Computer-based substrate specificity prediction for cytochrome P450

Cytochrome P450 is an important class of enzymes metabolizing numerous drugs. The composition and activity of these enzymes determine distribution of drug in the body, their pharmacological and toxic effects. Thus, prediction of the fate of compounds in the body is required at early stages of the development of new drugs. Different isoforms of cytochrome P450 can oxidize a wide range of chemical compounds and their substrate specificity does not correlate with their taxonomical classification. In this review we consider the main methods of cytochrome P450 substrate specificity prediction. These methods are subdivided by primary information used in the analysis: amino acid sequence based prediction, ligand-based (pharmacophore and QSAR models, expert systems) and structure-based (molecular docking, affinity prediction, interaction energy estimation) methods. The common problems complicating cytochrome P450 substrate prediction and advantages and disadvantages of these methods are discussed.