Re: [AMBER] force field issues using AT15

From: Peter Varnai <P.Varnai.sussex.ac.uk>
Date: Wed, 27 Jul 2011 12:26:58 +0000

Thanks, Jason for the response.

>> I prepare a protein prmtop file using tleap with current bug-fixed AmberTools15.
>> I obtain different results using the commands below:
>
> What are "results"?

I obtain prmtop files that differ, but I had not tested the corresponding energy values by sander. I realise that the CT to CX change must have effect on the format of the prmtop sections. An energy calculation does show indeed identical results thus no change in the underlying force fields. What was the reason for the CT to CX type change for CA?


>> source leaprc.ff99SBildn
>
> This is the "proper" way to load this force field.

section 2.2.1 of the manual describes this differently, leaprc.ff99SBildn is not mentioned.


>> give almost the same prmtop except the modified Bondi radii - I wonder where these radii are stored for leap and why the new ildn atom types changed those.
>
> These are stored in the leap source code. I forget where off the top of my head, but it's easily enough greppable. Assignments are done based on atom type and bonded neighbors (see the source code for details).
>
> I've written a program that will (among other prmtop modifications) allow you to adjust the radii to whichever set you want. The logic leap uses to assign radii can be found in that source as well, which I can share with you if you want (it may be easier to read the logic in python with separate cases for each radii set rather than in C all jumbled together).

probably the different mbondi assigned by leap are due to some perceived difference in the bonded neighbours depending on which way the prmtop was created.


>> I note that I do not see the latest OL modifications for RNA in ff10 in contrast to what is in the manual.
>
> What's the "latest"? Some chi-mods in addition to the bcn parameters are found in parm10.dat itself, rather than in individual frcmod files. Hopefully a guru can expand on this.

I meant Banas et al J. Chem. Theory Comput., 2010. I got confused as originally they had introduced new C1/C2 atom types that I could not find in parm10.dat. I can see now that these were renamed in parm10.dat to C4/C5 and the corresponding parameters are there indeed. However, isn't it risky to use the same atom type name (i.e., C4/C5) both in nucleic acids and proteins (see ildn modifications) in case of overlap when simulating protein-nucleic acid complexes? I understand that ildn (C3/C4/C4/C5) and CHI_YIL (C1/C2/C3/C4) are not meant to be used with ff10 to avoid potential confusion.

Thanks,
Peter



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 27 2011 - 05:30:04 PDT
Custom Search