[phenixbb] NCS, TLS and anisotropic b-factors

Pavel Afonine pafonine at lbl.gov
Tue May 8 11:14:29 PDT 2012


yes, we do observe cases when using NCS in ADP refinement is not useful. 
This does not suggest that using NCS as a priori knowledge should not be 
used in ADP refinement. However this does suggest that we still yet to 
find the best usage of it.


On 5/8/12 7:55 AM, Nathaniel Echols wrote:
> On Tue, May 8, 2012 at 7:45 AM, Morten Groftehauge
> <mortengroftehauge.work at gmail.com>  wrote:
>> Okay, brilliant. I had a base go off the rails with B-factor and it's NCS
>> mate got weird positive difference density. So I instantly looked for stuff
>> in the manual to explain it. Like I said it was a stupid question =)
> One word of caution: the use of B-factor restraints on NCS-related
> atoms frequently makes structures much worse - Jeff and I have found
> lots of examples where chains have nearly identical conformations but
> wildly different B-factors.  So you might want to experiment with
> setting the NCS B-factor weight to zero.  (The reason this isn't the
> default is a) historical baggage, and b) where appropriate, it does
> help a little bit with overfitting.)
> -Nat
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

More information about the phenixbb mailing list