VMD-L Mailing List

From: JC Gumbart (gumbart_at_physics.gatech.edu)
Date: Sun Jan 08 2017 - 23:08:15 CST

Indeed, it is subjective but 1.0 sounds really good. There is no guarantee
that subsequent rounds will monotonically improve.

Best,
JC

On Sun, Jan 8, 2017, 12:24 PM #YEO JINGJIE# <JYEO1_at_e.ntu.edu.sg> wrote:

> Thank you very much for the information, I will test that out and see how
> it goes.
>
> I have another query regarding bond and angle fitting. I obtained an
> objective value close to 1.0 for the first downhill iteration, but
> subsequent iterations drove it up to approximately 3.0 and 9.0. Is this
> erroneous or is the first iteration already sufficient? Is a value close to
> 1.0 considered low (although I assume this would be rather subjective)?
>
> Best,
> JJ
>
> On 7 Jan 2017, at 5:36 PM, JC Gumbart <gumbart_at_physics.gatech.edu> wrote:
>
> C8 has a penalty of 0.00, why are you fitting it?
>
> You need to balance atoms that are including with gaussian log files
> included. For example, if C8 is not being fit, don’t include the gaussian
> file for its interaction with a water. Ultimately, it takes some trial and
> error with atoms and water-interaction data to include, along with multiple
> rounds of fitting, to figure out the right balance.
>
> Best,
> JC
>
> On Jan 6, 2017, at 4:47 PM, #YEO JINGJIE# <JYEO1_at_e.ntu.edu.sg> wrote:
>
> Hi JC,
>
> I noticed that in one of the Gaussian water interaction files, one
> hydrogen was visualised by VMD to be bonded to the diazosulfonic acid
> carbons. If I include all the Gaussian files for optimisation and following
> your advice to exclude atoms with low penalties, I would get some funky
> numbers:
> C8 -0.950
> N2 -0.971
> C13 0.779
> N3 1.142
>
> However, even if I omit that anomalous Gaussian file, I get:
> C8 -0.195
> N2 0.402
> C13 -0.592
> N3 0.385
>
> Any ideas what might have gone wrong?
>
> Best,
> JJ
>
> On 6 Jan 2017, at 3:52 PM, JC Gumbart <gumbart_at_physics.gatech.edu> wrote:
>
> You just don’t even include those atoms with low penalties (< 10 at least,
> maybe higher) in the charge fitting. Their charges will be taken from the
> PSF file.
>
> Best,
> JC
>
> On Jan 6, 2017, at 1:52 PM, #YEO JINGJIE# <JYEO1_at_e.ntu.edu.sg> wrote:
>
> Dear VMD users,
>
> I would like to use FFTK to determine the CHARMM parameters for
> diazonium-modified tyrosine. I am setting up the problem using a fragment
> approach, where I have the tyrosine residue and diazosulfonic-acid
> fragment. The .mol2 and .str files from paramchem are available here:
>
> https://www.charmm.org/ubbthreads/ubbthreads.php?ubb=showflat&Number=36055#Post36055
>
> I've obtained the water interaction energies from Gaussian for the entire
> modified residue as well as only for the diazosulfonic-acid fragment alone.
> The question now is from which of this should I run the charge
> optimisations and which atoms ought to be constrained? From the paramchem
> results, the highest penalties are in the region around the -N=N- atoms--001a113dfc7685ea4f0545a25bfb--