Model Building

Why

Determining the structure of a macromolecule typically requires building an atomic model by fitting to maps obtained from experimental phasing or molecular replacement. For many cases, automated methods can be used to build models for both proteins and nucleic acids.

The accuracy and completeness of the model building is a function of the quality of the starting electron density map, and less critically the resolution of the data. At low resolution (3.0 Å or worse), partial models are usually built, which require manual completion.

How

In Phenix, several programs aid in automated model building; the main one is phenix.autobuild. The model building process integrates phase improvement using density modification methods with automated map interpretation in order to build and refine models. This minimally requires an experimental density map, experimental diffraction data, and the sequence of the molecule(s) in the crystal. The program outputs a model file with the automatically built model and the map coefficients for the best electron density map created during the model building process.

How to use the phenix.autobuild GUI: Click here

Phenix reference manual for phenix.autobuild

Common issues

• Frequently asked questions about model building

Related programs

• phenix.find_helices_strands: This program very rapidly locates protein secondary structure elements in an electron density map. This is extremely useful for quickly confirming that a map contains macromolecular features, or for interpreting low-resolution (4 Å or worse) maps. Optionally, you can provide the amino acid sequence and the program will fit residues where density permits.
• phenix.fit_loops: This program docks missing loops into protein structures given the current incomplete model, the sequence, and an electron density map. Model building procedures are sometimes unable to fit models to poor density in loop regions, especially in the initial stages of map interpretation. As the model becomes more complete and maps improve, phenix.fit_loops can perform simple real-space refinement to clean up the structure.
• phenix.map_to_model: This program is designed for lower-resolution model building. It is suitable for both crystallographic maps and cryo-EM maps. Taking into account symmetry used to create a map, it builds only the unique part of the model and expands to the remaining NCS-related copies.
• phenix.erraser: This program uses ERRASER (Enumerative Real-space Refinement ASsisted by Electron-density under Rosetta) to rebuild RNA structures by combining Rosetta and Phenix. By supplementing the Rosetta RNA scoring function with electron-density restraint, ERRASER can confidently reduce the errors in RNA crystallographic models while retaining a good fit to the diffraction data.