Gina,
I have found that TLS success or failure has little to do with either the size of the protein or the resolution. Have been successful with proteins as small as single domain, ~20 kDa (~1.0 Ang resolution), or as large as several NCS-related copies in A.U. (~200 kDa; about 20 TLS groups), and several in between.
Rather, TLS refinement can be exquisitely sensitive to changes that would seem not to be significant. For example, refining a kinase domain (~300 aa), single copy in A.U., ~2.3 Ang data was going fine w/ TLS, until I added in the first ~100 water molecules. TLS blew up. The solution was to alter the B-factors of the added waters (from ~30 to ~50 A^2; Wilson B is ~40 A^2). What worked was a B of ~30; whereas my first several attempts with the B's = Wilson B (or a bit higher) just blew up. Strangely, as expected, though, the B's for the waters did refine up, to ~45-50 A^2. TLS seems to be sensitive to the starting point.
Also, early in refinement, TLS can tend, in my experience, to overcompensate for the thermal motion, and thus the (residual) B's (in the PDB file) drive down to a hard limit (usually set in Refmac at 2.0). My workaround in this case is to delay the TLS until the model (coords & B's) gets a bit better, then try TLS again. (I think someone earlier mentioned on the CCP4BB that fixes are being contemplated to TLS refinement in Refmac to fix this behaviour.)
Hope this helps,
Dave
~~~~~~~~~~~~~~~~~~~~~~~~
David W. Borhani, Ph.D.
Structural Biology Group Leader
Chemistry Department
Abbott Bioresearch Center
Vox: 508-688-3944
Fax: 508-754-7784
Email: [EMAIL PROTECTED]
Smail: Abbott Bioresearch Center, Inc.
100 Research Drive
Worcester, MA 01605 U.S.A.
http://abbott.com
~~~~~~~~~~~~~~~~~~~~~~~~
This communication may contain information that is legally privileged, confidential or exempt from disclosure. If you are not the intended recipient, please note that any dissemination, distribution, use or copying of this communication is strictly prohibited. Anyone who receives this message in error should notify the sender immediately by telephone (508-849-2944) or by return e-mail, and delete it from his or her computer.
| Gina Clayton <[EMAIL PROTECTED]>
Sent by: [EMAIL PROTECTED] 06/22/2006 11:42 AM |
|
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Hi CCP4ers
I am having major problems with a TLS refinement. . .One of the things
that is happening is that the B factors are being assigned to 2.0 with
rfactor 80 and FOM 0.1 for some of my groups (CA or whole molecules)
during say cycle 7 of the refinement. So I tried less cycles (did not
work) or with just the whole molecule and CA only. Now the output is
some of the B factors are assigned values too low i.e 2 to 8 whilst
others seem reasonable. The latter I only got to work using a TLS
refined PDB in restrained refinement with a fixed TLS from the first TLS
I performed. When the TLS refinement goes bad the TLS matrix values leap
to very high numbers i.e. 220 or something.
I am using refined structures (CNS refined). Has anyone ane ideas that
may get this running? For instance should I "jumble" my molecule up
again in a simulated anneal i.e. is it becuase my model is refined
already and I am over refining? Or is it becuase the model is too small
each monomer is about 15-18kDa?
thanks for any guidance
Gina
