Re: [ccp4bb] Refining alt. confs for only part of a ligand

[Derek Logan]

Hi Nick, Pavel and Herman,

Thanks for the lightning-fast tip about not using an altconf label for the 
common part of the molecule. That was the solution. So obvious when you think 
about it... My colleague Esko Oksanen also pointed out offline that I need to 
include the last atom before the phenyl ring with two conformations, otherwise 
the torsion angle wouldn't be properly defined. Indeed, without including this 
atom it blew up when idealising geometry in Coot, but with that minor 
modification it behaved just fine in both phenix.refine and Refmac.

@Pavel, thanks but it wasn't a visualisation glitch, as the rings were clearly 
displaced from each other. (BTW my message got bounced from phenixbb because of 
the screen dump I included).

/Derek


On 27 Oct 2017, at 13:47, Nick Pearce 
> wrote:

Conformer A atoms don’t “see” conformer B atoms so as far as the program is 
concerned the B conformer only has the last ring.
You should set all confA atoms not in the last ring to blank conformers.

So:
last ring - A +B conformers
rest of ligand - no conformer

Thanks,
Nick

—

Nick (Nicholas) Pearce
Post-doctoral Researcher
Lab of Piet Gros
Crystal & Structural Chemistry Group
Universiteit Utrecht

On 27 Oct 2017, at 13:41, Derek Logan 
> wrote:

Hi,

This is a cross-post to ccp4bb and phenixbb. I'm trying to refine a largish 
small molecule ligand. It has two conformations that differ only in the 
orientation of a terminal aromatic ring, i.e. variation in the last torsion 
angle. I previously refined it as two independent conformations for the whole 
molecule, but that showed too much deviation between the conformations in the 
region that is "identical", i.e. that can be modelled just as nicely with a 
single conformation. With Occam's razor I split the molecule at the last 
torsion angle. Conformation A is the whole molecule; conformation B is just the 
last ring. Both have the same residue number, just different altloc labels. 
However when I refine this using phenix.refine, the B conformation refines as 
if it were not attached to the A conformation (see the attached screen dump). I 
tried it using Refmac and the two conformations separated even more. Clearly 
this didn't work as I was expecting. What am I doing wrong? Is it even possible 
to do it this way?

/Derek

Re: [ccp4bb] Refining alt. confs for only part of a ligand

[Nick Pearce]

Conformer A atoms don’t “see” conformer B atoms so as far as the program is 
concerned the B conformer only has the last ring. 
You should set all confA atoms not in the last ring to blank conformers.

So: 
last ring - A +B conformers
rest of ligand - no conformer

Thanks,
Nick

—

Nick (Nicholas) Pearce
Post-doctoral Researcher
Lab of Piet Gros
Crystal & Structural Chemistry Group
Universiteit Utrecht

> On 27 Oct 2017, at 13:41, Derek Logan  wrote:
> 
> Hi,
> 
> This is a cross-post to ccp4bb and phenixbb. I'm trying to refine a largish 
> small molecule ligand. It has two conformations that differ only in the 
> orientation of a terminal aromatic ring, i.e. variation in the last torsion 
> angle. I previously refined it as two independent conformations for the whole 
> molecule, but that showed too much deviation between the conformations in the 
> region that is "identical", i.e. that can be modelled just as nicely with a 
> single conformation. With Occam's razor I split the molecule at the last 
> torsion angle. Conformation A is the whole molecule; conformation B is just 
> the last ring. Both have the same residue number, just different altloc 
> labels. However when I refine this using phenix.refine, the B conformation 
> refines as if it were not attached to the A conformation (see the attached 
> screen dump). I tried it using Refmac and the two conformations separated 
> even more. Clearly this didn't work as I was expecting. What am I doing 
> wrong? Is it even possible to do it this way?
> 
> /Derek
> 
>