Re: Unequal branch lengths
- From: "Perplexed in Peoria" <jimmenegay@xxxxxxxxxxxxx>
- Date: Fri, 16 Sep 2005 13:50:36 -0400 (EDT)
"IRR" <iotarhorho@xxxxxxxxxxxxxxxxx> wrote in message news:dgdjnn$1eid$1@xxxxxxxxxxxxxxxxxxxxxx
> "Perplexed in Peoria" <jimmenegay@xxxxxxxxxxxxx> wrote in message
> news:dg84tv$2gr1$1@xxxxxxxxxxxxxxxxxxxxxx
> >I am puzzled by the phenomenon of unequal branch lengths in
> > phylogenetics. Not so much by the confounding effects
> > ('long branch attraction' and the like) as by what the
> > existence of unequal branches tells us.
> >
> > A long branch is one in which the rate of substitution
> > per year is high. (Well, in rooted trees, anyways. Is any
> > other interpretation of long branches possible?)
>
> The other interpretation would be that the branch length is an artifact, as
> can happen if the assumptions behind the canonical bifurcating tree model
> are violated, e.g. long-branch attraction, gene (or genome) fusions, etc.
Agreed, But see my reply to Guy.
> > Why the higher rate? It is faster evolution, in some sense,
> > but why? I can imagine a variety of explanations, including
> > weaker selection, stronger selection, faster mutation,
> > redundancy in genomes, and mutation pressure. Are there
> > other possible explanations? Is there a way of resolving
> > the cause?
>
> Yes (!) to all of your explanation, and certainly even neutral evolution can
> feasibly lead to unequal branch lengths (brings to mind the old rhetoric on
> why there always seem to be more birds on one side of the 'V' formation).
> Much effort has/is put into distinguishing these and other explanations,
> IIRC, but much focus is on resolving the signature of positive or purifying
> selection. This is typically done by looking at the ratio of synonymous
> versus nonsynonymous substitutions at specific sites in a protein coding
> gene, where positive selection is expected to skew toward nonsynonymous
> substitutions (and vice versa). This has been used (in the textbook
> example) to illustrate that the variable (antigen-recognition) regions of
> immunoglobulins are under positive selection for amino acid
> variability/diversity.
Yes. That is interesting. I can see how it might work. Unfortunately,
it is inapplicable to RNA datasets, or to deep branching trees in which
the synonymous substitution 'signal' has been degraded to noise. However,
I can also see how that idea of looking at the ratio of true synonyms to
presumably neutral changes might be used to construct an amino-acid
similarity table which would be more suitable for evaluating the optimality
of the genetic code (as I discuss on a different thread).
Well, maybe. As the phenomenon of codon usage bias (particularly in
microorganisms) shows, even 'synonymous' mutations may not always be
neutral.
> > In the case of the long eukaryotic branch in the rRNA
> > tree of life, I can imagine that it may be due to ur-eukaryotes
> > having redundancy in their rRNA genes due to multinucleate
> > or diploid genomic organization. This, I imagine, allows
> > a stem-weakening mutation to be tolerated as a pseudo-gene,
> > which is then later restored to functionality by a stem-fixing
> > mutation at the paired position. So redundancy allows you
> > to cross a chasm in a Wright landscape by effectively
> > making the chasm shallower. This is in contrast to the
> > effect of large populations (increasing the likelihood of
> > near-simultaneous mutations) which facilitate crossing the
> > chasm by, in effect, making the chasm narrower. Does anyone
> > know of a mathematical treatment of these issues which
> > supports or corrects these intuitions?
>
> :/ not sure about the specifics of eukaryotic rRNA as you ask here, but a
> good/seminal starting point for the idea of relaxed selection due to
> multiple/redundant gene copies is "The evolutionary fate and consequence of
> duplicate genes" (Lynch and Conery, Science 200[0/1]). Also interesting
> note that many prokaryotes have multiple copies of their rRNA operons -- as
> many as 7 "redundant" copies the last time I checked!
Thanks, that sounds like a review article that I definitely want to read.
Hmmm. I know I have read that there is little divergence among those
redundant copies in prokaryotes due to 'concerted evolution', probably
cause by 'gene conversion'. But then, the same mechanisms are available
in eukaryotes. So it appears that my 'redundancy hypothesis' needs some
work.
> > It seems clear that conserved secondary structures in rRNAs
> > must introduce correlations between positions in the sequence
> > data. It also seems possible that conserved structure in
> > proteins would introduce similar correlations. Is anyone
> > looking at these kinds of correlations - perhaps as a clue
> > to structure, or an aid in alignment, or in any other way?
>
> Google "ribosomal database"... there are a few projects doing exhaustive
> alignments of sequenced rRNA's. In most/all cases, these alignments are
> derived ultimately from a number of secondary and tertiary RNA structural
> models. Obviously the number of structures available is small, but with a
> well-conserved molecule such as the small subunit rRNA, the alignments and
> derived trees appear to be very robust.
Thx. Larry's comments sent me down that same path. Do you happen to
know whether anyone has tried doing a three domain tree using the small
subunit? Or is it the case that you can't go back that far with so little
data?
.
- References:
- Unequal branch lengths
- From: Perplexed in Peoria
- Re: Unequal branch lengths
- From: IRR
- Unequal branch lengths
- Prev by Date: Re: A fully developed creature can evolve?
- Next by Date: Re: A FUNDAMENTAL ISSUE
- Previous by thread: Re: Unequal branch lengths
- Next by thread: Re: Unequal branch lengths
- Index(es):
Relevant Pages
|
