[phenixbb] Whic mtz file to use and parameters for phenix.refine (GUI) to set?

Pavel Afonine pafonine at lbl.gov
Mon Jul 19 17:27:34 PDT 2010

  Hi Ivan,

> I thought you initially do not add waters until you have a reasonable 
> density fit with the protein molecules. Is it ok to update waters 
> right from the begining starting from an MR solution?

yes and no.

Yes, it is bad to add waters into poor map and very partial model IF you 
and the software you use are careless (for example, treat once placed 
water is God given and never critically re-think it at all subsequent 
steps). This is because you will have to deal with model bias later on.

No, it is good to add waters, because by adding waters (even temporarily 
into some residue side chain or ligand density), you in turn improve the 
map and the map improvement leads to a chance to build some more model 
and/or improve existing one in a next cycle, and building some model 
improves the map some more, and so on. The important thing is that you 
and the software you use should treat waters simply as density peaks 
(and not actual atoms) and do not hesitate to remove waters as necessary 
and build macromolecular into the density they are taking.

I guess AutoBuild does exactly this. phenix.refine option "fix_rotamers" 
does this too. I'm not sure if all tools follow this yet, but I hope it 
will be the case sooner or later.

Anyway, you can turn water recycling ON and remove all waters after the 
map is computed and before you go into manual model building in Coot or 
whatever program you use.


More information about the phenixbb mailing list