Python-based
Hierarchical
ENvironment
for
Integrated
Xtallography
Why
Once an atomic model has been built, you need to optimize it to best fit the experimental data while also preserving good agreement with prior chemical knowledge. This process alternates between automated refinement, validation, and either manual or automated model corrections.
A target function guides the refinement by linking the model parameters to the experimental data and by scoring the model-versus-data fit. For cryo-EM data, refinement of the model is done in real space and the target function is formulated in terms of a three-dimensional map. Because there are generally too many model parameters, refinement requires additional restraints that modify the target function by creating relationships between independent parameters.
How
The program used to refine atomic models in real space is phenix.real_space_refine. The algorithm uses a simplified refinement target function that speeds up calculations, so optimal data-restraint weights can be identified with little runtime cost. The procedure is robust, working at resolutions from 1 to 6 Å. The input and output models can be in PDB or mmCIF formats. The input map can be in CCP4 or MTZ format.
This program makes use of restraints on covalent geometry, secondary structure, Ramachandran plots, rotamer outliers, and internal molecular symmetry. The default mode performs gradient-driven minimization of the entire model. However, optimization can also be performed using simulated annealing, morphing, rigid-body refinement, and/or systematic side-chain improvement — at the potential expense of longer runtimes.
How to use phenix.real_space_refine: Click here
Common issues