Networks

Compare peptides by embedding similarity. ESM-2, pdb-predict, and ae-pdb GAE index from each result PDB.

Mock peptides

When the same allele, peptide, and docking method appear in multiple runs, only the most recent result is used in the network graph, peptide table, and HCA (newest artifact token).

Focal peptide is not indexed for: pdb-predict, ae-pdb GAE. Similarity uses only indexed methods — run the index actions below.

Embedding methods

pdb-predict0/12 indexed

Structure classifier logits from pMHC PDB. GitHub

ESM-212/12 indexed

Mean residue embedding from peptide chain C in final PDB (py-docktope API)

ae-pdb GAE0/12 indexed

Mean-pooled VGAE latent from pMHC interface graph. GitHub

Average selected methods

Similarity is the mean of per-method cosine scores.

Focal peptide

Type sequence

CVNGVCWTV

A0201

11 neighbors shown

Min. similarity0%

Highlighted peptides

Space- or comma-separated sequences. Shown in the network and tables. Presets fill highlight fields (Network 1: two groups; Network 2: five pairs).

Orange

Red

Yellow

Purple

Green

Blue

Cyan

Pink

Hide non-highlighted nodes

Add at least one highlighted sequence above.

HCA linkage

Add highlighted sequences first.

Similarity network

Radial distance from the focal peptide reflects averaged embedding distance (1 − similarity).

Neighbors

Sorted by similarity (highest first)

Peptide	Allele	Method	Similarity	pdb-predict	ESM-2	ae-pdb GAE
SSYRRPVGI	A1101	Gnina	22.0%	—	22.0%	—
FLKEKGGL	B0801	Gnina	18.4%	—	18.4%	—
NLVPMVATV	A0201	Gnina	14.7%	—	14.7%	—
RLRAEAQVK	A0301	Gnina	13.6%	—	13.6%	—
RPPIFIRRI	B0702	Gnina	7.6%	—	7.6%	—
GILGFVFTL	A0201	Gnina	0.0%	—	0.0%	—
RPPIFIRRL	B0702	Gnina	0.0%	—	0.0%	—
VMAPRTLVL	A0101	Gnina	0.0%	—	0.0%	—
VMAPRTLFL	A0101	Gnina	0.0%	—	0.0%	—
RLRAEAQVI	A0301	Gnina	0.0%	—	0.0%	—
SSYRRPVGF	A1101	Gnina	0.0%	—	0.0%	—

Peptide index

Peptide	Allele	pdb-predict	ESM-2	ae-pdb GAE