Can homoplasy be measured?

Guy Hoelzer hoelzer at med.unr.edu
Sun Jan 10 21:43:07 EST 1999


In article <778a9t$90b at net.bio.net>, marjon van dijk
<m.vandijk at bioch.kun.nl> wrote:

>Given a multiple alignment of ten mammal sequences, is it possible to judge
>objectively that two mammals with the same amino acid on one site and the
>other eight mammals having another amino acid on this site have a shared
>derived character or that it is homoplasy? So, are there methods to
>calculate the rate of homoplasy for each site in the alignment?

There are may methods for estimating the amount of homoplasy in a data
matrix.  See the recent book called "Homoplasy" by Sanderson and Hufford. 
Identifying particular homoplastic events, or estimating homoplasy on a
site-by-site basis, usually requires comparison with hierarchical patterns
in the rest of the matrix.  Maximum likelihood approaches might offer an
exception to this, but they usually invoke additional assumptions that you
might not be willing to make.

It seems odd to me to discuss a "rate of homoplasy", because I do not
consider homoplasy a process.  It is the result of homoplastic evolution
(convergence, etc.), which would have a rate.
>
>In order to compare different genes is it, knowing that not all sites
>change equally, possible to calculate the rate of homoplasy for a complete
>gene? 

This would simply be the mean of all of the site-specific rates.

-- 
Guy A. Hoelzer                 phone:  702-784-4860
Department of Biology          fax:  702-784-1302
University of Nevada Reno      email:  hoelzer at med.unr.edu
Reno, NV  89557




More information about the Mol-evol mailing list