Search papers, labs, and topics across Lattice.
AgenticPosesRanker, a novel agentic AI framework, leverages GPT-5 and six physics-based analysis tools to rank protein-ligand docking poses. It was evaluated on a balanced benchmark of 10 protein-ligand systems, achieving 50% best-pose accuracy, matching the Smina baseline but significantly exceeding random chance. Decision attribution analysis revealed high alignment between the agent's tool weights and objective metrics, suggesting performance limitations stem from tool coverage rather than reasoning errors.
GPT-5, combined with physics-based tools, can match traditional scoring functions in ranking protein-ligand docking poses, opening avenues for interpretable curation in drug design.
Scoring functions remain the principal bottleneck in molecular docking: they routinely fail to rank near-native poses above decoys, and their composite single-score design obscures the physicochemical basis of each ranking error. We present AgenticPosesRanker, an agentic AI framework that combines six deterministic, physically grounded analysis tools (interaction fingerprinting, solvent-accessible burial, conformational strain, steric-clash detection, unsatisfied-polar-atom penalty, and chemical-identity extraction) with large-language-model (GPT-5) chain-of-thought reasoning to evaluate and rank docking poses. On a curated benchmark of ten protein-ligand systems (162 poses) balanced by construction between Smina scoring-function successes and failures, the agent achieved 50.0% best-pose accuracy, matching the design-fixed Smina baseline of 50.0% and significantly exceeding a 7.7% uniformly random baseline (p<0.001, one-sided exact binomial test). The balanced-benchmark accuracy decomposes symmetrically: the agent retained 80% (4/5) of the Smina-success systems and recovered 20% (1/5) of the Smina-failure systems, so the aggregate 50% reflects one regression offset by one recovery rather than any net improvement over the Smina reference. Decision-attribution analysis showed high alignment between the agent's self-reported tool weights and objective metric separations of the selected pose (median \r{ho} = +0.83), consistent across correct and incorrect outcomes, localising the performance ceiling to tool-suite coverage rather than reasoning inconsistency. These results establish a methodological template for evaluating agentic AI against objective ground truth in the natural sciences and position the framework as an interpretable curation layer for late-stage pose refinement in structure-based drug design.
