Discrete diffusion model DTCR designs T-cell receptors for unseen immune targets
bravo_abad · x · 2026-08-27
Haoyan Wang and coauthors introduce DTCR, a discrete diffusion model that generates candidate T-cell receptors conditioned on a specific peptide–MHC target — designing receptors for immune targets it has never encountered.
The key idea: protein sequences aren't treated as arbitrary strings. The forward diffusion corrupts amino acids using a BLOSUM-based transition matrix derived from evolutionary substitution patterns, while the reverse process incorporates a TCR–epitope binding predictor to steer generation toward binding candidates. A notable showcase of generative AI for immunotherapy design.
More from Research
- USENIX Launches SAIS Conference Focused on Secure Agentic AI Systems — EarlenceF · 2026-08-28
- Margaret Mitchell's new paper: AI agents are pushing humans out of the loop — mmitchell_ai · 2026-08-28
- Robotics RL tutorial: adding commands to observations makes the balance bot remote controlled — ShawnHymel · 2026-08-28
- UniTS framework uses diffusion models to accelerate reaction discovery — bravo_abad · 2026-08-28
- Sai Agent Tops OSWorld 2.0 Benchmark at Lower Cost — TianbaoX · 2026-08-28
- Diverse pretraining reduces need for embodiment-specific data in robotics — chris_j_paxton · 2026-08-28