Each position is a real lead-optimization lever:
- R2 models a hydrophobic sub-pocket — CH₃, Cl, F and CF₃ pack in well; bulky polar groups (COOH) pay a burial penalty.
- R3 / R4 model complementary H-bond partners on the receptor: an acceptor residue needs a ligand donor (OH, NH₂, acids), a donor residue needs a ligand acceptor (F, OH, OCH₃, acids). Carboxylic acid and tetrazole are classic bioisosteres — similar acidity and charge, near-identical binding, different metabolism.
- Metabolic stability flags groups that CYP450 enzymes attack readily: anilines (–NH₂), O-methyl ethers (–OCH₃, O-demethylation) and benzylic methyls oxidize fast; –F and –CF₃ are classic metabolic blockers.
- Lipinski's Rule of Five (oral bioavailability heuristic): MW ≤ 500, cLogP ≤ 5, H-bond donors ≤ 5, H-bond acceptors ≤ 10.
Potency is a simplified score (H-bond matches + hydrophobic fit − steric/polar clash), not a real docking calculation.