DNA vs amino acid sequences
ts15 at cornell.edu
Fri Jun 23 14:37:11 EST 1995
In article <DA925E.7My at zoo.toronto.edu>, mes at zoo.toronto.edu (Mark
> In article <3rghpv$hm6 at nuscc.nus.sg> mcbbv at leonis.nus.sg (Venkatesh
> >Dear netters,
> > I am now in the process of generating a phylogenetic tree of actin
> >sequences from some lower
> >vertebrates. I am wondering which sequence I should use - DNA or amino
> >acid? Which is more appropriate for a highly conserved protein like the
> Your question is a wise one, yet it cannot be answered except empirically.
> It may be that it is so conserved in the scope of your question that
> aa does nto resolve all groups. It may be so poorly conserved that
> bases lead to inordinate homoplasy.
> I would recommend taking an "experimental" approach to the issue. That is
> look to see how sensitive your groupings are to different ways of
> handling the data.
> 1) aa translation only.
> 2) all bases equal weight
> 3) first and second position only
> 4) first and second position weighted heavier than third but still count third
> 5) all sites subject to TS/TV weighting.
> 6) first and second weighted to the same weight as TV in third and TS in
> third weighted less.
> Don't forget to consider the possibility of counting gaps.
> > Can someone lead me to recent publications that compare the
> >pros and cons
> >of using DNA vis-a-vis amino acid sequences for generating phylogenetic
> Mark E. Siddall "I don't mind a parasite...
> mes at vims.edu I object to a cut-rate one"
> Virginia Inst. Marine Sci. - Rick
> Gloucester Point, VA, 23062
Another method to try on your amino acid sequence characters is "protein
parsimony," in which it is made more costly to change between amino acids
that are separated (sequence-wise) by a third amino acid. (I.e., if by
changing only one base at a time it is only possible to transform one
amino acid into another by passing through a third amino acid, the cost
(in number of parsimony steps) is twice that of a change between two amino
acids that can be transformed directly into one another. This can be
implemented by generateing a protein parsimony step matrix in MacClade,
then running the analysis in PAUP after assigning the step-matrix
character type to the amino acid characters. I understand that protein
parsimony can also be applied in PHYLIP, though I have never tried this.
ts15 at cornell.edu
More information about the Mol-evol