Molecular docking plays a significant role in early-stage drug discovery, from structure-based virtual screening (VS) to hit-to-lead optimization, and its capability and predictive power is critically dependent on the protein–ligand scoring function. In this review, we give a broad overview of recent scoring function development, as well as the docking-based applications in drug discovery. We outline the strategies and resources available for structure-based VS and discuss the assessment and development of classical and machine learning protein–ligand scoring functions. In particular, we highlight the recent progress of machine learning scoring function ranging from descriptor-based models to deep learning approaches. We also discuss the general workflow and docking protocols of structure-based VS, such as structure preparation, binding site detection, docking strategies, and post-docking filter/re-scoring, as well as a case study on the large-scale docking-based VS test on the LIT-PCBA data set.

The pseudobond approach offers a smooth connection at the quantum mechanical/molecular mechanical interface which passes through covalent bonds. It replaces the boundary atom of the environment part with a seven-valence-electron atom to form a pseudobond with the boundary atom of the active part [Y. Zhang, T. S. Lee, and W. Yang, J. Chem. Phys. 110, 46 (1999)]. In its original formulation, the seven-valence-electron boundary atom has the basis set of fluorine and a parametrized effective core potential. Up to now, only the ${\mathrm{C}}_{\mathrm{ps}}{\mathrm{(sp}}^3\mathrm{)–}\mathrm{C}{\mathrm{(sp}}^3)$ pseudobond has been successfully developed; thus in the case of proteins, it can only be used to cut the protein side chains. Here we employ a different formulation to construct this seven-valence-electron boundary atom, which has its own basis set as well as the effective core potential. We have not only further improved ${\mathrm{C}}_{\mathrm{ps}}{\mathrm{(sp}}^3\mathrm{)–}\mathrm{C}{\mathrm{(sp}}^3)$ pseudobond, but also developed ${\mathrm{C}}_{\mathrm{ps}}{\mathrm{(sp}}^3\mathrm{)–}\mathrm{C}{\mathrm{(sp}}^2,\mathrm{carbonyl})$ and ${\mathrm{C}}_{\mathrm{ps}}{\mathrm{(sp}}^3\mathrm{)–}\mathrm{N}{\mathrm{(sp}}^3)$ pseudobonds for the cutting of protein backbones and nucleic acid bases. The basis set and effective core potential for the seven-valence-electron boundary atom are independent of the molecular mechanical force field. Although the parametrization is performed with density functional calculations using hybrid B3LYP exchange-correlation functional, it is found that the same set of parameters is also applicable to Hartree-Fock and MP2 methods, as well as DFT calculations with other exchange-correlation functionals. Tests on a series of molecules yield very good structural, electronic, and energetic results in comparison with the corresponding full ab initio quantum mechanical calculations.