Guacamol

该数据集是一个综合性基准，涵盖了与分子生成和优化相关的任务，旨在评估代理在有效探索化学空间方面的能力。此外，该数据集还包括了在药物设计中常见的任务，如优化定量估算药物相似性（QED）和受惩罚的logP等指标。其中一项具体任务是，评估生成的分子与目标分子塞来昔布的相似性。

This dataset is a comprehensive benchmark encompassing tasks related to molecular generation and optimization, designed to assess the capability of AI agents to efficiently explore chemical space. Additionally, this dataset incorporates tasks prevalent in drug design, such as optimizing metrics including Quantitative Estimate of Drug-likeness (QED) and penalized logP. One specific task entails evaluating the similarity between generated molecules and the target molecule celecoxib.

搜集汇总

背景与挑战

背景概述

Guacamol是一个用于分子设计模型基准测试的开源Python包，提供分布匹配和目标导向两种基准测试方法，并基于ChEMBL数据集提供标准化数据。