Supplementary file 1_AlgicideDB: a comprehensive database enhanced by large language models for algicide management and discovery.docx

Harmful algal blooms (HABs) are increasing in frequency and intensity worldwide, posing significant threats to aquatic ecosystems, fisheries, and human health. While chemical algicides are widely used for HABs control due to their rapid efficacy, the lack of systematic data integration and concerns over environmental toxicity limit their broader application. To address these challenges, we developed AlgicideDB, a manually curated database containing 1,672 algicidal records on 542 algicides targeting 110 algal species. Using this database, we analyzed the physicochemical properties of algicides and proposed an algicide-likeness scoring function to facilitate the exploration of compounds with antialgal properties. Additionally, we evaluated the acute toxicity of algicidal compounds to non-target aquatic organisms of different trophic levels to assess their ecological risks. The platform also incorporates a large language model (LLM) enhanced by retrieval-augmented generation (RAG) to address HAB-related queries, supporting decision-making and facilitating knowledge dissemination. AlgicideDB, available at http://algicidedb.ocean-meta.com/#/, serves as an innovative and comprehensive platform to explore algicidal compounds and facilitate the development of safe and effective HAB control strategies.

有害藻华（Harmful algal blooms, HABs）在全球范围内的发生频率与强度均呈上升趋势，对水生生态系统、渔业及人类健康构成严重威胁。尽管化学除藻剂因起效迅速而被广泛用于HABs防控，但由于缺乏系统化的数据整合体系，且存在环境毒性方面的顾虑，其进一步推广应用受到限制。为应对上述挑战，本研究开发了AlgicideDB——一款经人工整理的数据库，涵盖了针对110种藻类的542种除藻剂的1672条除藻记录。依托该数据库，本研究分析了除藻剂的理化性质，并提出了类除藻剂评分函数，以助力具有抗藻活性的化合物的筛选探索。此外，本研究还评估了除藻类化合物对不同营养级非靶标水生生物的急性毒性，以研判其生态风险。该平台还集成了经检索增强生成（Retrieval-Augmented Generation, RAG）优化的大语言模型（Large Language Model, LLM），可响应与HABs相关的各类咨询查询，为决策制定提供支持并促进知识传播。AlgicideDB可通过http://algicidedb.ocean-meta.com/#/访问，作为一款兼具创新性与全面性的平台，可为除藻化合物的探索以及安全高效的HABs防控策略开发提供有力支撑。