Hi Ed, that's all right. Both ways have clear pros and cons (I have a few pages long email that I wrote a year ago to someone that carefully lists them all), but are ok as long as a) things are self-consistent, and b) track about what was done is recorded. The only reason I made that remark is to point out that the previous behavior was not strictly speaking a bug, that's all. At some point Garib made a very nice comment about this (in ccp4bb list). All the best, Pavel On 4/6/13 8:13 AM, Ed Pozharski wrote:
Pavel,
On 04/06/2013 05:00 AM, Pavel Afonine wrote:
A quick comment on "B-factor problem" recently brought to attention: in fact it is not a bug at all, but a conflict of philosophies. It all boils down to whether you want to keep meaningless bulk overall B-factor in ATOM records in your PDB file (current phenix.refine behavior) or you say it is junk and keep it in overall scale factors (previous behavior that many of you called a bug). Personally I would prefer to keep in overall scale factors.
While my useless comment has a potential to start an endless and equally useless discussion, I'd like to nevertheless point out that from my understanding overall B-factor is not entirely meaningless. While some of it is due to overall static disorder which is truly of little interest, figuring out what part of it is contributed by true thermal motion (whether real at 100K or frozen out from 300K) is not exactly trivial. I appreciate your opinion that from mathematical standpoint it does not matter where this contribution goes, but having B=0 for some atoms because min(B) was subtracted and placed into a header comment line is at least equally meaningless. I do not see how B-min(B) is a better assessment of the "true atomic B-factor".
Cheers,
Ed.