LPT: Orchestrating bounded agentic loops for automated protein binder design
Current large language models (LLMs) can reason over scientific information and use code-based tools, but their training knowledge is broad rather than specialized, can be out of date and is hard to attribute to sources. Furthermore, unconstrained agentic workflows and unlimited tool use can consume large amounts of tokens. Here we present LPT (little protein tiger), a framework for fixed…
LPT (little protein tiger) is a novel framework designed to streamline the process of de novo protein binder design using large language models (LLMs). By orchestrating LLM stages into bounded agentic loops, LPT enables the use of a curated literature database to recall specialized scientific information, while limiting the amount of tokens consumed and computational resources required.
This approach has proven effective in three end-to-end design campaigns, where LPT autonomously identified design intents, selected targets, structures, and hotspot residues to create candidate binders for disrupting CALCRL/RAMP1, PD-1/PD-L1, and PVR-TIGIT interactions. In tests, LPT with database access correctly answered 41 out of 48 questions, while a model without database access provided no answers.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
