Section 2.2.6. CIF metadata and dictionary compliance

The development of several CIF dictionaries and fields of application has rapidly progressed beyond the specific purpose of describing a small-molecule or inorganic crystal structure for which CIF was devised. With these, a variety of application-specific metadata approaches have evolved to characterize the role of a particular CIF within a family of possible applications. These approaches use data definitions in dictionaries in which enumerated codes identify the file relationships. The mmCIF dictionary (see Chapter 3.6 ) allows informal identification of `external reference files' which act as libraries of standard molecular geometry. The pdCIF dictionary (see Chapter 3.3 ) specifies identifiers that may be included within data blocks of external files containing calibration results. It is the responsibility of the file users to manage a lookup table or database between the referenced identifiers and the location of the files to which they pertain.

Two categories of data items currently exist in the core dictionary to allow a file to indicate its relationship to CIF dictionaries and other data files. (Equivalent categories are also present in the mmCIF dictionary.) AUDIT_CONFORM is a category of data names identifying the dictionaries that hold definitions of the data names in the current CIF. Particularly where the referenced dictionaries include any of the various public dictionaries described in Part 3 of this volume, this serves to establish the discipline within the broad fields of crystallography, structural biology and structural chemistry to which the data are most relevant.

The category AUDIT_LINK allows an informal textual description of the relationship between the data blocks within the current file. It is `informal' in the sense that the relevant data items are free-text in nature. It would surely be useful to have a catalogue of more specific designations to allow automated software to track such relationships as the separate reference and modulated structures in an incommensurate compound, or the multiple trial refinements of a protein structure. The challenge is to determine and classify such standard relationships between data blocks.

In the future it is hoped that a common approach to metadata will be developed to enable all CIF instantiations to be uniquely identified and interrelated. Development of standard descriptions of the relationships between structural entities of this sort (reference geometries, calibration results, partial refinements, modulated superposed structures etc.) will be an important stage in the formalization of complete CIF metadata, and will become an important step towards categorization of data entities needed for interoperability between different file formats and across a wide range of scientific disciplines.