ModelCIF provides the data representation for describing structural models of macromolecules derived from computational methods. ModelCIF is developed and maintained by the ModelCIF Working Group. These models may be derived from already existing structural templates (also known as homology modeling or comparative modeling methods) or can be obtained from ab initio modeling methods.
ModelCIF data standards are used or supported by various model repositories (e.g. ModelArchive, ModBase, SWISS-MODEL Repository, AlphaFold Protein Structure Database), modeling applications (e.g. Modeller, SWISS-MODEL, D-I-TASSER, AlphaFold, Boltz, Chai), and visualization software (e.g. Mol*, ChimeraX).
The python-modelcif library facilitates the programmatic creation of ModelCIF files and includes convenience classes to streamline this process. A collection of ModelCIF conversion scripts, mostly based on python-modelcif, as well as a ModelCIF validation tool based on RCSB's CifCheck, are available in a dedicated modelcif-converters code repository maintained by the ModelArchive team.
ModelCIF is an extension of the PDBx/mmCIF dictionary. The PDBx/mmCIF dictionary is used by the wwPDB to archive structural models of macromolecules determined using X-ray crystallography, NMR spectroscopy and cryogenic electron microscopy.
ModelCIF dictionary is distributed in two forms:
- ModelCIF dictionary combined with definitions in the parent PDBx/mmCIF dictionary (available at dist/mmcif_ma.dic) and
- ModelCIF extension separately without the definitions in the parent PDBx/mmCIF dictionary (available at dist/mmcif_ma_ext.dic)
The primary publication about ModelCIF is available at https://doi.org/10.1016/j.jmb.2023.168021.
For more details regarding the dictionary, see the ModelCIF dictionary documentation and the mmCIF resources website.