BRD4 (bromodomain-containing protein 4)

// Independently verified against a real PDB structure — not pipeline output

Independently checked against PDB 3MXF (BRD4 BD1 + JQ1, the acetyl-lysine mimetic that seeded the whole BET-inhibitor chemical series). The rank-1 cluster (persistence 1.0) matches 12 of 14 real JQ1 contact residues — 86% — strict top-1, including the signature Asn140 hydrogen-bond residue and the WPF-shelf residues (Trp81, Pro82, Phe83). Full-length BRD4 is 1362 aa with two bromodomains and large disordered linker regions; the ranker correctly ignored the disorder and landed on the folded BD1 pocket, the same pattern already seen on MDM2. Structural check: CA RMSD vs. 3MXF is 0.89 Å over the 127 residues 3MXF resolves — near-experimental accuracy on the folded BD1 domain itself (the rest of full-length BRD4 — the disordered linker, BD2, the ET domain — isn’t part of this comparison, and the 3D viewer’s confidence coloring shows why: those regions render as low-confidence, extended chain rather than folded structure, because that’s what AlphaFold actually predicts for them). One limitation worth naming: no self-matching interaction-fingerprint structure was found for this target — the bromodomain family is broad (40+ human paralogs sharing the fold), and none of the cached ligand-bound family structures aligned closely enough to BRD4 BD1 specifically at the coverage threshold this pipeline requires. The pocket recovery above is unaffected — it comes from structural analysis, not interaction fingerprints — but per-residue PLIP contact data is not available in this report.