Observed arrival · 2026-10-06
A peptide–HLA benchmark that shows its work
Stability Explorer compares models predicting peptide–HLA binding stability on grooves the models have not seen.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Immunology researchers studying peptide–HLA binding
- Worth noticing
- It excludes NetMHCstabpan because that model trained on the full dataset, and labels its structure view as a template rather than a predicted structure.
Field notes
Model scores use median per-allele Spearman correlation across 21 leave-one-groove-cluster-out folds, and the page explains why NetMHCstabpan is not a fair direct comparator for this dataset. Its structure view places selected residues on PDB 3GSN but explicitly says the peptide is not folded; the design tab ranks single substitutions instead of optimizing freely across sequence space.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue