PLAS-5k: Dataset of Protein-Ligand Affinities from Molecular Dynamics for Machine Learning Applications

Abstract

Computational methods and recently modern machine learning methods have played a key role in structure-based drug design. Though several benchmarking datasets are available for machine learning applications in virtual screening, accurate prediction of binding affinity for a protein-ligand complex remains a major challenge. New datasets that allow for the development of models for predicting binding affinities better than the state-of-the-art scoring functions are important. For the first time, we have developed a dataset, PLAS-5k comprised of 5000 protein-ligand complexes chosen from PDB database. The dataset consists of binding affinities along with energy components like electrostatic, van der Waals, polar and non-polar solvation energy calculated from molecular dynamics simulations using MMPBSA (Molecular Mechanics Poisson-Boltzmann Surface Area) method. The calculated binding affinities outperformed docking scores and showed a good correlation with the available experimental values. The availability of energy components may enable optimization of desired components during machine learning-based drug design. Further, OnionNet model has been retrained on PLAS-5k dataset and is provided as a baseline for the prediction of binding affinities.

Fig. 1

Protocol for input preparation and simulations.

Fig. 2

Correlation plots between the experimental and calculated binding affinities for a subset with 2000 pdbids. The binding affinities are calculated (a) using Auto-dock Vina, and (b) using MM-PBSA.

Fig. 3

Prediction of binding affinity based on correlation with experimental data: FDA approved drugs for HIV-I protease targets (a) Experimental vs Docking, (b) Experimental vs MM-PBSA; For Tuberculosis targets - (c) Experimental vs Docking (d) Experimental vs MM-PBSA.

Fig. 4

Pearson correlation coefficient after training OnionNet on PLAS-5k database.