*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear friends,
Here I present a compilation of the discussion concerning my previous
post about distance between waters in refmac (at the end there goes the
whole e-mail).
Soheila Vaezeslami (SV), Bart Hazes (BH) and Eleanor Dodson (ED)
commented.
SV normally models these waters as alternate positions of the same water
molecules (a peanut shape for the electron density).
ED remembers many waters should have partial occupancy and when they are
close to each other they often have high B-factors. She remembers also the
care not to refine waters in protein alternate conformer positions (which
she says there might be ~ 10 % in each structure - I would like to see a
work on this, specially relating to the resolution and - I believe - solvent
exposed area). My new comments on this: i) someone pointed in this bb last
year - 2005 - that in many cases it is more appropriate to refine
occupancies rather than b-factors for water molecules; ii) in most of the
cases one or the two water molecules have a high b-factor, but in some few
cases both are reasonable and do not seem to be related to map noise.
BH remembered a recent suggestion about having O2 and N2 molecules in
the structures. I would say in fact that some of the 2 seed peanut like maps
might look so. At 1.6 A, hard to prove crystallographically.
Nevertheless, one question remains, whether there is a way to increase
the refmac weight on this type of interaction. I woud also be happy to know
of an article where the subject is treated (distance between waters in a
crystallographic structure), specially the average and standard deviation
...
Thanks to all,
Jorge
----- Original Message -----
From: "Jorge Iulek" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Wednesday, December 21, 2005 10:17 AM
Subject: [ccp4bb]: refmac5 distance between water molecules
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear all,
I am refining a structure (almost 1100 residues in the au) at 1.6 A
with refmac5. I could model a number of water molecules. At RCSB
validation, I see several close contacts (in the same asymmetric unit)
between water molecules (both at 1.00 occupancy). To investigate further,
I used at refmac5:
MONITOR MANY VDWR 1.5
In fact this lists several "bad" contacts including those reported by
RCSB validation. Also, at refmac5 output, I see that 2.740 is considered
to be the ideal distance between O's from water molecules, but it reports
the several negative deviations (and several distance values below 2.2,
2.1 and even 2.0 A). I went to inspect maps and the model in coot.
Interestingly, coot at its "check waters" tool does not list several of
these "water close contacts" even when I ask it to list the ones lower
than 2.2 A (does this distance work there also for between water molecules
? I could not find this at RTFM), but at measuring the distances there
they are (even in the same au). In coot, I could see a number of cases
where there is room for the molecule to go to "somewhere else" to avoid
the bump at 2.2 A (distance used by RCSB).
I thought this might be due to the fact that I am using higher weights
(around 0.9) because of the resolution (and this gives me very good rms
for bond angles and lengths). I lowered the weight to 0.3 and the problem
persists (although a little less serious). I also made a test, put the
same model into CNS and asked for only the water molecules to be refined.
Then, no close contacts between water molecules at all !
I had never seen this problem (this is the first time I have a
structure at 1.6 A).
My questions are:
a) How does refmac5 treat the distances between water molecules ? What is
the weight imposed to this ? Has this something to do with the fact that
the structure is 1.6 resolution and the weight is high ? If this is the
case, how to use a high general weight (which gives better statistics at
rms's and R factors) and not to have this problem with water molecules ?
Has this something to do with the fact that I have a good number of waters
(also because of the resolution) ?
b) How to overcome this ? What I know is that arp might deal with this in
real space, but there might still remain the fact thhat the new arp
waters is further refined in refmac5 to distances lower than 2.2.
Thanks,
Jorge
