C–N cross-coupling

化学家通常通过将催化剂应用于一系列简单化合物来发现反应。在药物研究中调整这些反应以容忍更多的结构复杂性非常耗时。Ahneman等人报告称，机器学习可以提供帮助。他们使用高通量数据集训练了一种随机森林算法，以预测在 C-N 键形成过程中哪种特定的钯催化剂最能耐受异恶唑（具有 N-O 键的环状结构）。这些预测还有助于指导对催化剂抑制机制的分析。

Chemists typically discover reactions by applying catalysts to a series of simple compounds. Optimizing these reactions to tolerate greater structural complexity in pharmaceutical research is extremely time-consuming. Ahneman et al. reported that machine learning could offer assistance. They trained a random forest algorithm using high-throughput datasets to predict which specific palladium catalyst would best tolerate isoxazoles (cyclic structures containing an N-O bond) during C-N bond formation. These predictions also help guide the analysis of catalyst inhibition mechanisms.

搜集汇总

背景与挑战

背景概述

该数据集聚焦于C-N交叉偶联反应，通过高通量数据训练随机森林算法，旨在预测钯催化剂在异恶唑存在下的耐受性，并辅助分析催化剂抑制机制。相关研究成果于2018年发表在《Science》期刊上。