited examples of AutoDock 4 identifying leads that had been published in JACS and Nat. Commun. in the last 2 years.2-4
Details of the systems in each of the three categories were given in Tables 2, 3, and 4:
Nguyen et al. also found that their “noncovalent bonding analyses indicate that the number of HBs between a ligand and a receptor obtained via the AD4 approach correlates more with experiments that [sic] found with Vina”.
Despite more recent advances, there is still a long way to go in achieving rapid scoring functions that can be used both in docking and predicting protein-ligand binding affinity.
References
- Nguyen, N.T., Nguyen, T.H., Pham, T.N.H., Huy, N.T., Bay, M.V., Pham, M.Q., Nam, P.C., Vu, V.V., and, Ngo, S.T. (2019). “Autodock [sic] Vina Adopts More Accurate Binding Pose but Autodock4 [sic] Forms Better Binding Affinity”, J Chem Inf Model, 60 (1), 204-211. doi: 10.1021/acs.jcim.9b00778. PubMed PMID: 31887035.
- Salveson, P. J., Haerianardakani, S., Thuy-Boun, A., Yoo, S., Kreutzer, A. G., Demeler, B. and Nowick, J. S. (2018). “Repurposing Triphenyl-methane Dyes to Bind to Trimers Derived from Aβ.” J. Am. Chem. Soc., 140 (37), 11745−11754. doi.org/10.1021/jacs.8b06568
- Corre, S., Tardif, N., Mouchet, N., Leclair, H. M., Boussemart, L., Gautron, A., Bachelot, L., Perrot, A., Soshilov, A., Rogiers, A., Rambow, F., Dumontet, E., Tarte, K., Bessede, A., Guillemin, G. J., Marine, J.-C., Denison, M. S., Gilot, D. and Galibert, M.-D. (2018). “Sustained Activation of the Aryl Hydrocarbon Receptor Transcription Factor Promotes Resistance to BRAF-Inhibitors in Melanoma.” Nat. Commun., 9 (1), 4775. doi.org/10.1038/s41467-018-06951-2
- Almaqwashi, A. A., Zhou, W., Naufer, M. N., Riddell, I. A., Yilmaz, Ö. H., Lippard, S. J., and Williams, M. C. (2019). “DNA Intercalation Facilitates Efficient DNA-Targeted Covalent Binding of Phenanthriplatin.” J. Am. Chem. Soc., 141 (4), 1537−1545. doi.org/10.1021/jacs.8b10252
