 |
 |
Dear CCP4bb, since the discussion about H-atoms is on, I wanted to ask about what I saw during my refinements: I did refinement with phenix of my 1.9-2.0 Angstroem structures and included the hydrogens (riding). However, when I checked on the statistics (refinement close to the end), the average B-factor was extremely high (in phenix.polygon it was higher than with any other structure in similar resolution range). It makes sense though that this happens, if you have a residue which has a high B-factor and carries a lot of hydrogens the average B-factor will raise quite a lot (since the B-factor of hydrogens is calculated 1-1.5x of B-factor from the atom it sits on), right? (When I removed the hydrogens again, the average B-factor was fine...) My question is now did I do something wrong in my refinement (-> do I have to change something that this does not happen), or is this something everybody sees? If this is common, what would happen (during evaluation) if you want to publish a structure and the statistics show such a high average B-factor? Is it better to have hydrogens on, but a bad average B, or no hydrogens on and a good average B... Obviously, I never published a structure ;) Thanks for sharing your opinion! Sara -----Ursprüngliche Nachricht----- Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag von Dirk Kostrewa Gesendet: Dienstag, 12. Januar 2010 16:38 An: [log in to unmask] Betreff: Re: [ccp4bb] Phaser: removing H in PDB increases the RFZ ?! Dear CCP4ers, <snip> > I think that characterize hydrogens as "transparent" to x-rays is > somewhat misleading. There are plenty of examples where hydrogens are > seen in ultrahigh resolution structures. Modern refinement programs all > use riding hydrogens and it improves models. More precise statement is > that at not-ultrahigh (non-atomic) resolution (i.e. worse than 1.2A) > optical resolution is too low to distinguish two atoms separated by ~1A > (typical value of X-H covalent bond). > </snip> if one looks at atomic scattering curves, hydrogens contribute to X-ray scattering mainly at low resolution, which makes them so difficult to detect even at atomic resolution. I think, it is generally a good idea to include hydrogens as riding atoms in all crystallographic tasks, and especially during refinement, both because of their scattering contribution and because of their geometrical contribution, if the latter is considered in form of anti-bumping restraints and even more important in form of hydrogen bonds. Best regards, Dirk. -- ******************************************************* Dirk Kostrewa Gene Center, A 5.07 Ludwig-Maximilians-University Feodor-Lynen-Str. 25 81377 Munich Germany Phone: +49-89-2180-76845 Fax: +49-89-2180-76999 E-mail: [log in to unmask] WWW: www.genzentrum.lmu.de *******************************************************