Pharmacophore — why these 3 drugs and not others

Common chemistry of the 3 hits

Drug	Scaffold	MW (Da)	LogP	HBD	HBA	Aromatic rings
PHENYTOIN	5,5-diphenylhydantoin	252	1.77	2	2	2
CLONAZEPAM	7-NO₂, 2′-Cl benzodiazepine	316	3.04	1	4	2
OXAZEPAM	7-Cl, 3-OH benzodiazepine	286	2.32	2	3	2

Common features:

Heterocyclic 5- or 7-membered ring with a free N-H imide (not N-methylated)
Two phenyl rings (one fused, one pendant)
Specific aromatic ring substituents (5,5-diphenyl for hydantoin; 7-NO₂/2′-Cl OR 7-Cl/3-OH for benzodiazepine)
MW 250–320 Da
LogP 1.7–3.0

The 25-compound stress test — 0/25 new robust hits

To test whether the pharmacophore is real, we docked 25 chemically-related compounds at pocket #1. Negative result: 0/25 new robust hits.

Compound	Family	Δ	Verdict
NITRAZEPAM	7-NO₂ benzo (no 2′-Cl)	−0.08	no signal
FLUNITRAZEPAM	7-NO₂ + 2′-F	+0.11	REVERSED
BROMAZEPAM	7-Br benzo	−0.01	no signal
CHLORDIAZEPOXIDE	7-Cl different scaffold	+0.24	REVERSED
LORMETAZEPAM	3-OH + N1-methyl	−0.23	collapsed
TEMAZEPAM	oxazepam + N1-methyl	−0.38	borderline only
11 other benzos	various	−0.10 to +0.65	weak / reversed
MEPHENYTOIN (N3-Me phenytoin)	hydantoin	+0.21	REVERSED
FOSPHENYTOIN	phenytoin prodrug	−0.25	collapsed
8 BARBITURATES	barbiturate	+0.03 to +0.45	all fail
4 Z-DRUGS (ZOLPIDEM…)	Z-drug	−0.05 to +0.59	no signal

This negative result is scientifically positive: it proves the 3 hits bind via a chemically specific pattern, not a generic hydrophobic interaction.

What we can deduce about binding

The pocket #1 cavity has 3 lysines (K126, K127, K249) — basic H-bond donors; 2 glutamates (E168, E170) — acidic H-bond acceptors; 1 tyrosine (Y251) — aromatic π-stacking; 2 hydrophobic residues (A229, I250). The pharmacophore matches:

Free N-H imide = H-bond donor to E168/E170 (a single methyl group on this NH abolishes binding — confirmed 3× with mephenytoin, temazepam, lormetazepam)
Phenyl rings = π-stacking with Y251
MW range = exact pocket volume match