Re: [AMBER] sleap/tleap differences for parmbsc0 DNA

From: Jason Swails <jason.swails.gmail.com>
Date: Mon, 21 May 2012 08:56:24 +1000

On May 16, 2012, at 9:38 PM, "C.M. Baker" <cmb84.cam.ac.uk> wrote:

> On May 15 2012, case wrote:
>
>> On Tue, May 15, 2012, C.M. Baker wrote:
>>>
>>> I'm running some simulations of DNA (canonical B-form) using amber with
>>> the parmbsc0 parameter set. Everything seems to work fine, but I've
>>> noticed that I get different results depending on whether I use tleap or
>>> sleap for my initial system set-up.
>>
>>>
>>> source leaprc.ff99bsc0
>>> dna1 = loadPdb 1BNA_no_waters.pdb
>>> saveAmberParm dna1 1bna_99.prmtop 1bna_99.prmcrd
>>
>> At this point, you could use readparm to print and examine the dihedral
>> angle parameters (say esp. around delta) in the prmtop files created by
>> tleap and sleap. Looking at energies of the initial coordinates should
>> also show whether or not there are any differences that would eventually
>> result in different distributions of angles.
>>
>> leaprc.ff99bsc0 reads in a main parameter file, followed by two frcmod
>> files. This leads me to suspect that the differences may arise from how
>> frcmod files are interpreted. One way to test this is to try sleap vs.
>> tleap using leaprc.ff10, which does not make use of frcmod files. Again,
>> I'd think that close examination of the prmtop files, or of single point
>> energies, should suffice; one presumably doesn't need to run converged
>> simulations from each option.
>>
>> ....dac
>>
>
> Thanks for the quick reply. As you suggested, I've been running some single
> point calculations, and looking at the resulting energies.
>
> With parmbsc0, if I read in my DNA molecule and calculate a single point
> energy (no ion, no solvent), I get exactly the same results regardless of
> whether I use tleap or sleap.
>
> However, if I add solvent then things change. Now if I do a single point
> calculation, the angle energies are very different. And if I look at the
> resulting prmtop files, is shows that there are number number of atom types
> in each case:
>
> tleap:
>
> %FLAG POINTERS
> %FORMAT(10I8)
> 12953 17 12467 544 640 836 1294 1520 0 0
> 20470 4089 544 836 1520 43 86 48 28 1
> 0 0 0 0 0 0 0 2 34 0
> 0
>
> sleap:
>
> %FLAG POINTERS
> %FORMAT(10I8)
> 12953 16 12467 544 1280 1672 1294 1520 0 0
> 20470 4089 544 1672 1520 32 59 48 16 0
> 0 0 0 0 0 0 0 2 34 0
> 0
>
> Looking at the "%FLAG ATOM_TYPE_INDEX" section of the prmtop files, the
> first atom in the parameter file (HO) is given type 1. When I use sleap,
> all of the water H atoms are also give type 1; when I use tleap, the water
> H atoms are given type 17.
>
> This seems a little strange, and like it might well result in differences
> in simulations run with the resulting prmtop files. If you have any
> thoughts on what is going on here, I would be grateful to hear them.

This will not result in differences. The reason is that water atom types have the same Lennard jones parameters as hydroxyl groups, so they don't really need to be separated. It's not wrong if they are (which is what tleap does) or if they are (which is what sleap does). Determining prmtop equality is very difficult.

Do you use the same inpcrd file for both tests? Placing solvent is up to how the software decides to do it, so tleap and sleap may do different thing. Try both prmtops on both sets of coordinates (so 4 total calculations) and see if the numbers match for both prmtops on the same sets of coordinates.

HTH,
Jason

--
Jason Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun May 20 2012 - 16:00:03 PDT
Custom Search