ABSTRACT
In the early 1990s, several successful structure-based drug designs (SBDDs) were reported,1-4 and this approach has become common recently, implying that rational drug design based on the physical theory can compete with the conventional trialand-error drug discovery. For example, interaction analysis is often used to identify
8.1 Introduction .................................................................................................. 171 8.2 Modeling of Protein-Ligand Complexes ...................................................... 173
8.2.1 Addition of Missing Hydrogen Atoms and Checkpoints of Experimental Structure Data ........................................................ 173
8.2.2 Structural Refinements by Geometry Optimization ......................... 175 8.2.3 Optimized Structure and Distortion Due to the Packing Effect
in Crystal .......................................................................................... 176 8.3 Protein-Ligand Interactions ......................................................................... 176
8.3.1 FMO Calculations and Pair Interaction Analysis ............................. 176 8.3.2 Binding Energy ................................................................................. 178 8.3.3 Interaction Energy Decomposition ................................................... 181 8.3.4 Importance of the Polarization and the van der Waals
Interactions ....................................................................................... 186 8.4 Solvent Effects on Protein-Ligand Binding ................................................. 187
8.4.1 Continuum Solvent Models .............................................................. 187 8.4.2 Desolvation Penalty for the Association of the Protein
and Ligand ........................................................................................ 188 8.5 Conclusion .................................................................................................... 189 References .............................................................................................................. 190
which amino acids are significant for ligand recognition and to obtain ideas for ligand modification. Docking simulations including the virtual in silico screening are increasingly used to find novel seed or lead compounds, because the number of real compounds suitable for the high-throughput screening is growing, and the screening costs rise whereas the hit rates decrease. Protein modeling is essential if experimental protein structures are unavailable. Affinity prediction is one of most important but the most difficult research areas. There are many methods for affinity prediction, for example, quantitative structure-activity relationship (QSAR) or more rigorous simulation-based methods such as the free energy perturbation (FEP)5 and thermodynamic integration (TI)6 methods.
