A training set of 35 compounds whose calcite growth inhibition potencies we
re measured was used to construct a 4D-QSPR model. A site-specific binding
pattern of atom types in space, that is a pharmacophore, consisting of six
interaction sites between the inhibitors and the surface to which they bind
was identified and represented by the 4D-QSPR model. Three of these pharma
cophore sites dominate the 4D-QSPR model. One pharmacophore site indicates
that its occupancy by any inhibitor atom decreases inhibition potency, sugg
esting this region of space is occupied by the binding surface. A second ph
armacophore site involves an oxygen of a PO3H2 group, which is common to al
l compounds of the training set, hydrogen bonding to the surface. The third
major pharmacophore site identifies a nonpolar region of space as being fa
vorable to inhibition potency. A virtual library of 20 analogues to the tra
ining set was evaluated by using the 4D-QSPR model as a virtual high throug
hput screen, VHTS. Seven of the compounds in the virtual library are predic
ted to be better calcite growth inhibitors than the most potent inhibitor o
f the training set.