arXiv cs.AI· Bogdan Zagribelnyy, Ivan Ilin, Nikita Bondarev, Anton Morgunov, Arkadii Lin, Maksim Kuznetsov, Rim Shayakhmetov, Vladimir Aladinskiy, Alex Aliper, Alex Zhavoronkov·· 4 小时前AI 评分37
URSA:面向逆合成评估的化学感知基准
URSA: Chemistry-Aware Benchmark for Utilitarian Retrosynthesis Assessment
AI 导读
研究者推出 URSA(Utilitarian RetroSynthesis Assessment)评估框架,可从收敛到商业可得起始原料的正式角度与化学合理性角度评测逆合成路线,模拟专家化学家的评审方式。
正文 · 原文
Abstract:Synthesis planning aiming to find pathways of reactions for a target molecule is one of the most important and challenging tasks in drug discovery. Recent progress has produced both specialized deep-learning retrosynthesis systems and general-purpose large language models, but objective comparison remains difficult due to the lack of flexible, chemically interpretable benchmarking protocols. In the current study, we are introducing the URSA (Utilitarian RetroSynthesis Assessment) evaluation framework that provides the opportunity to benchmark the synthetic routes not only from a formal perspective, such as convergence to commercially available starting materials, but also from a chemical plausibility perspective, mimicking the way expert chemists evaluate the reactions and routes. The study covers a comprehensive evaluation of both conventional end-to-end retrosynthesis solutions and LLMs for the synthesis planning task on a set of novel, diverse target molecules with undisclosed synthetic routes, which represent realistic tasks in the daily drug design routine. Although LLMs can support high-level strategic planning and even achieve SOTA performance in the OOD set, the Pareto frontier for estimated run cost and chemically plausible route solvability consists entirely of specialized retrosynthesis planners.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL) |
| Cite as: | arXiv:2607.04688 [cs.LG] |
| (or arXiv:2607.04688v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2607.04688 arXiv-issued DOI via DataCite |
Submission history
From: Maksim Kuznetsov [view email]
[v1]
Mon, 6 Jul 2026 05:32:09 UTC (3,412 KB)
[v2]
Fri, 13 Nov 2026 18:36:28 UTC (3,630 KB)
来源:arXiv cs.AI · arxiv.org