Dear Eleonor and Bullitin board people,
I would like to comment on the "weird numbering schemes" mentioned in the email below. Although they make life (slightly) more difficult for developers of crystallographic software, they make life much easier for many of the rest of the world and have been introduced for good reasons by some very good crystallographers. I work with 6 different serine proteases, all numbered according to the chymotrypsin numbering scheme, introduced by Wolfram Bode. Here one immediately knows which residues are the active site residues and one can easily superimpose the structures by superimposing identical residue numbers without first having to do a sequence alignement, or looking into some table to find equivalent residues. Also by now, a vast amount of literature has accumulated with this numbering scheme. It is very nice to be able to read these papers without having to frequently consult sequence alignments. I also work with different protein families where no such consensus sequences exist and this is much more cumbersome.
The programs I have been using until very recently: CNS and Quanta did not have any trouble handling these consensus sequences. However, the more recent programs Refmac, Coot and Pymol are not able to handle these consenses sequences properly. Of course, workarounds like renumbering to some linear sequence before refining, and renumbering back to the original sequence afterwards are possible. However, these kinds of workarounds to compensate for weaknesses in software are something I find archaic. But that is again a personal view.
Herman Schreuder
-----Ursprüngliche Nachricht-----
Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag von Eleanor Dodson
Gesendet: Dienstag, 10. April 2007 11:09
An: [log in to unmask]
Betreff: Re: [ccp4bb] antibody refinement in REFMAC with Kabat numbering
John Pak wrote:
> Hi all.
>
> I'm currently refining an Fab structure. All was going reasonably
> well until I renumbered the PDB according to the Kabat numbering
> convention. After which, REFMAC does not refine the inserted residues
> properly (ie. residues 82A, 82B, 100A, 100B, 100C, etc.). It refines
> these residues into a steric jumble. I think REFMAC can't recognize
> that the residues are connected or something.
>
> So my question is, how do you refine this type of PDB in REFMAC? Or
> do you just wait until the refinement is done, then change all the
> residue numbers and connect entries in the header afterwards?
>
>
Yes! - or better, let the deposition site renumber if they must.. I feel these weird numbering schemes are archaic - we can easily align sequences and see that 82A is actually 83 etc.. But that is a personal view.
Eleanor
|