[buster-discuss] Water Picking

[Clemens Vonrhein]

Dear Dale,

On Tue, Nov 16, 2010 at 12:12:08PM -0800, Dale Tronrud wrote:
>    I'm having a real disagreement with your water picker.  It likes
> to build water molecules way out in open space.

Yes, the default picker isn't very clever as a water picker - partly
by design: we're using it also to place dummy atoms to describe
density e.g. for unmodelled ligands (or loops, domains etc). In that
regime it is pretty effective and can help getting density that is
easier to interpret for yet unmodelled parts. But the resulting
'water' atoms need to be taken with a pinch of salt, I agree.

> Do you have a filter program that I could use to automatically get
> rid of this junk?

We haven't got anything automatic for that at the moment (but see
below).

> How do people deal usually deal with this problem?

I tend to use the water checking in Coot after a round of automatic
water update: deleting all waters with 2Fo-Fc density below
1.0 and those with distances that are too long/short.

After that I wouldn't use any automatic water update anymore (no
matter what method/program is used). I'd rather cycle through the
difference peaks that eg. Coot finds and add sensible waters by hand
after each refinement round.

In my experience, any automatic water picking program gets somewhere
into the 80-90% range of having all waters correct. The last checks
and additions need to be done by hand anyway it seems.

>    I tried to override your picker by adding the line (which old
> notes of mine indicated was a possibility)
> 
> WaterUpdate_program=ARP_WATERS
> 
> to the "refine" command but it didn't seem to make any difference.

The command line is

  WaterUpdateProgram="ARP_WATERS"

see

  http://www.globalphasing.com/buster/manual/autobuster/manual/autoBUSTER6.html#WaterUpdateProgram

... so just a little typo ;-)

However, this requires the 'arp_waters' binary being available. Recent
version of CCP4 have moved this into the unsupported section and don't
compile it by default.

Also, in the same appendix1.html there is a link to an example script
for writing your own water update tool: if you have your own set of
peak-picking and checking tools that can be incorporated into such a
script.

Cheers

Clemens

PS: the next release will include two more options for water picking -
    a hook to Coot's findwater program and a an option to turn on a
    distance criteria to potential hydrogen donors. The all give
    slightly different results, some add a few too many, others not
    enough ...

-- 

***************************************************************
* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
*
*  Global Phasing Ltd.
*  Sheraton House, Castle Park 
*  Cambridge CB3 0AX, UK
*--------------------------------------------------------------
* BUSTER Development Group      (http://www.globalphasing.com)
***************************************************************