simon.brocklehurst at camb-antibody.co.uk
Fri Mar 5 05:08:23 EST 1999
> Dear All,
> My question is this: Can anyone help me in how to determine if atoms are
> bonded in PDB files?? Someone recommended that I bond all atoms < 1.7A
> together, but this is yielding too many bonds in comparison to the
> number that RasMol generates, but RasMol uses a Voxel engine in
> determining bonds from what I see.
You had good advice in the first place (although, I would use 1.8 Anstroms
myself) - your algorithm should work fine for proteins, but you need to bear
a couple of things in mind:
o The structures you're looking at need to have good stereochemistry - that
is, the techniques used to build the co-ordinates in the first place should
have been correctly applied.
o If you are looking at an all-atom structure - that is, hydrogen atoms have
been included explicitly then you need it won't work quite properly
(hydrogen atoms can be closer than 1.7 Angstroms and not be covalently
bonded). Thus you will need to make a special case of hydrogen atoms
o Your cut-off won't recognise disulphide bonds, i.e. in PDB-speak, SG-SG
bonds between CYS residues won't be identified. Use a 2.3 Angstrom cut-off
in this case.
Good luck with your structure viewer!
Simon M. Brocklehurst, Ph.D.
Head of Bioinformatics & Advanced IS
Cambridge Antibody Technology
The Science Park, Melbourn, Cambridgeshire, UK
mailto:simon.brocklehurst at camb-antibody.co.uk
More information about the Proteins