Datasets:

Tc-43
/

CRBN_Binders

license: cc-by-4.0
pretty_name: Cereblon (CRBN) Binders — Ligand–Receptor Complexes
task_categories:
  - graph-ml
  - other
size_categories:
  - n<1K
language:
  - en
tags:
  - chemistry
  - drug-discovery
  - molecular-docking
  - protein-ligand-complex
  - CRBN
  - cereblon
  - molecular-glue
  - targeted-protein-degradation
  - E3-ligase
  - biology
  - de-novo-design
  - generative-design
  - structure-based-drug-design
  - synthetic
  - synthetic-data
  - ai-generated
  - computational-chemistry
  - virtual-screening
  - cheminformatics
  - SMILES
  - E3-ubiquitin-ligase
  - CRL4-CRBN
  - thalidomide
  - lenalidomide
  - pomalidomide
  - IMiD
  - PROTAC
  - TPD
  - degrader
  - neosubstrate
  - ligase-handle
  - generative-ai
  - small-molecule
  - SDF
  - genetic-algorithm
  - GA-II
  - autodock-vina
  - TC-43
  - technetium

Cereblon (CRBN) Binders — Ligand–Receptor Complexes

Why this target matters. Cereblon is the most clinically validated E3 ligase handle in medicine — the direct target of thalidomide, lenalidomide and pomalidomide, and the recruiting engine behind most molecular glues and PROTACs now in the clinic.

221 computationally designed small-molecule ligands docked into Cereblon (CRBN), each provided as a single-file protein–ligand complex in PDB format (221 unique ligand structures).

CRBN is the substrate-receptor subunit of the CRL4^CRBN E3 ubiquitin ligase and the central recruiting handle for molecular glues and PROTACs in targeted protein degradation. Designs occupy the thalidomide-binding domain (the classic tri-tryptophan glutarimide pocket).

Receptor note: coordinates correspond to the CRBN thalidomide-binding domain (chain A res 20–123). TODO: add the source RCSB PDB accession for the receptor template used to generate these complexes.

Dataset summary


Complex files	221 (`*_cmpx.pdb`)
Unique ligand SMILES	221
Receptor	CRBN thalidomide-binding domain (chain A res 20–123)
Generator	Technetium GA-II pocket-conditioned generative platform
Pose scoring	AutoDock Vina

These are de novo, scaffold-constrained generative designs produced by the Technetium GA-II pocket-conditioned generative platform. Each design is docked into the target pocket and scored with AutoDock Vina; a REMARK CORE record preserves the scaffold/attachment context.

Each complex file is self-contained — receptor structure, the ligand's 3D docked pose, and a 2D↔3D atom map all travel inside the single PDB.

Property profile

Physicochemical ranges are computed with RDKit over the 221 unique ligand structures; docking energy is from the generation/docking pipeline.

Property	Range	Median
Docking energy (AutoDock Vina)	≤ -8.4 kcal/mol (down to -9.5)	—
Molecular weight	216.2 – 299.3 Da	282.3
cLogP	0.0 – 3.0	1.9
TPSA	38.3 – 102.4 Å²	57.1
Fsp3 (fraction sp³ C)	0.2 – 0.6	0.3
H-bond donors	0 – 4	1
H-bond acceptors	2 – 6	3
Rotatable bonds	1 – 3	1

File format

Each *_cmpx.pdb bundles the receptor and one docked ligand pose:

Record	Content
`REMARK VINA RESULT <energy> …`	AutoDock Vina docking score (kcal/mol)
`REMARK CORE <smiles>`	the scaffold / attachment context of the design
`REMARK SMILES <smiles>`	the docked ligand (2D structure)
`REMARK SMILES IDX <pos> <serial> …`	map of each SMILES heavy-atom position ↔ its ligand atom serial (the 2D↔3D key)
`ATOM … <chain>`	receptor heavy atoms
`ATOM … UNL` (after `MODEL 1`)	ligand 3D pose (residue name `UNL`)

Usage

import glob

def read_complex(path):
    smiles, idx = None, {}
    with open(path) as fh:
        for line in fh:
            if line.startswith("REMARK SMILES IDX"):
                toks = line.split()[3:]            # flat list of (smiles_pos, atom_serial)
                for i in range(0, len(toks), 2):
                    idx[int(toks[i])] = int(toks[i + 1])
            elif line.startswith("REMARK SMILES"):
                smiles = line.split(None, 2)[2].strip()
    return smiles, idx                              # idx[smiles_atom_position] -> ligand atom serial

for f in glob.glob("*_cmpx.pdb"):
    smi, idx = read_complex(f)
    # ligand atoms are the `ATOM ... UNL` records following `MODEL 1`

Provenance & intended use

These are computationally generated designs and docked poses — not experimentally validated binders. No claim of activity or selectivity is made.
Intended for machine-learning, cheminformatics, generative-model benchmarking, and docking-pose research on a well-defined target.

Citation

Generated by Technetium Therapeutics. Poses scored with AutoDock Vina.