ABSTRACT
Comparative binding energy (COMBINE) analysis is a computational method for deducing quantitative structure-activity relationships using structural data from ligandmacromolecule complexes (1). It can be applied to the formation of macromolecule-small molecule complexes and macromolecule-macromolecule complexes; in this article, these complexes will be referred to generically as macromolecule-ligand complexes. The “COMBINE” acronym refers to two aspects of the technique (2): (i) macromoleculeligand structural data are combined with experimental binding data and (ii) empirical molecular mechanics energy calculations are combined with Partial Least-Squares Projection to Latent Structures (PLS) chemometric analysis. COMBINE analysis systematically explores the relationships between experimental binding affinities for a set of ligands and selected interaction energies with the macromolecule. COMBINE analysis is formally similar to CoMFA (comparative molecular field analysis) (3) in as much as both methods deal with data matrices containing a large number of energy descriptors that are subjected to chemometric analysis. On the other hand, the energy descriptors differ: in CoMFA they are interaction fields calculated for the ligand alone, whereas in COMBINE analysis they represent residue-based ligand-receptor interactions. Compared to classical molecular mechanics calculations of binding energies, the advantages of subjecting ligand-macromolecule interaction energies to statistical analysis are that the
noise due to inaccuracies in the potential energy functions and molecular models can be reduced and that mechanistically important interactions can be identified. Compared to classical Quantitative Structure-Activity Relationships (QSAR) analysis, COMBINE is expected to be more predictive as it incorporates more physically relevant information about the energetics of ligand-receptor interactions (1).
