Re: [AMBER] Amber force-field compatibility

From: Ross Walker <>
Date: Mon, 8 Nov 2010 13:32:50 -0800

Hi Jarmila,


I saw your email from several days ago but have not had time to properly
reply. You should post such questions to the amber mailing list since then
you will get a more prompt reply and from people who can better comment on
the appropriateness of certain force fields.


My gut instinct would be to use FF99SB with BSC0 - we tend to see slightly
better performance for proteins with FF99SB and the charge model is the same
as that for BSC0 - I have not seen anything showing the compatibility of the
point charge model between FF03 and other force fields that use a slightly
different approach and so I am wary of this. I would suggest looking at what
the authors used as the basis when they fitted BSC0 and go with this.


Others might be able to comment more.


As for Gromacs 4.5 with AMBER make sure you use double precision and NOT the
pure single precision implementation as we have seen issues with this
regarding energy conservation and stability.


Good luck,



From: Jarmila Husby []
Sent: Monday, November 08, 2010 2:38 AM
Subject: Amber force-field compatibility


Dear professor Walker,

I am about to set up an MD simulation of a large protein-DNA complex on
Gromacs 4.5 (on Nvidia Tesla card) with Amber ports for gromacs. I would
like to ask you about the compatibility of the Amber FF03 (for proteins)
with bsc0 (for NA) versus the FF99sb with bsc0.

My main focus of the simulation is the protein-DNA interaction, so I am
searching the most favorable force-field combination. Personally, I would
rather combine the parm99sb with bsc0 as they are both derived from
parm94/99 with corresponding concept of deriving the partial charges. But
recently I came across a number of papers (by Ramsden & Kunkel) where they
used the combination of FF03+bsc0. Would you recommend this FF combination
despite the FF03 is polarized and bsc0 is the parm99 "derivative"? What is
the rational of these two force-fields being used together? And would you
expect significantly different results between the parm99sb+bsc0 and
FF03+bsc0 simulations?

I would greatly appreciate your comments and suggestion.

Thank you in advance for your reply,

Best regards


Jarmila Husby, PhD Candidate

CRUK PPI Drug Discovery/ BMSG Group

Department of Pharm & Biol Chemistry

The School of Pharmacy, University of London


tel: (44) 0207 753 5996

AMBER mailing list
Received on Mon Nov 08 2010 - 14:00:03 PST
Custom Search