AMBER Archive (2009)

Subject: Re: [AMBER] B-factor

From: Thomas Cheatham III (
Date: Fri Oct 16 2009 - 17:16:35 CDT

> I know that lots has already been discussed about the b-factor, but I've
> noticed some strange behaviour when calculating the b-factor with ptraj.
> If I ask to calculate just the bfactor without calculating previously
> the rms I obtain some very strange/huge values for the b-factor. More

Yes! If you do not do the RMS fit, then you're atomic fluctuations will
include rotation and translation too and hence be large (although since
you are centering, you have removed much of the translation)... Ideally,
you would create an average structure over your trajectory, RMS fit to
this, and then do the "atomicfluct". You can RMS fit to first, but this
is not exactly the same as RMS fitting to the average structure.

> The two answers are completely different. I am not sure why is so. I
> know that b-factor is calculated as 8/3*p^2*fluct, but aren't the two
> actions done independently?

Yes, B-factor is independent of RMS fit, i.e. it does not implicitly do an
RMS fit... --tec3

AMBER mailing list