Re: NAMD support for lone pairs

From: Gumbart, JC (gumbart_at_physics.gatech.edu)
Date: Tue Mar 12 2019 - 08:35:03 CDT

Are you sure that it’s not an error in CGenFF or some other oversight? I have trouble imagining them putting a free lone pair in there. That CHARMM doesn’t crash probably just means no one thought to check it. I would follow up with Alex et al. before writing a lot of code to support it.

Best,
JC

On Mar 12, 2019, at 8:50 AM, Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:

Tabitha,

Thanks for updating us with the results
of those tests! Apparently my previous guess regarding atom order was incorrect (though I will still contend this is a good idea and best practice).

To summarize - NAMD does not appear to like PSFs that do not contain an explicit bond to lonepairs, but CHARMM accepts them just fine. The canonical force fields appear to always include those bonds, but CGenFF may not when adding a lonepairs on chlorine (any halogen?). The developers will have to discuss how best to address this odd edge case.

Thanks for working through this with us,

Brian

On Mon, Mar 11, 2019, 9:01 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
Adding the bond indices and incrementing the NBOND count by one worked! I tried it also on the PSF/PDB pair that did not have the lone pair close to its parent, and that worked as well. The results of the minimization look reasonable.

On Mon, Mar 11, 2019 at 5:27 PM Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:
That's odd. The BOND command in the Residue definition should cause the PSF to have a bond entry - regardless the ordering.

You can try adding a manual bond by incrementing the ! NBOND count by one and adding two extra bond indices to the list (note that the PSF indices start at one). This is not an ideal solution, but hopefully it will work.

Can you send me your initial pdb offline so that I can play with this? We are looking to support this kind of construction in psfgen and this kind of lonepair support might be one of the first things we can get going.


On Mon, Mar 11, 2019, 7:51 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
Hi Brian,

I tried renumbering the atoms in my PSF and PDB file as you suggested. I also added the BOND statement in my stream file.
However, NAMD is complaining that the lone pair is an isolated particle. My edited files are attached in this email.
Do you other ideas as to what might be the problem?

Best,
Tabitha

On Mon, Mar 11, 2019 at 4:10 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
My input PDB to parameterize with CHARMM did not include the lone pair. The lone pair information was in the stream file, but I had to explicitly list it in the input file for CHARMM. So I would have to renumber the resulting PSF/PDB atoms. Do you know of a way I can do this automatically and not by hand?

On Mon, Mar 11, 2019 at 2:39 PM Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:
Ah, I think I see the problem now that I look at your PSF more closely,

NAMD, I believe, requires lonepairs to be close to their parents (right after, unless there is also a Drude particle) in the PSF, or at least this is a convention of the CHARMM force fields. You similarly have all of your hydrogens at the end of the residue rather than near their associated heavy atoms.

Was this molecule parameterized via CHARMM-GUI also? I've heard of ParamChem being sensitive to atom ordering and connectivity, so it might be worthwhile for you to reorganize the atoms in your input PDB. Without using mol2 format I don't know how you would get bonded information - maybe proximity in ordering is used heuristically in this case?

On Mon, Mar 11, 2019 at 5:30 PM Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:
I would probably not recommend trying to manually alter a PSF (though it is possible).

I'm just confused by this behavior from CHARMM-GUI. It would be nice to know if their output can be run with CHARMM natively, which could indicate an overly strict response from NAMD (and maybe require a change).

On Mon, Mar 11, 2019 at 5:02 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
Hi Brian,

Would the parent be defined in the bonds section of the PSF? Is this something I can manually add?

I haven't run it in CHARMM, no, but if you think it's helpful I can try to get that going. Ultimately I would want to be using NAMD though.

-Tabitha

On Mon, Mar 11, 2019 at 11:41 AM Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:
Hi Tabitha,

NAMD seems to interpret unbonded lonepairs as an error - there needs to be one (and only one) entry in the PSF indicating the parent for that lonepair (not just in the NUMLP block). I don't know if CHARMM includes this by default or not.

I assume your PSF runs just fine in CHARMM?


On Fri, Mar 8, 2019 at 6:46 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
Yes, I actually used CHARMM to build the PSF. When I tried the resulting files in a NAMD minimization, I got the error "FATAL ERROR: FOUND ISOLATED LONE PAIR PARTICLE 24".
Do I need to specify anything in the configuration file about the lone pair? You mentioned the "lonepair on" keyword is deprecated, and I didn't see mention of anything else in the current user guide.

I attached my stream file from CGenFF, the PSF/PDB files generated from CHARMM, and my NAMD input and output files.
Is there other information I can provide that would shed more light on this problem?

Thank you!

On Fri, Mar 8, 2019 at 12:14 PM Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:
This should be supported now for a valid PSF (did you try it and have it not work?).

The current lack of functionality is in the RTF parser inside psfgen, so that you cannot yet build these PSFs. I would recommend CHARMM-GUI for that purpose, although there are plans to support this in the future.

On Fri, Mar 8, 2019, 12:10 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
Hi Brian,

Thank you very much for your response. This might be a naive question, but if I have a chloro group with a lone pair generated from CGenFF, is there a way I can convert that to a currently supported format from NAMD?

Thank you,
Tabitha

On Fri, Mar 8, 2019 at 7:45 AM Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:
Hi Tabitha,

As of 2.13 there should be full support for all lonepair types that exist in currently published/released CHARMM force fields, both fixed charge and Drude. This includes the "bisector" lonepairs on water and carbonyls as well as the "collinear" lonepairs most associated with halogens (e.g. chlorobenzene). Furthermore, the old "lonepair on" keyword should be essentially deprecated and instead auto-detected based on the PSF contents (it should be very obvious if this does not work). However, NAMD does not yet support all CHARMM compatible lonepairs because they are not defined in the PSF standard.

Cheers,
BKR

On Thu, Mar 7, 2019 at 9:27 PM Tabitha Nobel <hydrargyrumate_at_gmail.com<mailto:hydrargyrumate_at_gmail.com>> wrote:
Dear NAMD developers,

Are CGenFF lone pairs supported in NAMD version 2.13? This question is similar to the one asked here<https://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l.2018-2019/0992.html>, but I saw in the release notes<https://www.ks.uiuc.edu/Research/namd/2.13/features.html> that there was improved lone pair support, and it wasn't clear to me whether that was only for the Drude force field.

Thank you,
Tabitha

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:20:36 CST