Hi Ryan,

Thanks for the explanation. useDynamics=1 is the default in 
regularize.fixupCovalentGeometry, so if you didn't touch it, you are already 
using it.

If you did the protonation within Xplor, there's a problem I can think of. The 
side chain of asparagines and glutamines might have their O- and N-atoms 
incorrectly assigned (swapped) in the X-ray structures. Putting protons on an N 
that is actually an O with no room for them around it will result in clashes.

So, before moving on, I recommend assessing the clashes of the input structures 
independently using the MolProbity server 
(http://molprobity.biochem.duke.edu/). It will add protons and try to figure 
out the correct assignment for Asn and Gln side chains, "flipping" them if 
necessary. You'll also get a detailed report of the clashes.

Best,

Guillermo
________________________________________
From: Ryan Hoffman [[email protected]]
Sent: Thursday, August 27, 2015 2:53 PM
To: Bermejo, Guillermo (NIH/CIT) [E]
Cc: [email protected]
Subject: Re: [Xplor-nih] high VDW question (and mailing list is offline?)

I've cobbled-together an initial model by joining some separate domains. 
They're all crystal structures on their own, so I think the high energies come 
from protonating the sidechains, mostly. There aren't any obvious occluded 
regions. I'm just trying to get minimized initial coordinates. I'm using 
version 2.39, by the way.

I'm trying to avoid defining a lot of artificial distance restraints and 
minimizing against them, which I expect would work.

"Note that regularize.fixupCovalentGeometry does not enforce VDW interactions 
unless you set the useVDW argument to True."

Yup, I've set useVDW=1. Can you comment on whether I should also set 
useDynamics=1?

By inputting the heavy atoms and protonating them, the energies get very high 
(with no further dynamics.) The regularize.fixupCovalentGeometry greatly 
decreases these energies but it seems the VDW term remains high. The main chain 
doesn't seem to move at all and the side chains only very slightly.

Here's an outputted PDB header:

REMARK summary    Name       Energy      RMS     Violations
REMARK summary total         2202.19                  31964.0
REMARK summary ANGL           792.98    0.375      0.0
REMARK summary BOND            11.27    0.001      0.0
REMARK summary IMPR            96.82    0.244      0.0
REMARK summary RAMA           -17.90
REMARK summary VDW           1319.02           31964.0

Thanks for mentioning torsionDBPot. I'll take a closer look at 
eginput/gb1_rdc/refine.py to see how it compares with my script.

Thanks,

Ryan




On Thu, Aug 27, 2015 at 10:51 AM, Bermejo, Guillermo (NIH/CIT) [E] 
<[email protected]<mailto:[email protected]>> wrote:
Hi Ryan,

Could you describe a bit what you are trying to do?

Note that regularize.fixupCovalentGeometry does not enforce VDW interactions 
unless you set the useVDW argument to True.

Also note that the replacement of RAMA, torsionDBPot, yields slightly fewer 
clashes. For an example of its setup see eginput/gb1_rdc/refine.py within your 
Xplor-NIH directory.

Best,

Guillermo


________________________________________
From: Ryan Hoffman [[email protected]<mailto:[email protected]>]
Sent: Thursday, August 27, 2015 12:49 PM
To: [email protected]<mailto:[email protected]>
Subject: [Xplor-nih] high VDW question (and mailing list is offline?)

Hi there!

https://dcb.cit.nih.gov/pipermail/xplor-nih/
is apparently offline?

My question's fairly simple...what is the current state of the art for relaxing 
VDW violations? The rest of the bonded parameters are good, but there are 
apparently many, many VDW violations.

I have no experimental restraints at this point although I am using the RAMA 
potential.

I've been trying regularize.fixupCovalentGeometry, as well as deleting the 
hydrogens and rebuilding them, and using Powell minimization following brief 
(and low-T) dynamics. Especially the former removes all of the violations but 
they seem to accumulate in the VDW term.

Thanks,

Ryan


_______________________________________________
Xplor-nih mailing list
[email protected]
https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih

Reply via email to