Why a Good Docking Score Does Not Prove Binding

By BioDockify Computational Research Team · August 8, 2026 · Critical Analysis

The most important article for computational drug discovery credibility. Explains why docking scores are hypothesis-generating, not proof of binding. Covers false positives, scoring function limitations, and what constitutes proper validation.

Read the complete article with tables, code and references on BioDockify.

Key References

• Trott, O. & Olson, A.J. AutoDock Vina. J. Comput. Chem. 31, 455-461 (2010). DOI: 10.1002/jcc.21334
• McNutt, A.T. et al. GNINA. J. Cheminform. 13, 43 (2021). DOI: 10.1186/s13321-021-00522-2
• Li, H. et al. RF-Score v3. J. Cheminform. 12, 20 (2020). DOI: 10.1186/s13321-020-00425-4
• Jain, A.N. Scoring noncovalent protein-ligand interactions. Curr. Opin. Drug Discov. Devel. 7, 396-403 (2004).
• Kitchen, D.B. et al. Docking and scoring in virtual screening. Drug Discov. Today 9, 993-1000 (2004).

Scope & Limitations

This article is about limitations — that IS the point Scoring functions are improving but still imperfect Experimental validation is always required