protein structure refinement using structure factor file or any other methods

Question

[elmira b]

Hello All,

I am using PHENIX (http://www.phenix-online.org/) to run MolProbity Analysis and refine a predicted protein structure. PHENIX needs the structure factor data for my protein to start analysis.

Structure factor data provides a topology of the protein such as connected atoms, important torsion angels, etc.

There is almost no structure factor data associated to my protein in the protein data bank and no NOE in biological Magnetic resonance bank.

Does anyone know how to refine my protein structure and fix issues such as poor rotamers, ramachandran outliers, bad bonds and angles?

Thanks,
Elmira

Answer 1

[nathaniel.echols]

MolProbity (in Phenix, or from the web server) should run without a structure factor file, as far as getting validation statistics is concerned. However, Phenix is not intended to improve predicted models, only experimentally derived ones. You may have some luck simply running the geometry regularization app but this will only fix the really basic issues like bad bonds. There are many other software packages that are specifically designed to refine theoretical models and are much more thorough - Rosetta would be an excellent choice to start with.

[elmira b]

Rosetta Design and Chimera energy minimization to refine small proteins: http://rosettadesign.med.unc.edu/

In general GROMAC has a good tutorial for refining proteins without NOE data: http://www.gromacs.org/@api/deki/files/203/=tutorial.pdf

Also, there is a paper explaining other good methods to refine protein structures without NOE data: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0108888

In this paper they introduce STAP algorithm which can be run here: http://psb.kobic.re.kr/NmRe/index.cgi

Other algorithms: RECOORD, AMBER.