Figure 3.

(a) Shapes, (b) chemical structure and (c) half-maximal inhibitory concentration against pure recombinant mouse Nat2 for the query molecule and the three most potent active molecules found by USR (from left to right: S₀^USR = 1.000, S₁^USR = 0.969, S₃^USR = 0.966 and S₁₇^USR = 0.950). The shapes of these conformers are aligned manually in order to appreciate their similarity (alignments are not a byproduct of USR operation, as this method does not require alignments to compare molecules). While being very similarly shaped, these USR hits have no common scaffold and thus belong to different chemical series. These are also remarkably different in structure with respect to the query molecule, as evidenced by the corresponding MACCS Tanimoto score (S₀^MACCS = 1.000, S₁^MACCS = 0.333, S₃^MACCS = 0.222 and S₁₇^MACCS = 0.377). Such scaffold hopping is a very valuable feature of a virtual screening method, as each chemical series can be considered as an alternative starting point for the lead optimization process.