From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Tue Oct 30 2018 - 09:16:39 CDT
>
> Both paramchem (faster) and FFTK (slower) will automatically identify the
> missing parameters.
>
However, after patching for a covalent bond between ligand and protein, it
is only by running a minimization on .psf/.pdb, obtained from psfgen, that
tells you about missing parameters for the new situation. Am I wrong, or is
anything more efficient?
cheers
fp
On Mon, Oct 29, 2018 at 7:07 PM Gumbart, James C <gumbart_at_physics.gatech.edu>
wrote:
> So autogenerate just means that all possible angles and dihedrals given
> the topology of the molecule are added to the PSF. Each of these will
> require a corresponding parameter. If you are being asked for a parameter
> that doesn’t involve at least one of the changed atoms, then something
> might be wrong. But consider even just changing one atom type has the
> potential to require dozens of new angle and dihedral parameters if it
> hasn’t been used previously in that context in the force field.
>
> Both paramchem (faster) and FFTK (slower) will automatically identify the
> missing parameters.
>
> Best,
> JC
>
> On Oct 29, 2018, at 3:07 AM, Francesco Pietra <chiendarret_at_gmail.com>
> wrote:
>
> I was surprised that parameters were also requested for bonds rather
> removed from NE2-C9, such as between C10 and its adjacent C11. Besides
> that, requests seemed to be endless, for situations with non-obvious
> analogies. So that, with a natural-product ligand, covalently bonded to a
> protein receptor, running a decent MD may take months of efforts with QM
> calculations.
> When that arises as a small part of a large project, involving many
> ligand-protein cases, one (myself) becomes frustrated. Unless one decides
> that each case is a publication, which, unfortunately, is a decision by
> many.
> Thanks for your view
> francesco
>
> On Mon, Oct 29, 2018 at 12:09 AM Vermaas, Joshua <Joshua.Vermaas_at_nrel.gov>
> wrote:
>
>> That seems perfectly normal to me. You changed the atomtypes on C9 and
>> NE2 (and some others it looks like, but I haven't been following your
>> exploits all that closely), so any term that touches those atoms (and there
>> are likely a ton of them) will be missing parameters. What behavior did you
>> expect?
>>
>> -Josh
>>
>>
>>
>> On 2018-10-27 01:34:07-06:00 owner-namd-l_at_ks.uiuc.edu wrote:
>>
>> The following PRES generated correctly C9-NE2 bonding between organic
>> ligand and HSP of protein
>>
>>> PRES ZAHI 1.116 ! patch for ligand-HSP bonding
>>> ! Patch must be 1-HSP and 2-ZPN
>>> ! use in patch statement
>>> ! follow with AUTOgenerate ANGles DIHEdrals command
>>> GROUP !
>>> ATOM 2C10 CG311 0.332 !
>>> ATOM 2C9 CG314 0.199 !
>>> ATOM 2H91 HGA1 0.09 !
>>> ATOM 2C8 CG321 0.127 ! CD2--NE2--CE1
>>>
>> GROUP ! /
>>> ATOM 2CD2 CPH1 0.219 ! C10--C9--C8
>>> ATOM 2NE2 NR3 -0.20 ! |
>>> ATOM 2CE1 CPH2 0.349 ! H91
>>> DELETE ATOM 2H92
>>> DELETE ATOM 1HE2
>>> BOND 2C9 1NE2
>>> IMPR NE2 CD2 CE1 C9
>>> IMPR NE2 CE1 CD2 C9
>>
>> Problems arise, however, on minimization. Besides the expected request for
>> BONDS
>> CG314 NR3 261.00 1.4900 !
>> CG311 HGA2 309.00 1.1110 !
>> ANGLES
>> HGA1 CG314 NR3 51.5 107.5 !
>> CG311 CG314 NR3 117.00 124.00 !
>> minimization also required angle
>> CG2O5 CG321 CG314
>> and many other angles outside the strict area of bonding. Even on
>> satisfying these requests, dihedrals were also requested.
>> Question: is that normal despite the <AUTOgenerate ANGles DIHEdrals> or
>> did I miss something?
>> thanks for advice
>> francesco pietra
>>
>>
>
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:20:17 CST