SEISMO: Increasing Sample Efficiency in Molecular Optimization with a Trajectory-Aware LLM Agent

2026-01-31

AI Summary

Molecular optimization is often hindered by expensive and inefficient experimental evaluations (oracles), resulting in poor sample efficiency. To address this, this work proposes SEISMO—a trajectory-aware large language model agent that uniquely integrates complete optimization trajectories with structured domain knowledge to enable rigorous online molecular optimization during inference. After each oracle query, SEISMO immediately updates its strategy, generating new molecules by leveraging natural language task descriptions, scalar rewards, and explanatory feedback. Notably, it operates without batch or population-based learning. On 23 Practical Molecular Optimization benchmark tasks, SEISMO achieves 2–3× higher area under the optimization curve than existing methods, frequently approaching optimal solutions within just 50 oracle calls. Incorporating explanatory feedback further enhances its efficiency in medicinal chemistry tasks.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Optimization
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Optimizing the structure of molecules to achieve desired properties is a central bottleneck across the chemical sciences, particularly in the pharmaceutical industry where it underlies the discovery of new drugs. Since molecular property evaluation often relies on costly and rate-limited oracles, such as experimental assays, molecular optimization must be highly sample-efficient. To address this, we introduce SEISMO, an LLM agent that performs strictly online, inference-time molecular optimization, updating after every oracle call without the need for population-based or batched learning. SEISMO conditions each proposal on the full optimization trajectory, combining natural-language task descriptions with scalar scores and, when available, structured explanatory feedback. Across the Practical Molecular Optimization benchmark of 23 tasks, SEISMO achieves a 2-3 times higher area under the optimisation curve than prior methods, often reaching near-maximal task scores within 50 oracle calls. Our additional medicinal-chemistry tasks show that providing explanatory feedback further improves efficiency, demonstrating that leveraging domain knowledge and structured information is key to sample-efficient molecular optimization.

Problem

• molecular optimization
• sample efficiency
• oracle efficiency
• drug discovery
• molecular design

Innovation

• sample-efficient optimization
• trajectory-aware LLM agent
• online molecular optimization
• explanatory feedback
• oracle-efficient drug discovery