Model building

From Media Wiki

Model building is the process of interpreting an electron density map in terms of atomic coordinates. We cannot see the atomic model itself, only the electron density. However the description in terms of atoms and bonds is more useful from a chemical or biological point of view, and therefore we try to infer it from the density. At low resolutions this process is not without error, and requires the application of chemical knowledge. Unexpected features (e.g. in the solvent region) are often not identifiable unambiguously.

Model building can be performed automatically by a variety of methods (e.g. software such as ARP/WARP, RESOLVE, BUCCANEER, TEXTAL), or using graphical tools such as Coot and O. When building a protein model manually, it is normal to trace the main chain first, align the sequence with the chain, build the side chains, and then build waters and ligands. Tracing the main chain is often performed by use of a 'skeleton' trace of ridges in the density, in combination with a 3.8A 'baton', which is used to connect consecutive Cα positions. Once the Cαs have been identified, the connecting peptide atoms can be added automatically.

At high resolutions, the model can be built by picking electron density peaks.

--Kevin Cowtan 02:37, 25 April 2008 (CDT)