enFoldX turns AlphaFold3 ensemble noise into a TCR–peptide–MHC predictor
quaidmorris · x · 2026-07-22
enFoldX uses AlphaFold3 ensemble noise to predict TCR:pMHC binding
The post explains that enFoldX does not rely on a single predicted structure. Instead, it extracts structural, confidence, and biophysical features from an AlphaFold3 ensemble and uses them to classify cognate vs non-cognate TCR–peptide–MHC pairs.
- SHAP points to iPAE as the dominant feature, with iPTM, pLDDT, peptide:MHC interface signals, and ensemble mean/SD also contributing.
- On the neoantigen task, where peptide pairs differ by just one residue, enFoldX leads 8 mutational-scan datasets with median AUC 0.71.
- It also beats sequence- and structure-based baselines on IMMREP25’s unseen-peptide challenge.
- The model appears to generalize across species: human-only training still reaches AUC 0.76 on mouse TCR:pMHCs.
More from Research
- OpenAI shares new reward-seeking research and a method to measure it — OpenAI · 2026-07-22
- Reddit points to OpenAI’s ChatGPT Ads page — EcstaticAsparagus509 · 2026-07-22
- Open-source runtime lets each repo define its own AI code reviewer — ibabufrik · 2026-07-22
- OpenAI and Apollo Research introduce Contrastive SDF to measure reward-seeking — OpenAI · 2026-07-22
- NVIDIA says to tune the harness before tuning the model with LangChain — NVIDIAAI · 2026-07-22
- The Thimble and the Waterfall: AI's Data Bottleneck and Feedback Loops — dyamins · 2026-07-22