Expanded Benchmark

T cell receptors (TCRs) play a crucial role in the adaptive immune system. They are responsible for recognising antigenic peptides presented at the surface of infected cells by major histocompatibility complex (MHC) molecules, and triggering downstream immune responses to fight off disease.

This repository contains structural data of TCRs and pMHC complexes crystallised in bound and unbound form, serving as a set of benchmark test cases for computational docking tools. This dataset contains the 30 TCR-pMHC docking cases present in the original TCR benchmark, as well as 14 new cases identified in the STCRDab and Protein Databank databases. This data was curated for work described in Peacock and Chain (2021).

Structures in the raw folder contain structural data as stored in the Protein Databank. Original names for these PDB structures can be found in the table below.

Structures in the imgt folder have been preprocessed. Solvent and heteroatoms have been removed, as have disordered atoms. Residues with non-standard names in PDB format have been updated to their standard equivalents. Raw data that contained multiple co-crystalised protein structures has been reduced such that only one appears in each PDB file. Structures with missing atoms or residues at the binding interface were repaired using Modeller using a protocol described in the Modeller-Repair repository. TCRs in both the unbound and bound files have been re-numbered according to the IMGT numbering scheme. TCR chains have been relabelled with the IDs D and E, MHC chains with IDs A and B, and peptide chains with ID C. The position and orientation of each structure has been randomised.

Each docking case is named after the PDB code of the raw bound TCR-pMHC structure. Bound TCR-pMHC structures contain the suffix _b.pdb, unbound TCRs contain the suffix _r_u.pdb (where r stands for "receptor", and u stands for "unbound") and unbound pMHCs contain the suffix _l_u.pdb (where l stands for "ligand", and u stands for "unbound"). Corresponding codes for the unbound TCR and unbound pMHC structures can be found in the table below.

Bound Complex	Unbound TCR	Unbound pMHC	MHC Class	IRMSD	Fnon-nat	Difficulty
1AO7 *	3QH3	1DUZ	I	1.25	0.33	rigid
1MI5 *	1KGC	1M05	I	1.25	0.48	medium
1MWA *	1TCR	1LEK	I	1.14	0.3	rigid
1OGA *	2VLM	2VLL	I	1.36	0.43	medium
2BNR *	2BNU	1S9W	I	0.72	0.23	rigid
2CKB * ‡	1TCR	1LEG	I	1.17	0.45	medium
2IAM *	2IAL	1KLG	II	0.87	0.24	rigid
2IAN *	2IAL	1KLU	II	0.82	0.3	rigid
2NX5 * †	2NW2	1ZSD	I	1.16	0.38	rigid
2OI9 *	1TCR	3ERY	I	1.1	0.41	medium
2PXY *	2Z35	1K2D	II	1.18	0.55	medium
2PYE *	2PYF	1S9W	I	0.88	0.3	rigid
3DXA * †	3DX9	3DX8	I	1.48	0.39	rigid
3H9S *	3QH3	3H7B	I	1.31	0.42	medium
3KPR *	1KGC	3KPQ	I	1.37	0.55	medium
3KPS *	1KGC	3KPP	I	1.31	0.48	medium
3PWP *	3QH3	3PWL	I	1.24	0.36	rigid
3QDG * † ‡	3QEU	1JF1	I	0.91	0.31	rigid
3QDJ * ‡	3QEU	2GUO	I	0.94	0.28	rigid
3SJV * ‡	3SKN	1M05	I	0.96	0.41	medium
3UTT *	3UTP	3UTQ	I	0.75	0.4	rigid
3VXR	3VXQ	3VXN	I	0.82	0.38	rigid
3VXS * †	3VXQ	3VXP	I	0.89	0.35	rigid
3W0W * † ‡	3VXT	3VXO	I	0.94	0.42	medium
4JFD * †	4JFH	4JFP	I	1.51	0.51	medium
4JFF	4JFH	1JF1	I	1.54	0.52	medium
5C07	3UTP	5C0E	I	0.57	0.15	rigid
5C08	3UTP	5C0F	I	0.65	0.43	medium
5C09	3UTP	5C0G	I	0.59	0.24	rigid
5C0A	3UTP	5N1Y	I	0.5	0.3	rigid
5C0B	3UTP	5C0I	I	0.59	0.25	rigid
5C0C	3UTP	5C0J	I	0.64	0.35	rigid
5HHM	2VLM	5HHN	I	1.42	0.51	medium
5HYJ	3UTP	5C0D	I	0.55	0.34	rigid
5IVX	5IW1	3ECB	I	1.29	0.38	rigid
5NME	5NMD	2V2W	I	1.07	0.34	rigid
5NMF * †	5NMD	5NMH	I	1.05	0.38	rigid
5NMG * † ‡	5NMD	5NMK	I	1.07	0.42	medium
6AMU	3QEU	6AMT	I	1.16	0.41	medium
6AVF * †	6AT6	6AT5	I	1.95	0.72	medium
6CQL * †	6CPH	6CPN	II	0.78	0.23	rigid
6CQQ * †	6CPH	6CPO	II	0.83	0.21	rigid
6CQR * †	6CPH	6CQJ	II	0.85	0.26	rigid
6EQB	4JFH	2GUO	I	1.62	0.55	medium

^* TCR docking cases that feature in the TCR3d database
^† Cases that differ in I-RSMD score to those in the TCR3d database
^‡ Cases that differ in docking difficulty category in the TCR3d database

Notes:

• A structure for 2NX5_r_u with improved modelling of the missing atoms and residues in one of its CDR3 loops has been uploaded to the imgt folder, since the publication of Peacock and Chain (2021). Minor differences in the IRMSD and F_non-nat values have been updated accordingly in the table above.