Surface descriptors for protein-ligand affinity prediction. Academic Article uri icon

start page

  • 25

end page

  • 33


  • Molecular descriptors calculated by the VolSurf program have been extensively used to model pharmacokinetic properties, e.g., passive permeability through the gastrointestinal tract or through the blood-brain barrier. These descriptors quantify steric, hydrophobic, and hydrogen bond interactions between model compounds and different environments. Since these interactions are the same as those involved in the ligand-receptor binding, VolSurf descriptors could potentially be relevant in modeling this process as well. We obtained a significant model (r(2) = 0.85, q(2) = 0.75) using VolSurf descriptors derived from the ligand, the protein, and the ligand-protein complex for a diverse set of 38 structures previously used in the VALIDATE (ref 23) training set. Furthermore, a statistically significant model (r(2) = 0.94, q(2) = 0.89) was obtained using the same type of descriptors for a homogeneous set of glycogen phosphorylase inhibitors (ref 25). Using the VolSurf computational framework, both ligand-receptor binding and the ligand's pharmacokinetic behavior can be modeled simultaneously during the preclinical aspects of drug discovery.

date/time value

  • 2003

Digital Object Identifier (DOI)

  • 10.1021/jm011051p

PubMed Identifier

  • 12502357


  • 46


  • 1


  • Ligands
  • Models, Biological
  • Models, Molecular
  • Pharmacokinetics
  • Protein Binding
  • Proteins
  • Quantitative Structure-Activity Relationship