Should we remediate small molecule structures? If so, who should do it?

Problems can arise in crystallographic databases with errors and omissions in the representation of data that impede searches, and with errors in the actual data. While the Cambridge Crystallographic Data Centre with its Improvement Projects has solved many of the first category of problems, errors in atomic coordinates and other crystallographic data are surprisingly common. Although modern software warns of many types of error, such errors appear even in recently deposited Crystallographic Information Files. Richard Marsh found many examples of missed symmetry in assignment of the space group; such errors are now waning. Hydrogen atoms are commonly placed in calculated positions. Particularly for OH and NH groups involved in hydrogen bonds, occupancy factors may need to be reduced to 0.5 or the hydrogen atom positions may require amendment. Examples of acids and imidazole derivatives are provided, showing that sometimes only consideration of bond distances and angles at the heteroatom can distinguish between OH or NH and unprotonated O or N. Significant work by other researchers correcting mis-positioned hydrogen atoms in dihydrogen phosphates and water aggregates as well as mis-identified elements is also summarized. This review concludes with some suggestions for more comprehensive detection and correction of errors in deposited data.

晶体学数据库中可能出现两类问题：一是数据表征存在错误与遗漏，阻碍检索操作；二是实际数据本身存在误差。尽管剑桥晶体学数据中心（Cambridge Crystallographic Data Centre）依托其改进项目解决了第一类问题中的诸多难题，但原子坐标及其他晶体学数据的错误依旧相当常见，这一点出人意料。尽管现代软件可对多种类型的错误发出预警，但这类错误仍存在于近期提交的晶体学信息文件（Crystallographic Information Files）中。理查德·马什（Richard Marsh）曾发现大量空间群指派中遗漏对称性的案例，这类错误如今已逐渐减少。氢原子通常被置于计算得到的位置上。对于参与氢键作用的OH与NH基团而言，其占位因子可能需要下调至0.5，或是氢原子的位置需要进行修正。文中提供了酸类与咪唑衍生物的相关案例，表明有时仅通过考察杂原子的键长与键角，即可区分OH/NH基团与未质子化的O/N原子。其他研究者针对磷酸二氢盐、水聚集体中位置错误的氢原子以及元素识别错误所开展的大量修正工作，本文也一并进行了总结。本综述最后针对提交数据中错误的更全面检测与修正方法提出了若干建议。