Pharmacophore feature

Another family of fingerprints available are the MOE pharmacophore fingerprints accessible through software from the Chemical Computing group [51]. In this system, the atoms are generalized into a smaller vocabulary of pharmacophore features, after which the fingerprint is constructed based on connected paths. 

A third and slightly older method available is the UNITY package from Tripos Inc. This also relies on the user to identify pharmacophore features and spatial arrangement. When multiple compounds and biological activity is known, this can be used to focus on a limited number of features or to exclude specific volume from the molecule. The compounds in the database are then compared to the query... 

Pharmacophoric features are red for positively ionizable, green for hydrogen bond acceptors, light blue for hydrophobic and orange for ring aromatic. Shape restraints are shown in light... 

Figure 16.7 (I) A graphical representation ofthe most important 3D pharmacophoric GRIND features forthe astemizole inhibitor as a reference compound [54]. (II) Similar pharmacophoric features found forthree potent P-gp inhibitors [55] as an example, these features are reported for compound 3a. The colored areas around the molecule are the GRID MIFs calculated with the DRY (yellow) and N1 (blue) probes.

Why should Gox2 activity be at all related to the presence or absence of certain pharmacophore feature pairs in a molecule Certainly, it can be argued - using the typical interpretation given to QSAR models based on autocorrelogram-type descriptors [63] - that the selected pharmacophore feature pairs stand for pairs of functional groups that are involved in direct interactions with the site, and these functional groups must be found at a fixed relative distance. However, this model... 

At identical variable number, the overlay-based model outperforms its overlay-independent equivalent training RMS = 0.712 (pICso units), training = 0.712, validation RMS = 0.698, training = 0.724. ComPharm field descriptors therefore appear to have a better predictive power than pairwise pharmacophore feature counts alone. 

---