enFoldX extracts 106 interface and confidence features from AF3 ensembles
quaidmorris · x · 2026-07-22
enFoldX details its ensemble-based feature pipeline
This reply explains how the model works:
- It folds an ensemble of 10 seeds × 5 samples.
- For each prediction, it extracts 106 structural, confidence, and biophysical features.
- The emphasis is on the TCR:pMHC interface and peptide–CDR3 contacts.
- Each feature is summarized with both the mean and standard deviation across the ensemble.
The result is a classifier that can exploit structure uncertainty rather than collapsing everything into one best guess.
Related event: enFoldX Turns AlphaFold3 Noise into TCR Recognition Signals(11 posts)→
More from Research
- DeepMind paper: LLMs can't sample randomly on their own, only convert given seeds — rohanpaul_ai · 2026-10-27
- DeepMind paper: LLMs can't sample randomly on their own — pass random numbers from code instead — rohanpaul_ai · 2026-10-27
- World Model Assigns Weight and Friction to Objects, Then Simulates with a Physics Engine — jonstephens85 · 2026-10-27
- Token Collapse: Study Finds Graph Soft Prompting Tokens in LLMs Fail to Encode Structure — Sapienza · 2026-10-27
- Mathathon 2.0 launches with Terence Tao's SAIR, $1.5M in compute for teams — _sathvikr · 2026-10-27
- The 'sweet lesson' is real: two kinds of ML problems need opposite strategies — mike64_t · 2026-10-27