Pharmacophore protein site-derived

In general, pharmacophore models can be derived by characterizing the binding site on the receptor protein, by comparing the structures of known ligands, or by computational docking analysis, which is essentially a combination of the first two approaches. 

A computational elucidation of putative binding sites using the PASS algorithm revealed an internal site as most interesting for small organic molecules to interact with [49], Subsequent more detailed analysis resulted in the derivation of a protein-based pharmacophore (Figure 8.8). 

Finally, dynamic structure-based pharmacophore models can be derived through a method first described by Carlson et al that uses multiple conformations of the target protein, which are obtained either by molecular dynamics simulation or by the use of multiple experimentally determined conformations. The binding sites of the respective snapshots are flooded with small molecular probes (e.g., methanol for hydrogen-bond interactions and benzene for aromatic hydrophobic interactions) and while the protein structure is held rigid the probe molecules are subjected to a low-temperature Monte Carlo minimization where they undergo multiple, simultaneous gas-phase... 

---