2.85M reaction mechanisms from the USPTO reaction dataset

Reaction mechanisms of 226,669 reactions in USPTO-480K reaction dataset¹ originally extracted from US patents (USPTO) reaction dataset² generated by MechFinder³ to train Reactron. This dataset includes approximately 2.85 million mechanistic steps.Reference:1.Jin, W., Coley, C., Barzilay, R., &amp; Jaakkola, T. (2017). Predicting organic reaction outcomes with weisfeiler-lehman network. <i>Advances in neural information processing systems</i>, <i>30</i>.Lowe, Daniel (2017). Chemical reactions from US patents (1976-Sep2016). figshare. Dataset. https://doi.org/10.6084/m9.figshare.5104873.v1Chen, S., Babazade, R., Kim, T., Han, S., &amp; Jung, Y. (2024). A large-scale reaction dataset of mechanistic pathways of organic reactions. <i>Scientific Data</i>, <i>11</i>(1), 863.<br>

本数据集收录了USPTO-480K反应数据集¹中的226,669个反应的反应机理，该数据集最初源自由MechFinder³生成的美国专利（USPTO）反应数据集²，用于训练Reactron。本数据集包含约285万个反应机理步骤。
参考文献：
1. Jin, W., Coley, C., Barzilay, R., & Jaakkola, T. (2017). 基于魏斯费勒-莱曼网络（Weisfeiler-Lehman Network）预测有机反应产物. 《神经信息处理系统进展》, 30.
2. Lowe, Daniel (2017). 美国专利中的化学反应数据（1976年—2016年9月）. figshare. 数据集. https://doi.org/10.6084/m9.figshare.5104873.v1
3. Chen, S., Babazade, R., Kim, T., Han, S., & Jung, Y. (2024). 面向有机反应机理路径的大规模反应数据集. 《科学数据》, 11(1), 863.