AMBER Archive (2008)

Subject: Re: AMBER: B-factor

From: Thomas Cheatham III (
Date: Wed Apr 02 2008 - 12:18:18 CDT

> I`ve calculated the B-factor of my Molecular Dynamic
> Simulation with AMBER 9.0 (ff99SB) and I`ve used the script:
> reference ../Leap/INHA.pdb
> trajin mdcp_teste_fit.crd
> rms first out rms_bb_8_10ns.dat :1-268_at_CA,C,N,O
> atomicfluct out bf_8_10ns.dat @CA,C,N,O byres bfactor
> The B-factors calculated by AMBER are below experimental ones.
> P.S. The B-factor were calculated on the last 2 ns of a 10 ns running.

...and what is your question?

Note that I would probably calculate the atomicfluct over all the protein
atoms, however this may not alter your observations. Also, seeing "less"
fluctuation is contrary to experience, since historically, MD simulation
tended to over-estimate motion. However, with recent improvements in the
force fields, such as with ff99sb, ff03 or CHARMM c22/CMAP, motion more in
line with experiment is typically observed. Very nice (independent from
the AMBER force field developers) recent work is that from Bruschweiler's
group, specifically JACS 129, 4158-4159 (2007), JACS 129, 14146-14147
(2007) and JCTC 3, 961-975 (2007). Now these papers all relate to NMR
observables rather than crystal B-factors. If the MD fluctuations are
"less", perhaps adding in the hydrogens and/or including MD averages over
longer time scales may bring the B-factors up more in line with
experiment. Perhaps long timescales are needed to access low frequency
fluctuations in the structure? Alternatively, perhaps something in the
simulation protocol is inhibiting motion (restraints, elevated pressure,
reduced temperature, ...).

-- tec3
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to