Ed, <snip>
> Maybe what we need is the new graphic representation, where higher > B-factors will be shown not by red color, but by some kind of > "fuzziness". I don't think thicker tubes really deliver the message, > not to mention how ugly they look. </snip> I have thought that combining a thermal ellipse (or sphere for isotropic Bs) with transparency, so that the atom sphere fades as it grows larger would convey the meaning of high B-factors more readily to non-crystallographers. What do you think, Warren, about adding this to PYMOL? On the topic of what to include. I have tried multiple copy refinement, and the problem is a practical one of insufficient data to restrain the refinement when these extra parameters are added. At the conceptual level multiple conformation refinement should be used for all the residues in the protein, and the backbone should be free to move also. Again, we are limited by our data and TLS refinement is a practical, and very powerful method of modeling disorder. Tom Blundel's RAPPER program is another, very inventive, method of modeling disorder. The RAPPER output has the advantage of providing several conformations, which hopefully, sample the true conformational freedom of the protein and as such are more readily interpretable by non-crystallographers. Alas, even I have not applied this technique to any of my deposited structures, but I do recommend that you try it. Best regards, Mark PS. The standard B-factor restraints are too tight, resulting in too small a B-factor for most sidechains. I recommend increasing the sigma values, using TLS, or Ian Tickle's variable restraints as implemented in PMB/CNS (http://xray.utmb.edu/PMB). Also, PHENIX has a different method of restraining B-factors which may be less prone to over-restraint. Peter or Paul, do you have any information about this? On Wed, 2007-01-10 at 13:16 -0500, Edwin Pozharski wrote: > *** For details on how to be removed from this list visit the *** > *** CCP4 home page http://www.ccp4.ac.uk *** > > > Nick, > > as you see from the responses you got, there are really three options: > > 1. Set the occupancies of atoms in question to 0.0 > 2. Remove the atoms in question > 3. Let refinement take care of it by "inflating" B-factors > > In terms of refinement, first two options may seem to give identical > results, but they in fact differ. Zero occupancies will give zero > contribution to Fcalc, whereas with atoms removed the corresponding > volume will be filled with bulk solvent. IMHO, the first option (zero > occupancies) thus lead to the biggest possible error introduced to your > model. The third option will contribute to Fcalc about the same way the > second option does, because the corresponding volume will be "filled" > with diffuse density resulting from high B-factor. > > So, I guess the zero occupancy option should be rejected because it > introduces larger error into your model. I would also reject it on > semantic grounds. Crystallographic models with isotropic B-factors > describe every atom in your structure using 5 numbers - three > coordinates (x,y,z), occupancy (q) and B-factor. The meaning of it can > be summarized as follows: > > "This atom (known from chemical information to be present in the > structure) spends the q fraction of time in the vicinity of (x,y,z), > with the probability to find the atom at a certain distance from it > falling off according to the normal distribution whose width is defined > by the B-factor" > > Zero occupancy is thus complete nonsense - what it means is that the > atom is NEVER in the vicinity of (x,y,z) given in your pdb-file. It is > not even zero-information statement, it is plain wrong, given that you > choose, say, the highest probability rotamer. The atoms missing from > density in fact has some probability to be where you put them! > > So what about omitting atoms from the structure? You are then just > saying that you have no idea where these atoms are, which sounds OK. > Indeed, structural models are always incomplete. If I replace > individual waters and whatever else is in my mother liquor by bulk > solvent, why should disordered sidechains be special? Here is why: > > No protein structure (except for crambin, I presume) is ever determined > without prior knowledge about its chemical structure. Unless you have > strong reasons to believe that an xray fairy has cut off this lysine > sidechain with a tiny magic saw with an explicit purpose to drive you > crazy, you should leave atoms in because you know from sequence that > they are there. The B-factor of 200 corresponds to rmsd of ~1.6A, which > transforms to "atom spends 95% of time inside a sphere of 6.4A in > diameter", so you are not really narrowing down its spatial position too > much. In fact, covalent bonds will keep it in much smaller volume. > > Crystallographers worry, however, that non-crystallographers may misuse > their models because non-crystallographers don't understand the basic > concepts behind B-factors. IMHO, it is crystallographer's job then to > educate non-crystallographers. But at the end of the day, it is > driver's responsibility to learn the meaning of red light, and if a > reasonable effort was made by RMV to deliver the information, that is > all what they can do. > > Maybe what we need is the new graphic representation, where higher > B-factors will be shown not by red color, but by some kind of > "fuzziness". I don't think thicker tubes really deliver the message, > not to mention how ugly they look. > > Ed. > > Nicholas Noinaj wrote: > > *** For details on how to be removed from this list visit the *** > > *** CCP4 home page http://www.ccp4.ac.uk *** > > > > > > Hi, > > > > i would like to get opinions on whether or not one removes side-chain atoms > > where there is no density. for example, if one can only observe density up > > to the beta-carbon for lysine (say at > 0.5 sigma), does one leave the > > lysine side chain intact, knowing it must be disordered, or does one > > terminate at the beta-carbon, making the coordinates reflect what is > > actually observed in the density. > > > > It seems both approaches are published and people seem to have conflicting > > opinions on the topic. It would be nice to come to some concensus, > > possibly clear up the issue for us newbies. > > > > Thanks in advance for all feedback! > > > > > > > > Cheers, > > NIck > > > > > > > > > > ________________________________________ > > > > Nicholas Noinaj > > University of Kentucky College of Medicine > > Department of Molecular and Cellular Biochemistry > > The Center for Structural Biology > > Biomedical Biological Sciences Research Building, Rm 236 > > 741 S. Limestone > > Lexington, Ky 40536 > > Lab: 859-323-8183 > > Cell: 859-893-4789 > > Home: 859-228-0978 > > [EMAIL PROTECTED] > > noinaj.com > > > > > > > > > > > > > > Sincerely yours, Mark A. White, Ph.D. Assistant Professor, Dept. Biochemistry and Molecular Biology, Manager, Sealy Center for Structural Biology and Molecular Biophysics X-ray Crystallography Laboratory, Basic Science Building, Room 6.660 C University of Texas Medical Branch Galveston, TX 77555-0647 Tel. (409) 747-4747 Fax. (409) 747-4745 mailto://[EMAIL PROTECTED] http://xray.utmb.edu http://xray.utmb.edu/~white
