Curated Experimental MOF Structures with Assigned Charges for Simulation

This dataset contains curated experimental metal–organic framework (MOF) structures with assigned atomic charges, prepared for adsorption and binding-site simulations. It was used in the study “How well do conventional atomistic simulations predict adsorption binding sites in MOFs compared to experiment?” The dataset enables direct benchmarking of common force fields and charge-assignment schemes. Directory Naming Convention Each directory label encodes the key details of the corresponding simulation. The label is composed of sections separated by a double underscore (__). The directory labels closely match the MOF names reported in Table 1, using filesystem-compatible formatting, following the convention: <Structure_Name>__<Guest_Molecule>__<Temperature>__<Pressure> Structure Preprocessing All structures were obtained from experimental crystallographic data, either from the Cambridge Structural Database (CSD) or from the supplementary information of the original publications. Each structure was carefully corrected to remove artifacts such as missing ligands, disordered fragments, and missing protons. The corrected structures were converted to P1 symmetry and their protons were geometry-optimized using DFT, as detailed in the Methods section of the paper. Atomic charges were then assigned using several commonly employed charge-assignment methods in adsorption simulations. Each folder contains the following charge-assigned cif files: Optimized structure: *_opt.cifDDEC6 charges (DFT-fitted via Chargemol): *_opt_ddec6.cifREPEAT charges (DFT-fitted): *_opt_repeat.cifMEPO-QEQ empirical charges: *_opt_mepoqeq.cifMEPO-ML machine-learned charges (trained on REPEAT): *_opt_mepoml.cif Reference CSV File These charge-assigned experimental structures are ready for GCMC, isotherm prediction, or binding-site localization studies. The dataset also includes a reference index file named Experimental_MOF_Dataset.csv. This CSV file provides a readable mapping between each directory and its reference DOI. In addition, the columns below were added as well: In_Table1_Set: Identifies the MOFs included in Table 1 of the main text, where the core binding-site analyses were conducted using UFF and REPEAT. In_RMSE_RMSD_Set: Marks the MOFs used to calculate RMSE and RMSD differences across all force-field and charge-method combinations. Has_Isotherms: Indicates the structures for which adsorption isotherms were computed and included in the study.本数据集包含经过精心甄选的、已分配原子电荷的实验性金属有机骨架（metal–organic framework, MOF）结构，专为吸附与结合位点模拟任务构建。该数据集曾应用于题为"相较于实验，传统原子模拟在预测金属有机骨架中的吸附结合位点方面表现如何？"的研究中，可用于对常见力场与电荷分配方案开展直接基准测试。 目录命名规范 每个目录标签均编码了对应模拟的关键细节，标签由双下划线（__）分隔的多个部分组成。目录标签采用兼容文件系统的格式，与正文章节表1中报告的MOF名称高度匹配，遵循以下约定： <结构名称>__<客体分子>__<温度>__<压力> 结构预处理 所有结构均源自实验晶体学数据，获取途径包括剑桥晶体学数据库（Cambridge Structural Database, CSD）以及原始文献的补充材料。每一个结构均经过严格修正，以去除配体缺失、片段无序、质子缺失等伪影。修正后的结构被转换为P1对称性，并使用密度泛函理论（Density Functional Theory, DFT）对其质子进行几何优化，具体细节可参见论文的方法章节。随后，采用吸附模拟中几种常用的电荷分配方法为结构分配原子电荷。 每个文件夹均包含以下已分配电荷的晶体学信息文件（CIF）： - 优化后结构：*_opt.cif - DDEC6电荷（通过Chargemol拟合的DFT电荷）：*_opt_ddec6.cif - REPEAT电荷（DFT拟合电荷）：*_opt_repeat.cif - MEPO-QEQ经验电荷：*_opt_mepoqeq.cif - MEPO-ML机器学习电荷（基于REPEAT数据集训练）：*_opt_mepoml.cif 参考CSV文件 这些已分配电荷的实验结构可直接用于巨正则蒙特卡洛（Grand Canonical Monte Carlo, GCMC）模拟、吸附等温线预测或结合位点定位研究。本数据集还包含一个名为`Experimental_MOF_Dataset.csv`的参考索引文件，该CSV文件提供了每个目录与其对应参考文献DOI之间的可读映射关系。此外，该文件还新增了以下列： 1. `In_Table1_Set`：用于标识正文章节表1中包含的MOF，核心结合位点分析采用通用力场（UFF）与REPEAT方法完成。 2. `In_RMSE_RMSD_Set`：标记用于计算所有力场与电荷方法组合下的均方根误差（Root Mean Square Error, RMSE）与均方根偏差（Root Mean Square Deviation, RMSD）差值的MOF。 3. `Has_Isotherms`：表示已计算吸附等温线并纳入本研究的结构。