weiventee/Repository_for_Orthosteric_and_Allosteric_Design

ROAD — Repository for Orthosteric and Allosteric Design

ROAD is a non-redundant set of 27,164 ligand-binding sites (6,207 allosteric, 20,957 orthosteric) identified from 128,141 PDB structures, with ligand, site-geometry, residue-composition and protein–ligand contact features for each site.

• Paper: TODO
• Browse it: https://huggingface.co/spaces/weiventee/Repository_for_Orthosteric_and_Allosteric_Design
• Zenodo repository (includes complete PDB files): https://doi.org/10.5281/zenodo.21901988
• Code: https://github.com/WeiVen-Tee/ROAD-database

Files

Config / file	Rows	Description
site_ligand_features.csv	27,164	One row per site: annotation plus ligand, site and contact features
nonredundant_sites_annotated_enriched.csv	27,164	Site definitions and protein annotation
clusters_foldseek_all.csv	27,164	Foldseek structural cluster of each site's protein
clusters_foldseek_allosteric.csv	6,207	The same, clustered over structures with an allosteric site
hierarchy.csv	27,164	Sites grouped as EC subclass > protein > location

Every column is described in the matching _fields.txt file. Cluster IDs (cluster_struc) are assigned independently in the two cluster files and cannot be compared across them.

Each site is labelled by min_dist: the shortest heavy-atom distance between any residue lining the site and any residue annotated in UniProt as Active site or DNA-binding. Using an operational definition of allosteric sites: allosteric if min_dist > 5 Å, else orthosteric.

Residues are written Chain.Resnum (insertion code appended, e.g. H.60A) and ligands Chain.Resnum:CCD, using the author chain and residue numbering of the RCSB {PDB_ID}-assembly1.cif file, where symmetry copies of a chain are named e.g. B-2.

import pandas as pd
df = pd.read_csv("hf://datasets/weiventee/Repository_for_Orthosteric_and_Allosteric_Design/site_ligand_features.csv", dtype={"source_pdb": str})

Read source_pdb as a string: IDs such as 1E20 otherwise parse as numbers.

The structure files (PDB.tar.gz, PDB_allosteric.tar.gz) are on Zenodo only.

Checksums (MD5, identical to Zenodo version 1.0.0)

62b4001086cba0b56c55da3f43f02a52  site_ligand_features.csv
405bf11781a6531a2b8c900d03c69956  site_ligand_features_fields.txt
744c14513c51edf024687d6d7c7cf87e  nonredundant_sites_annotated_enriched.csv
513866a3af30dd44c757e585e7255c22  nonredundant_sites_annotated_enriched_fields.txt
077e26862e85772035693c74076bbc51  clusters_foldseek_all.csv
cae1d777bf903ac5621d435c7954230d  clusters_foldseek_allosteric.csv
78b112fa7277af2012d9a0834e71fa78  clusters_foldseek_fields.txt
4529b5bf5b072dcb466f198e624c62fa  hierarchy.csv
585ccc970626636bd9f94d3ceababa4f  hierarchy_fields.txt

Licence and attribution

Released under the GNU General Public License v3.0. Protein names, EC numbers and functional-site annotations are derived from UniProt (CC BY 4.0); structures and ligand definitions from the wwPDB (CC0).

Citation

TODO