Abstract:Applications of machine learning in chemistry are often limited by the scarcity and expense of labeled data, restricting traditional supervised methods. In this work, we introduce a framework for molecular reasoning using general-purpose Large Language Models (LLMs) that operates without requiring task-specific model training. Our method anchors chain-of-thought reasoning to the molecular structure by using unique atomic identifiers. First, the LLM performs a zero-shot task to identify relevant fragments and their associated chemical labels or transformation classes. In an optional second step, this position-aware information is used in a few-shot task with provided class examples to predict the chemical transformation. We apply our framework to single-step retrosynthesis, a task where LLMs have previously underperformed. Across academic benchmarks and expert-validated drug discovery molecules, our work enables LLMs to achieve high success rates in identifying chemically plausible reaction sites ($\geq90\%$), named reaction classes ($\geq40\%$), and final reactants ($\geq74\%$). Ultimately, our work establishes a general blueprint for applying LLMs to challenges where molecular reasoning and molecular transformations are key, positioning atom-anchored LLMs as a powerful solution for data-scarce chemistry domains.
| Comments: | Alan Kai Hassen and Andrius Bernatavicius contributed equally to this work |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Biomolecules (q-bio.BM) |
| Cite as: | arXiv:2510.16590 [cs.LG] |
| (or arXiv:2510.16590v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2510.16590 arXiv-issued DOI via DataCite |
From: Alan Kai Hassen [view email]
[v1]
Sat, 18 Oct 2025 17:27:44 UTC (1,331 KB)
[v2]
Thu, 21 May 2026 16:21:38 UTC (1,431 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。