Abraham D J, Kellogg G E
Department of Medicinal Chemistry, Medical College of Virginia, Virginia Commonwealth University, Richmond 23298-0540.
J Comput Aided Mol Des. 1994 Feb;8(1):41-9. doi: 10.1007/BF00124348.
Physical organic structural properties of small molecules and macromolecules such as bond count, branching and proximity between multiple polar fragments contribute significantly to measured hydrophobicity (log P). These structural properties are encoded in the Rekker and Leo methods of calculating log P as structural-dependent factors. Regardless of the size of the atom primitive set, methods predicting log P with only atom primitives can miss subtle structural detail within series of related compounds. The HINT (Hydropathic INTeractions) model for inter- and intramolecular noncovalent interactions calculates atom-based hydrophobic constants, but uses all Leo-type factors in the calculation rather than a large set of atom primitives. Two types of applications of HINT are discussed: evaluation of the binding of an inhibitor (A74704) to HIV-1 protease, where it is shown that modeling of the protonation state (i.e., Asp25, Asp125) in the protein can strongly influence perceived substrate binding; and the use of HINT to calculate a third (hydropathic) field for CoMFA can yield a statistically enhanced and predictive model for molecular design.
