Tc-43/GID4_7SLZ_20260620

GID4 (PDB 7SLZ) — GA-II Designed Ligand–Receptor Complexes

1,567 de novo, generative-AI–designed small-molecule ligands docked into the substrate-binding pocket of human GID4 (Glucose-Induced Degradation protein 4), each provided as a protein–ligand complex in PDB format.

GID4 is the substrate-recognition subunit of the human CTLH E3 ubiquitin-ligase complex and an emerging handle for targeted protein degradation / molecular glues. Receptor coordinates are based on PDB 7SLZ (crystal structure of GID4, Structural Genomics Consortium).

Dataset summary

Complex files	1,567 (*_complex.pdb)
Unique ligand SMILES	1,497
Receptor	GID4 substrate-binding β-barrel (chain A, ≈ residues 125–289)
Generator	Technetium GA-II pocket-conditioned generative platform
Generation date	2026-06-20

Each 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

The library was generated within a drug-like envelope and every pose clears a docking-energy bar. Physicochemical ranges below are computed over the 1,497 unique structures; docking energy is from the generation/docking pipeline.

Property	Range	Median
Docking energy	≤ −8.0 kcal/mol (down to ≈ −13.5)	—
Molecular weight	323 – 499 Da	420
cLogP	0.0 – 5.0	2.4
TPSA	34 – 130 Ų	86
Fsp3 (fraction sp³ C)	0.15 – 0.81	0.36
H-bond donors	1 – 6	2
H-bond acceptors	1 – 11	4
Rotatable bonds	2 – 10	5

The set sits in Lipinski/Veber-friendly space (MW ≤ 500, cLogP ≤ 5, rotatable bonds ≤ 10) with notable three-dimensional character (median Fsp3 0.36) — appropriate for GID4's shallow, partly peptidic substrate pocket.

File format

Filenames follow <MW>_7SLZ_II<id>_complex.pdb (<MW> = rounded molecular weight; <id> = compound id).

Record	Content
REMARK SMILES <smiles>	the bound ligand (2D structure)
REMARK SMILES IDX <pos> <serial> …	map of each SMILES heavy-atom position ↔ its HETATM serial (the 2D↔3D key)
HETATM (resname LIG)	ligand 3D pose
ATOM	GID4 receptor heavy atoms

Quick start (tabular index)

A browsable, machine-readable table (data/gid4_7slz.parquet) ships every ligand with its canonical SMILES, source complex filename, and RDKit physicochemical descriptors (mol_weight, clogp, tpsa, fsp3, h_bond_donors, h_bond_acceptors, rotatable_bonds, ring_count, num_heavy_atoms). It powers the Dataset Viewer above and loads in one line:

from datasets import load_dataset

ds = load_dataset("Tc-43/GID4_7SLZ_20260620", split="train")
print(ds[0])          # compound_id, smiles, complex_file, descriptors ...
df = ds.to_pandas()

The full 3D docked poses live in GID4_7SLZ_20260620.zip (one *_complex.pdb per row, joinable on complex_file / compound_id).

Usage (3D poses)

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, hetatm_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] -> HETATM serial

for f in glob.glob("GID4_7SLZ_20260620/*.pdb"):
    smi, idx = read_complex(f)
    # ... map any SMILES atom to its 3D coordinate via idx -> HETATM serial

A combined SMILES index (GID4_7SLZ_20260620.smi, SMILES<TAB>compound-id) is included for quick cheminformatics use.

Provenance & intended use

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

Citation

Generated by Technetium Therapeutics (GA-II platform). Receptor template: RCSB PDB 7SLZ.