Welcome, Guest. Please Login
YaBB - Yet another Bulletin Board
likelihood ratio tests- four-fold degenerate sites (Read 1963 times)
Donna Toleno

likelihood ratio tests- four-fold degenerate sites
Jul 13th, 2004 at 10:47am
I am one of Mike Clegg's graduate students and we have just moved from UC Riverside to UC Irvine.  I apologize for the hotmail address but my new e-mail is not set up yet.  I'm really glad there is a HYPHY message board.  I have met very few people who have experience with HYPHY.   

I would like to partition my data so that I can do likelihood ratio tests for rate parameters using only four-fold degenerate sites.   I know that if I get a list of four-fold degenerate sites I can select them for my partition, but since I am looking at between species data I think that I need to take the tree into consideration when considering which changes happened at four-fold degenerate sites.  Supposedly someone wrote some code in Python that was designed to do this, but I think there is probably a way to do it in HYPHY.

I hope this makes sense.

I appreciate any helpful comments.

Back to top
IP Logged
YaBB Administrator

Datamonkeys are forever...

Posts: 1658
Gender: male
Re: likelihood ratio tests- four-fold degenerate s
Reply #1 - Jul 13th, 2004 at 11:04am
One has to be very careful about how a four-fold degenerate site is defined. As you point out, there is also the need to take the evolutionary history of the sample into account. This filtering can be accomplished in HyPhy - although not in an entirely straightforward manner, but before I go through with the instructions on how to do it, may I suggest that you try testing relative rates in the context of coding models by constraining only synonymous rates? That way the notion of four-fold degeneracy is subsumed by a more general notion of synonymous substitution. You can do this type of test by choosing:

Standard Analyses
Codon data
Any codon model with local parameters (so that each branch has a separate dN and dS)
Constraining only the synonymous rate for the test.

If you still want to use nucleotide four-fold degenerate sites (which must be defined in terms of neighboring nucleotide positions anyway), I can tell you how to do that, but codon models still have to be involved in order to recover codon ancestral states implicit in the definition.

Back to top

Associate Professor
Division of Infectious Diseases
Division of Biomedical Informatics
School of Medicine
University of California San Diego
IP Logged