***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


Hi Tommi,
your points are well taken.  I am well aware of such examples partly
from my own work on helical cytokines which share the same fold with
sometimes barely 10% seq.id.  By structural similarity I meant
structural agreement as judged by r.m.s.d values. Afterall, we often
find that fold agreement does not necessarily mean that a structure can
serve as a good model in MR approaches. 

My intention was actually to mention some of the original papers that
formed the basis for the 'rule of thumb' that Marilyn Yoder wanted to
know about.  You are correct about the interpretation of fig 2 in the
1986 paper in EMBO J. and in fact  this had already showed that this
rule of thumb may be a bit too conservative.  What we do need to point
out, however, is that the plot in fig 2 arose from aligning core
residues of the structures and not the entire structures. As one other
contributor to the discussion pointed out, the alignment length is an
important parameter when trying to draw conclusions about sequence
divergence and structural similarity. And that is nicely illustrated in
the more recent papers on the topic that have already been mentioned in
the discussion. 
Best wishes
Savvas



-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
Tommi Kajander
Sent: donderdag 22 september 2005 16:00
To: [email protected]
Subject: RE: [ccp4bb]: sequence/structure homology


***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


On Thu, 22 Sep 2005, Savvas N. Savvides wrote:

> In that paper you will also find a plot of r.m.s.d vs sequence
> identity which reveals that the 3 angs rmsd (which we generally take 
> as an upper-boundary for structural similarity) occurs at about 25-30%

> seq.id.
>  
> Best wishes
> Savvas
> 
 
-I dont think this true a) personally i know explames -at least 
beta-propellers with seq id less than 15% and rmsd around 2 A, 
and i am sure many more.
b) If you check the EMBO J Chothia paper (cant find it now, but the same
data is in Creightons book) you see rmsd 2 A occurs below 20% identity. 

The other question is what do you mean by structural similarity? surely
proteins can have the same fold wihtout 
practically any >20% seq. identity ---if this about feasibility of use
as a molecular replacement, but thats another thing.

Tommi



Reply via email to