*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
Hi, There is an error estimate for every parameter of refined model. However, for B and O (or N, whatever you want to call occupancy) the parameters are closely related - therefore the refinement of occupancy is difficult at low data/parameter ratios - such as encountered at low resolution :) Realistically, if you see enough of the minor component of the disorder, even at such a low resolution, then you can and should model it, but I personally wouldn't bother too much with trying to establish 'accurate' occupancy at this resolution. In my experience, a guess is about as accurate as trying to match B-factors. You can try SHELX - it can refine independent occupancy parameters. At 2.5 A I would use one B-factor per residue and restrain a whole lot, to make sure that the 'anatomy' of the structure is realistic. So in the case in question, there will be one B-factor parameter per residue, and then for the few residues that are disordered there will be an additional parameter - the occupancy (N and 1-N for the two components of each pair, so still only one extra). There shouldnt be an extra B-factor parameter since SHELX can link B-factors of the disordered pairs. This way you won't be adding too many parameters (you still have all the extra coordinates to refine, though!). Free variables FTW*! Artem ex-PGP * - FTW is gamer slang acronym 'For The Win' - meaning 'great'. -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of nic steussy Sent: Friday, March 17, 2006 5:43 AM To: [EMAIL PROTECTED] Subject: Re: [ccp4bb]: Refmac ligand occupancy refinement *** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk *** All, How reasonable is it to estimate the occupancy of a ligand by assuming it has the same B factors as the adjacent side chain atoms? I don't mean to pick on Ekstrom, as I have seen this assumption asserted in a number of structure papers. I was a bit skeptical, especially at modest (mid 2.0A) resolution. What do you all think? Nic out ============================ C. Nicklaus Steussy, M.D., Ph.D. Purdue University [EMAIL PROTECTED] ============================ Ekström Fredrik wrote: > > > I'm struggling with a 2.5 Å structure of a protein-ligand complex. > Current R/Rfree is 19/24. > > The occupancy of the ligand is approximately 0.80 (estimated by B-factor > matching of neighbouring residues) >
