1 million hypothetical MOFs

This project includes CIF files of about 1 million hypothetical MOFs (exactly 1,118,074 structures) generated by PORMAKE, which were used for pre-training MOFTransformer. The details of generating MOFs and optimization with UFF forcefield are summarized in the maunscript of MOFTransformer. <br> An example of file format: puc+N131+E128.cif The first token (e.g., pcu) means topology and other tokens mean names of building blocks, which can be identified in supporting information of the PORMAKE paper [computational screening of trillions of metal-organic frameworks for high-performance methane storage])

本项目包含由PORMAKE生成的约100万个假想金属有机框架（Metal-Organic Frameworks，MOFs，准确数量为1,118,074个结构）的CIF（晶体学信息文件，Crystallographic Information File）文件，上述文件被用于MOFTransformer的预训练。关于MOFs的生成方法以及采用UFF力场进行结构优化的详细细节，已在MOFTransformer的相关研究论文中进行了汇总说明。
文件格式示例：puc+N131+E128.cif。其中首个Token（Token）代表拓扑类型，其余Token代表结构基元的命名标识，相关命名规则可在PORMAKE论文的支撑材料中查阅，该论文的研究主题为《面向高性能甲烷存储的万亿级金属有机框架计算筛选》