Re: [COOT] Wild refinement of waters
On Thu, 2021-06-03 at 09:02 -0700, Philip Kiser wrote: > Thank you for spotting the problem and introducing the new restraint. I have > to say that it was easy to get in the > habit of creating poorly placed water molecules when filling difference map > peaks as the prior refinement method for > isolated waters would just move them into the nearby 2Fo-Fc peak. You can do that - that's reasonable - but it makes more sense (to me) to allow the protein to move in response to a badly placed water - and that stabilizes the refinement (in my hands). > Any residual placement errors or clashes would get fixed during reciprocal > space refinement and all was good. Yes, that's the idea with Coot's RSR. > But I will take your advice and change my ways. Well, refine with sphere refine rather than just one water and you should be good. And if not, the soft restraints will do the trick. Paul. To unsubscribe from the COOT list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1 This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
Re: [COOT] Wild refinement of waters
Hi Paul, Thank you for spotting the problem and introducing the new restraint. I have to say that it was easy to get in the habit of creating poorly placed water molecules when filling difference map peaks as the prior refinement method for isolated waters would just move them into the nearby 2Fo-Fc peak. Any residual placement errors or clashes would get fixed during reciprocal space refinement and all was good. But I will take your advice and change my ways. Much appreciated, Philip On Thu, Jun 3, 2021 at 3:12 AM Paul Emsley wrote: > On Wed, 2021-06-02 at 15:56 +0100, Paul Emsley wrote: > > On Tue, 2021-06-01 at 20:45 +0100, David M Dranow wrote: > > > Does anyone have a decent set of refinement parameters for Coot 0.9.5 > in linux? My colleagues and I find that using > > > the "r" shortcut causes most waters to fly off and center on carbons, > sometimes as far away as 45 Å! I've tried to > > > change the overall weight and Lennard-Jones epsilon, and > Geman-McClure alpha and I find no combination that keeps > > > all > > > waters within the density that they already are in. I'd be happy to > provide a model and map to demonstrate this > > > phenomenon. Our current work-around is to use R, but this is not > ideal. If anyone has any idea how to fix this, we > > > would greatly appreciate any advice. > > > > > > > This reminds me of the exchange about the new refinement on CCP4BB some > time ago. Briefly, people reported problems > > that > > I wasn't seeing. > > > > So I'd like to see what you see. Either make a screencast of your > session or conjure up a coordinates file and a list > > of coot operations that makes the issue apparent. Until then, I am in > the dark. > > > > > > Thanks to files provided by Philip Kiser I now see the problem. > > First, one had to create an unorthodox model - one that included a bad > non-bonded contact distance to a water > (this is not something I do) > Second, one had to refine just the water, and not the residues to which it > was hydrogen-bonded or other atoms in > the environment > (this is not something I do either) > > By its nature, refinement of this sort will give a considerable kick to a > bad water. > I have therefore introduced an optional "Soft Mode" for non-bonded-contact > restraints. > set_refinement_use_soft_mode_nbc_restraints(flag) > > This is available now in the repo and will be in the next release. > > But for me, the lessons are: Don't create models with poorly placed waters > in the first place. But if you do, don't do > isolated refinement of the waters one by one. > > Paul. > > > > To unsubscribe from the COOT list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1 > > This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing > list hosted by www.jiscmail.ac.uk, terms & conditions are available at > https://www.jiscmail.ac.uk/policyandsecurity/ > To unsubscribe from the COOT list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1 This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
Re: [COOT] Wild refinement of waters
On Wed, 2021-06-02 at 15:56 +0100, Paul Emsley wrote: > On Tue, 2021-06-01 at 20:45 +0100, David M Dranow wrote: > > Does anyone have a decent set of refinement parameters for Coot 0.9.5 in > > linux? My colleagues and I find that using > > the "r" shortcut causes most waters to fly off and center on carbons, > > sometimes as far away as 45 Å! I've tried to > > change the overall weight and Lennard-Jones epsilon, and Geman-McClure > > alpha and I find no combination that keeps > > all > > waters within the density that they already are in. I'd be happy to > > provide a model and map to demonstrate this > > phenomenon. Our current work-around is to use R, but this is not ideal. > > If anyone has any idea how to fix this, we > > would greatly appreciate any advice. > > > > This reminds me of the exchange about the new refinement on CCP4BB some time > ago. Briefly, people reported problems > that > I wasn't seeing. > > So I'd like to see what you see. Either make a screencast of your session or > conjure up a coordinates file and a list > of coot operations that makes the issue apparent. Until then, I am in the > dark. > > Thanks to files provided by Philip Kiser I now see the problem. First, one had to create an unorthodox model - one that included a bad non-bonded contact distance to a water (this is not something I do) Second, one had to refine just the water, and not the residues to which it was hydrogen-bonded or other atoms in the environment (this is not something I do either) By its nature, refinement of this sort will give a considerable kick to a bad water. I have therefore introduced an optional "Soft Mode" for non-bonded-contact restraints. set_refinement_use_soft_mode_nbc_restraints(flag) This is available now in the repo and will be in the next release. But for me, the lessons are: Don't create models with poorly placed waters in the first place. But if you do, don't do isolated refinement of the waters one by one. Paul. To unsubscribe from the COOT list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=COOT=1 This message was issued to members of www.jiscmail.ac.uk/COOT, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/