These domain assignments are primarily taken from the literature. Where
this stategy was inappropiate (e.g. not defined in publication), domain
assignments were by visual inspection. These assignments were made
for a non-redundent representative
set of the April 1996
Brookhaven protein data bank (PDB).
Domain assignments for protein homologous
to this set were assigned by projecting the author assignment onto a multiple
sequence alignment.
We prefer to use authors' assignment of domains, where possible, as they generally
incorporate knowledge from a variety of sources in their decision making
process. For example, structural similarity between proteins and identification
of functional modules augment consideration of distinct spatial segments. To
date, neither visual inspection nor computer algorithms incorporate all of these
concepts.
The non-redundent set was created by using
BLASTP and complete linkage cluster
analysis to created groupings of structures all having BLAST probabilities
of 10e-10 or lower. Alignments of all sequences in these sets were
calculated using the AMPS package.
Our domain assignments can be viewed using Rasmol or PREPI (Silicon Graphics Computers Only). The domain assignments are shown using different colour schemes.
For further information on the proteins, we have provided links to file header of the Brookhaven protein data bank (PDB) entry, SCOP (Structural Classification of Proteins) and The NCBI WWW Entrez Browser