[phenixbb] rotamers

Pavel Afonine pafonine at lbl.gov
Mon Mar 28 08:20:11 PDT 2011

  Dear Ed,

nice survey page! I have a few comments / questions:

- what you mean by "no density", I mean in sigmas? zero? 0.5? multiple 
apparently alternative locations with 0.7 sigma? or..? Most of the time 
it is not that the density is zero or negative due to artifacts, but 
"weak" (I used "" because that needs definition, and probably another 
survey -:)). Therefore this raises another item for your questionnaire: 
model each such weak density state with ensemble of alternative 
conformers (each corresponding to plausible rotamer) and refine group 
occupancy for these atoms (one occupancy per all atoms in question - the 
occupancy typically will refine to something less than 0.5 or so). The 
sum of occupancies may not necessarily add up to 1 because you will 
still be missing a few conformers with truly no density.

- Item #3: I guess if a program lets you to do this during restrained 
refinement of individual isotropic (or anisotropic) B-factors, then you 
should submit a bug report. All the flavors of restraints used in 
individual ADP refinement are essentially similarity restraints that 
make sure the B-factors of connected atoms are more or less similar. 
This trick with smearing out an atom by B-factor may only work for 
isolated (single) atoms  such as waters because they are not bonded to 
anything through restraints.

All the best!

On 3/28/11 7:25 AM, Ed Pozharski wrote:
> On Sat, 2011-03-26 at 10:06 +0000, Antony Oliver wrote:
>> I keenly await the (hopefully) consensus opinion on what we, as
>> crystallographers, should do with residues with poor or absent
>> side-chain
>> density.
> Antony,
> thanks for passing along the information.  I won't hold my breath for
> the consensus though - this is one of the recurring discussions (myself
> being guilty of repeatedly stoking the fire) that never result in clear
> conclusion, simply because the choice here has zero effect regarding
> conclusions drawn from your data.  It is simply a question of semantics.
> To get some idea if there is any indication of consensus, I've set up a
> survey at google docs
> https://spreadsheets.google.com/viewform?hl=en&formkey=dHVNa3VodUtfbVQtZ2pnUFcxQkx6RHc6MQ#gid=0
> I'll post the results once it levels off - please don't vote more than
> once (but feel free to get your lab/family members to have a say).
> Cheers,
> Ed.

More information about the phenixbb mailing list