TCR–pHLA interface analysisTCRshows

Upload a TCR–peptide–HLA ternary complex in PDB format. The chains are assigned for you, and the tool computes contacting residue pairs across the interface, hydrogen bonds, salt bridges, hydrophobic and π-system interactions, buried solvent-accessible surface area and an estimate of the binding free energy, then breaks the contribution down by CDR and by peptide position. Everything runs locally in your browser; the structure file is never uploaded.

1 · Load a structure

Predicted output from tFold-TCR, AlphaFold or TCRmodel2 works, as does an experimental structure from the RCSB PDB. Only the first model is read; hydrogens and waters are ignored.

Click to choose a .pdb / .ent / .cif file, or drop one here

A TCR–pHLA complex usually has five chains: TCRα, TCRβ, peptide, HLA heavy chain, β2m

Fetch a PDB ID from the RCSB

Paste PDB text instead

2 · Chain assignment

Inferred from chain length and conserved sequence motifs. Correct anything that is wrong — every analysis below builds on these role assignments.

Chain	Residues	Role	Basis	Sequence (N-terminal fragment)

Contact cutoff (heavy atoms, Å)

Temperature for the ΔG calculation (℃)

3 · Interface overview

Computed with TCR (α+β) on one side and pHLA (peptide + heavy chain + β2m) on the other.

Use this for relative comparison only — never as an absolute affinity. The binding free energy is estimated with PRODIGY’s contact-count / non-interface-surface (IC-NIS) regression model (Vangone & Bonvin, eLife 2015), at a fixed 5.5 Å residue-level contact cutoff. That model was trained on general protein–protein complexes, and TCR–pMHC is the textbook exception: a large interface with weak affinity. Measured natural TCRs mostly sit at Kd 1–100 µM, yet this model consistently returns about −12 kcal/mol, i.e. sub-nM, for the classic crystal structures (1AO7, 2BNR, 5MEN) — a systematic overestimate of roughly 3–4 kcal/mol, or about 1000-fold too tight in Kd. So it is only good for ranking within one system: affinity-matured variants of the same TCR, or different TCRs on the same pHLA. Cross-system comparison and absolute numbers are both unreliable, and the error grows further if the structure is predicted rather than crystallographic.

4 · Contribution by chain pair

Contacting residue pairs and atom pairs for each pair of chains — this is how you see the α / β docking balance, and whether the TCR is reading the peptide or the HLA framework.

Chain pair	Contacting residue pairs	Contacting atom pairs	Share of the TCR–pHLA interface

5 · Interaction detail

Identified geometrically. Predicted structures rarely contain hydrogens, so hydrogen bonds are called from the heavy-atom donor–acceptor distance, with no angular criterion.

6 · Contribution by CDR

CDR3 is located exactly, from the conserved Cys…[FW]G·G motif; CDR1 and CDR2 are inferred intervals anchored on IMGT positions, so check them against the sequences shown. Sequences follow the IMGT CDR3 definition (105–117) and exclude the conserved anchors Cys104 and Phe/Trp118 — to compare against junction notation such as CASSxxxxF, add one character at each end.

Segment	Sequence	Residue range	Peptide contacts	HLA contacts	Total

7 · Peptide position by position

Contacts made by each peptide residue with the TCR and with the HLA. Positions pointing up, towards the TCR, are usually the ones that determine specificity; those pointing down into the HLA pockets are the anchors.

Position	Residue	TCRα contacts	TCRβ contacts	TCR total	HLA contacts	Orientation

8 · Contact map

Rows are contacting residues on the TCR side, columns on the pHLA side; shade shows the minimum heavy-atom distance.

9 · 3D view

Cartoon for the overall fold, sticks for interface residues. Drag to rotate, scroll to zoom.