This is in reply to secondary structure based alignments of 18S rRNA:
When I began trying to find out how alignments could be done to "take
secondary structures into account" I encountered a lot of hand waving
(I mean a LOT of hand waving) and finally realized that most people
either use the RDP alignments or adjust alignments by eye using some
kind of sequence editor. I was not satisfied with the RDP
alignments for my work with overall animal phylogeny and finally
discovered the DCSE editor and a database of aligned rRNA sequences
at an FTP site in Belgium (ftp://uiam3.uia.ac.be) that is maintained
by De Rijk, Van De Peer, and De Wachter.
I find their alignments work better than those at RDP (at least for
animal phylogeny). The DCSE editor allows you to do real secondary
structure based alignments. It has extensive built in abilities for
helix (stem) matching and you can include helix and loop labels so
you know which helix is which. It has a simple mechanism to include
bump-outs and mismatches in the sequence.
In my work on lower metazoans, DCSE allows me to find good
alignments for several regions that Clustal mis-aligns. The
uncertainty of using secondary structure is when there are helices
or loops of different sizes between taxa, the sequences are very
different, and thus it can sometimes be difficult to determine
exactly how the corresponding helices and loops should be aligned,
expecially when there are few conserved sites in the helix or loop.
The advantage is that you always know where you are within the model
and it is unlikely that a large region will be improperly aligned
The biggest problem with DCSE is that there is a fairly steep
learning curve, but I think it is well worth it. I find that
the alignments result in trees with higher bootstrap values than a
Clustal alignment. However, I work with very diverse organisms, ie
entire metazoan phylogenies. When working with closely related
groups (ie within a slow evolving phylum), the advantage of secondary
structure based alignments may not be as great.
In summary, I agree that there are limitations to secondary structure
alignments, but at a practical level, I don't think it is due to
problems with the secondary structure models used (you can find the
matched stems fairly easily with DCSE), but more with deciding how to
align the more variable stems and loops among diverse taxa.
Jim Garey
Department of Biological Sciences
Duquesne University
Pittsburgh PA 15282
garey at next.duq.edu
> To: mol-evol at net.bio.net> From: mes at zoo.toronto.edu (Mark Siddall)
> Subject: Re: Alignment programs
> Date: Thu, 21 Dec 1995 05:33:57 GMT
> In article <carmean-1512951828090001 at berbee.botany.ubc.ca> carmean at sfu.ca (Dave Carmean) writes:
> >
> >I find it interesting that there is no discussion here of using homology
> >(such as secondary structure) to align sequences. Karl Kjer has been
> >working on this aspect of alignment.
>> I think that secondary structure is interesting.
> I am concerned though about the assumptions it makes.
> That is, much of the SSU rDNA work which purports to use
> secondary structure in assessing alignments fails to acknowledge
> that these are model-based and that there are vanishingly few
> models to work with.
> I mean is it reasonable to be using a murine model for ciliates?
> Moreover, there is the problem (like the Ribosomal Database
> Project) wherein you end up with an upsidedown pyramid of knowledge.
> So much becomes predicated on what preceded it. How would those homology
> statements look if the models and database began with anthozoans, or
> with trypanosomes for example?
>> Mark
> --
> 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
>>>
_____________________________________
Dr. James R. Garey
Department of Biological Sciences
Duquesne University
Pittsburgh, PA 15282
412-396-6322 (voice)
412-396-5907 (fax)
garey at next.duq.edu
