Development of the Knowledge-Based and Empirical Combined\nScoring Algorithm (KECSA) To Score Protein–Ligand Interactions

Abstract

We\ndescribe a novel knowledge-based protein–ligand scoring\nfunction that employs a new definition for the reference state, allowing\nus to relate a statistical potential to a Lennard-Jones (LJ) potential.\nIn this way, the LJ potential parameters were generated from protein–ligand\ncomplex structural data contained in the Protein Databank (PDB). Forty-nine\n(49) types of atomic pairwise interactions were derived using this\nmethod, which we call the knowledge-based and empirical combined scoring\nalgorithm (KECSA). Two validation benchmarks were introduced to test\nthe performance of KECSA. The first validation benchmark included\ntwo test sets that address the training set and enthalpy/entropy of\nKECSA. The second validation benchmark suite included two large-scale\nand five small-scale test sets, to compare the reproducibility of\nKECSA, with respect to two empirical score functions previously developed\nin our laboratory (LISA and LISA+), as well as to other well-known\nscoring methods. Validation results illustrate that KECSA shows improved\nperformance in all test sets when compared with other scoring methods,\nespecially in its ability to minimize the root mean square error (RMSE).\nLISA and LISA+ displayed similar performance using the correlation\ncoefficient and Kendall τ as the metric of quality for some\nof the small test sets. Further pathways for improvement are discussed\nfor which would allow KECSA to be more sensitive to subtle changes\nin ligand structure.