IUBio Biosequences .. Software .. Molbio soft .. Network News .. FTP


GGUTMAN at vmsa.oac.uci.edu GGUTMAN at vmsa.oac.uci.edu
Wed Jan 26 17:25:33 EST 1994

Genevieve Fourel asked a while ago about GCG'c DIVERGE program, and
I am posting my reply to her:
I have not used the GCG DIVERGE program, but I have a fair bit of
experience with the calculations on which it is based (Perler et al., Cell,
20:555-566, 1980).  To answer your three questions:

     First, it is expected that the results are independent of the order in
which they are presented.  The calculations are done in such a way that
*both* directions of change are evaluated and averaged (Sequence A goes to
Sequence B, and vice versa.)

     Second, no confidence intervals have been calculated for these
numbers, to my knowledge.  How big a difference between two comparisons is
required before you consider it meaningful is completely up to you.

     Third,  the original Perler et al. paper *does* present a single
average value for the divergence of two sequences at silent sites and
another for replacement sites; I can't imagine why DIVERGE does not do
this.  (The average is not simple, however; it is actually the average of
*two* averages, the first weighted by the number of sites in each of the
three categories, the second by the number of changes in each category.)

Having said all this, I must point out that the Perler calculations have
long been recognized as being seriously flawed because of their
assumption that transitions and transversions are equally likely.  Brown et
al. (J. Mol. Evol. 18:225-239, 1982) pointed this out, and provided a
modified calculation which took into account the transition/transversion
ratio.  It is unfortunate that DIVERGE does not do this, but I have written
a DOS (PC-compatible) program to do it, which is available to anyone

* George A. Gutman                                            *
* Department of Microbiology            Tel:(714)856-6593     *
*          and Molecular Genetics       FAX:(714)856-8598     *
* University of California           BITNET: GAGUTMAN at UCI     *
* Irvine CA 92717                  INTERNET: GAGUTMAN at UCI.EDU *

More information about the Info-gcg mailing list

Send comments to us at biosci-help [At] net.bio.net