Multiple alignment of protein sequences

doelz at urz.unibas.ch doelz at urz.unibas.ch
Fri Feb 7 05:13:35 EST 1992


In article <1992Feb6.141356.9142 at athena.cs.uga.edu>, WEISE at gandal.dnet@server.uga.edu (MICHAEL WEISE) writes:

> In trying to decide between meaningful homology and garbage, also note that
> GCG7's BESTFIT has an optional parameter ( /RANDOMIZE=n ) that provides a mean
> and std.dev. for n comparisons where one of the sequence pair is randomized. 
> If the score for the nonrandom alignment is 'significantly' greater than the

True, true. But remember that statistical significance must not necessarily
reflect what you consider to be biological significance. I remember that there
was an old paper of Lipman et al in the middle 80's on statistical problems 
in the sequence comparison field. While BESTFIT's randomize is rather crude 
I should recommend W.Pearson's RDF2 which also permits to suffle *windows*
of the sequence, thus leaving the sequence more intact than the pure 
monte carlo would permit. RDF2 is described in various papers including 
Pearson and Lipman, PNAS (1988) 85:2444-2448, and available from William 
Pearson. 

Regards 
Reinhard

************************************************************************
    Dr. Reinhard Doelz            *     EAN     doelz at urz.unibas.ch
      Biocomputing                *     DECNET  20579::48130::doelz
Biozentrum der Universitaet       *     X25     psi%022846211142::embnet
   Klingelbergstrasse 70          *     FAX     x41 61 261- 6760 
     CH 4056 Basel                *     TEL     x41 61 267- 2076 or 2247   
************************************************************************



More information about the Proteins mailing list