The benefit of Vagabond is derived primarily on a complete redesign of the model for proteins, which involves a dramatic reduction in the number of parameters compared to conventional atomic coordinates (to about 27-40% of the original). At the moment, use cases for Vagabond include:
At the heart of the algorithms behind Vagabond is a shift to a bond-based model: the protein's positional information and flexibility is associated and propagated through the properties of the bonds. This change in parametrisation better matches that of how molecules flex and move in reality, and allows asymmetrical distributions of atom positions which cannot be captured with a single B factor.
Vagabond has higher R factors than those of other programs which also refine models to match diffraction data. The R factors will decrease in the future as the protocol improves, but Vagabond's benefits lie elsewhere. If you're looking for better R factors for a publication, Vagabond (currently) will not help you. Vagabond will provide benefit from reduction in overfitting and thus revealing information in the structure which may be masked by other methods.
Vagabond is a young project and is in constant development. Vagabond development relies very heavily on feedback from the community, so please get in touch with your thoughts and problem structures.
Talk at CCP4 study weekend 2020
At the CCP4 study weekend 2020, I had the fortunate opportunity to give a talk on Vagabond and the underlying methodology: