I cant comment on the occupancy refinement of overlapping ligands; this
is always difficult, and I must say as a reader I am often sceptical of
the claims in papers - 3A data and two conformations???
However - if you label the alternatives according to the PDB conventions
C1 A C2 A nres etc for conformaer 1 then C1 B C2 B with the same nres
for the next, and mae sure the sum of the occupancies does not exceed
1.0 then REFMAC should not push the coordinates apart - the Xray
gradients wont know what to move where, and cant help refinement much,
but the geometric restraints should hold the two conformers together.
Eleanor
Jorge Iulek wrote:
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear ZhiYi,
It seems that this subject comes times to times to either the
ccp4bb or straight to me, yet I raised this question some years ago. I
could count several people with the same situation; that makes me
wonder how frequently this happens and, above all, how frequently
people do take care to make the job the best possible.
Going to your question: as it was raised at that time and as I
could notice (with the program version at that time), there is an
issue with refmac such that it increases the B-factor of one of the
molecules and pushes it ouside of the density, even if their
occupancies sum <=1. The solution I found was to use CNS with the
capabilities to turn on and off the interactions. That allowed me to
refine well my molecules, sharing positions (occ sum = 1). On the
other hand, refmac has the TLS tool, im my humble opinion, a MUST for
anyone who cares on providing the best refined possible model (at
least to try it, and now I must say I will try the TLSMD server in
every structure I have in my hands) (this a question, "MUST's" I am
yet for some time to elaborate and propose to the ccp4bb...).
So, I made a huge labour on swapping between CNS (to refine the
overlapping ligands) and refmac (to use TLS, important for that
structure); although this labour, I feel myself confortable as to have
made the most possible to achieve the best final model. At describing
this process in the paper, one referee wrote "...akward description of
the refinement method..." and I hope that was due to my inabilities
with the redaction on the english language, yet concerning methodoly
itself I cannot think of any other better way. If there is so, please,
educate me.
If you find difficulties to prepare the CNS files for turning
on-off interactions, let me know and I can try to dig this from my
archives, as a template for you.
Concerning occupancies, there is a crude general recommendation to
attribute values such that B-factors end up approx. the same, although
there might be distinct cases to analyse. I think other people can add
more useful comments on this.
Jorge
----- Original Message ----- From: "ZhiYi Wei"
<[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Thursday, September 15, 2005 12:11 PM
Subject: [ccp4bb]: Questions on ligand refinement
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Hi, all
How to refine two different ligands in the same position and figure
out their
occupancy? Which program will be better to deal with this situation,
CNS or
refmac? I know that some similar questions have been posted in bb,
but I can not
find any answers in detail.
Any answers will be appreciated.
begin:vcard
fn:Eleanor Dodson
n:Dodson;Eleanor
email;internet:[EMAIL PROTECTED]
tel;work:+44 (0) 1904 328259
tel;fax:+44 (0) 1904 328266
tel;home:+44 (0) 1904 424449
version:2.1
end:vcard
