T-REX: open-source LLM agent controller orchestrates de novo protein binder design
ArashVahdat · x · 2026-09-25
The Stanford ml-struct-bio team (first author Minkyu Jeon, amplified by Arash Vahdat) released its first protein design paper, T-REX (Target-adaptive Rescue-Explore-eXploit): an agentic campaign controller for high-throughput de novo protein binder design, fully open source with a preprint.
Key ideas:
- Frames protein design as an online allocation problem over heterogeneous tools (Proteina-complexa, Boltzgen, Bindcraft, etc.), with agents reasoning over which tools and settings to run;
- Planner and Supervisor LLMs propose and prioritize jobs while deterministic code validates them and manages generation, redesign, and evaluation—efficiently saturating GPUs;
- A granular agent taxonomy: Rescue / Explore / eXploit roles;
- Campaigns need only four user choices (target, GPUs per node, duration, output dir) via trex setup -> init -> check -> submit;
- Verified on Linux x8664 with NVIDIA H100 GPUs under Slurm.
A solid AI4Science case of LLM agents acting as lab experiment orchestrators.
More from coding & agent
- LangChain webinar: fast, cheap Jev models make real-time filtering and guardrails viable — LangChain · 2026-09-25
- Peter Yang launches 25-lesson AI workflow course with 40+ copy-paste prompts — petergyang · 2026-09-25
- Claude Code's most-hated flaw: message queuing, 244 issues and still broken after months; Codex does it right — sytelus · 2026-09-25
- Dev runs parallel subagents to build 3D games; Claude writes its own GPU/memory task allocator — BLUECOW009 · 2026-09-25
- Cognition AI says it has surpassed $1 billion in annualized revenue run rate — Polymarket · 2026-09-25
- opencode makes $60/month of DeepSeek v4.1 Flash usage permanent — intellectronica · 2026-09-25