arXiv subjects

Andrea Hunklinger

Publications and source records attributed to Andrea Hunklinger.

3 recordsLinked to original sources

arXiv2026

SEISMO: Explanation-Aware, Trajectory-Conditioned LLM Agents for Sample-Efficient Molecular Optimisation

Optimizing 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 for inference-time molecular optimisation that turns information routinely available alongside the oracle score, but discarded by existing methods, into an explicit guidance signal. Rather than treating the oracle as a scalar black box, SEISMO conditions each proposal on a natural-language task description, the full optimization trajectory, and machine-readable feedback derived from post-hoc explainability methods and sub-score decompositions. Across a wide range of drug-discovery-relevant tasks, this consistently improves sample efficiency over existing optimisers as well as zero-shot LLM generation, with gains growing as explanatory feedback is enriched. In practice, medicinal chemists can inspect the agent's reasoning and intervene to steer generation in natural language, keeping them central to molecular optimisation projects.

cs.AI↗

arXiv2025

Toward the Explainability of Protein Language Models

Protein language models (pLMs) excel in a variety of tasks that range from structure prediction to the design of functional enzymes. However, these models operate as black boxes, and their underlying working principles remain unclear. Here, we survey emerging applications of explainable artificial intelligence (XAI) to pLMs and describe the potential of XAI in protein research. We divide the workflow of protein AI modeling into four information contexts: (i) training sequences, (ii) input prompt, (iii) model architecture, and (iv) input-output pairs. For each, we describe existing methods and applications of XAI. Additionally, from published studies we distil five (potential) roles that XAI can play in protein research: Evaluator, Multitasker, Engineer, Coach, and Teacher, with the Evaluator role being the only one widely adopted so far. These roles aim to help both protein scientists and model developers understand the possibilities and limitations of implementing XAI for predictive and generative tasks. While our analysis focuses on pLMs, both this categorization and roles are broadly applicable to any other model architectures. We conclude by highlighting critical areas of application for the future, including risks related to security, trustworthiness, and bias, and we call for community benchmarks, open-source tooling, domain-specific visualizations, and wet-lab characterization to advance the interpretability of protein AI.

q-bio.BM↗

arXiv2024

Reinforcement Learning Guides Generative Protein Language Models

Protein engineering can optimize molecules for biotechnology and therapeutics, but navigating the high-dimensional sequence landscape remains challenging. Protein language models (pLMs) have shown to to generate functional proteins far from natural sequences, yet their outputs tend to reflect prevalent properties in training data, limiting discovery of rare properties such as high catalytic activity or thermostability. Here, we introduce ProtRL, a reinforcement learning framework for pLMs that iteratively updates model parameters to maximize externally defined reward functions. Across diverse design tasks, ProtRL shifts generation toward specified objectives while maintaining sequence diversity. We demonstrate the optimization of target folds, bounded and continuous fitness predictors, and multi-objective optimization in binder design. As a proof of concept, we applied ProtRL to experimental feedback for the engineering of epidermal growth factor receptor binders. Testing fewer than 100 designed variants across the experimental campaign, ProtRL-guided optimization provided a final round in which 16 of 22 variants bound EGFR. The best variant showed a dissociation constant of 5.5 nM, representing a nine-fold improvement over wild-type EGF and higher affinity than previously reported EGF variants identified through substantially larger screening campaigns. Our code and models are publicly available at github.com/AI4PDLab/ProtRL

q-bio.BM↗