[phenixbb] ordered solvent removing metal ion coordination

Pavel Afonine pafonine at lbl.gov
Mon Nov 24 16:31:34 PST 2008

Hi Engin,

this is for your first question and I hope someone knows a better 
solution for the second one:

First of all, sorry for the problem. As it is now, if 
"ordered_solvent=true" is used phenix.refine treats water the way that 
it can be removed any time if it does not satisfy the selection 
criteria; or/and the water can be added as well. There is no way to fix 
particular waters (unless you re-name them somehow like WT1, WT2, ... ). 
The idea behind this seeming inflexibility is that the selection 
criteria are for the purpose to not let a wrong water be in your 
structure. If you believe the selection criteria are too strict you may 
loosen them till they keep your favorite water molecules.

I now realize that it's possible that one may want to keep certain 
waters even though they are beyond the selection criteria and lowering 
the criteria (to keep those waters) may add some spurious "waters". On 
the other hand, fixing selected waters will make it a user's 
responsibility to monitor their quality (which is dangerous in general 
and is against the idea of automation).

Anyway, I will provide a functionality to keep selected waters in future 
(this will require some code modifications) (it's on my list already but 
I'm not sure when I get to it); although a long-term goal is to make 
water updating procedure smarter so there will be no need to manually 
fix anything.

As for the second question (I don't really know the best solution..., 
just something that would practically work...), I would manually create 
a custom CIF file and give a name to that complex (so the waters that 
make it are not treated as real waters anymore). So basically this is 
what you did already...

Please let know if you have any other questions or problems!


On 11/24/2008 3:16 PM, Engin Ozkan wrote:
> Hi,
> Is there a way to shut off solvent checking for a few waters?  I have 
> what looks like sodium-coordinating waters that get taken out by ordered 
> solvent filtering, even as I am loosely restraining them to be at 
> distances expected for sodium. These waters -I have 4 NCS-related 
> copies- are taken out by filtering, which messes up water numbering, and 
> makes my restraint descriptions irrelevant.  I have tried putting them 
> in a separate chain, but they get fused to the solvent chain by phenix.  
> My other option is to describe those waters as a different type of atom 
> (not HOH), but they are water molecules, so I am reluctant. And I want 
> to keep ordered_solvent True.
> For cif files already available in the default ccp4 library, NA seems to 
> be elemental sodium, NAW, NA5 and NA6 seem to be specific sodium+water 
> complexes with three, five and six waters coordinating (where they are 
> no longer HOH, obviously). So, I have created bond/angle restraints 
> specifically to my atoms, but is there an easy way to do this? Or should 
> I just create a two-water NA (NA2.cif) specific for my case and call 
> waters as some O atom? That's easy, but I wonder if that's the common 
> practice (or I am missing something completely).
> Any ideas?
> Thanks
> Engin
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb

More information about the phenixbb mailing list