Developing Deep Learning-based Large-scale Organic Reaction Classification Model via Sigma-profiles

The "<b>Train_AE.zip</b>" contains the scripts for training an auto-encoder.The "<b>Train_DL_Models.zip</b>" contains the scripts for training deep learning-based models.The "<b>sigma_profiles_dict.npy</b>" contains the sigma-profiles of millions of different molecules. The SMILES of a molecule is used as key to query the corresponding sigma-profiles.The "<b>sorted_agent_dict.npy</b>" contains the statistical results of USPTO_TPL dataset concerning the frequency of occurrence of agents. The agents are shown in an descending manner.The "<b>sorted_agent_combination_dict.npy</b>" contains the statistical results of USPTO_TPL dataset concerning the frequency of occurrence of agent combinations. The combinations are shown in an descending manner.<br>The "<b>USPTO_TPL_own_version.xlsx</b>" contains the reactions that used for training/validation/testing.<br>

<b>Train_AE.zip</b> 包含用于训练自编码器（auto-encoder）的脚本文件。<b>Train_DL_Models.zip</b> 包含用于训练基于深度学习模型的脚本文件。<b>sigma_profiles_dict.npy</b> 存储了数百万种不同分子的σ轮廓（sigma-profiles），以分子的简化分子线性输入规范（SMILES）作为键即可查询对应的σ轮廓。<b>sorted_agent_dict.npy</b> 包含USPTO_TPL数据集的试剂出现频率统计结果，统计结果按试剂出现频率降序排列。<b>sorted_agent_combination_dict.npy</b> 包含USPTO_TPL数据集的试剂组合出现频率统计结果，统计结果按试剂组合出现频率降序排列。<br><b>USPTO_TPL_own_version.xlsx</b> 包含用于训练、验证及测试的化学反应数据。