Hi Both
a further comment on the porcelain crab analysis.
The EST dataset is stunning, and looks really useful for lots of
things, but the claim that the data as presented support one tree over
another is in fact not there.
The claim is based on a pair of venn diagrams (the figure 5 in the
paper) that show the numbers of EST clusters or genes match between
the porcelain crab (Pc), Daphnia pulex (Dp), and either Drosophila
melanogaster (Dm) or Apis mellifera (Am). The venn diagram shows
clearly that more Dp genes match Dm or Am than do Pc EST clusters.
Hence Dp must be closer to Hexapoda.
However, like is not being compared with like. The Dp, Dm and Ap
datasets are hopefully-near-complete proteomes from the respective
genomes, while the Pc data are a partial transcriptome. Thus a Dp gene
that matches Dm but doesnt match Pc could simply be due to missing
data in the Pc dataset.
The correct comparison is to look at the number of Pc genes that match
ONLY Dp versus ONLY Dm, or ONLY Dp versus ONLY Am. These genes are
presumed to be present in the last common ancestor of the species
compared, and the tree that implies the fewest gene losses is arguably
the better tree. The data are:
Pc clusters in Dahnia pulex but NOT in Dm or Am: 2449
Pc clusters NOT in Daphnia pulex but in Apis: 2811
Pc clusters NOT in Daphnia pulex but in Dm: 2071
So the comparison to Drosophila places Pc closer to Dm than Dp
(2811>2449): Pc is closer to Hexapoda
But he comparison to Apis places Pc closer to Dm than Dp (2071<2449):
Daphnia is closer to Hexapoda
So, actually the data as presented are entirely equivocal, but since
2811>2071 could be used to argue the opposite of the authors
conclusions (I wouldnt recommend doing so).
As I have posted earlier, BLAST is not a good phylogenetic tool, and
phylogenies should be addressed using model-driven phylogenetic
methods. In addition it is very dangerous to use presence-absence data
such as these in cases where there are incomplete data (such as the Pc
'partial genome') or there is an ascertainment bias of any kind (eg:
were not many of the Apis and Daphnia gene models inferred partly on
the basis of the wonderfully dense annotation of the Drosophila
genomes? So isnt there a circularity: Daphnia looks more like
Drosophila because Daphnia was modeled on Drosophila).
In this comment I do not intend to make any claims as to the truth or
falsity of the hypothesis of relationships within Pancrustacea, just
to say that the data in the Pc paper do not address it cogently...
they could though!
Mark
On 19 Feb 2010, at 21:24, Don Gilbert wrote:
>> Darren,
>> This Porcelain crab transcriptome is compared to only Daphnia and
> two insects
> (in the figure I looked at), so it is much narrower in species scope
> than that
> recent Arthropod phylogeny paper. On the other hand it is comparing
> 19,000 EST
> clusters (~genes), whereas the Nature paper compared only ~60
> genes. My take,
> and I'm not a phylogenist, is these are both accurate
> interpretations of their
> data, but neither is a complete or final determination of the
> phylogenies.
> We need more data for that :)
>> I just took another look at the phylo tree in Regier et al Nature
> paper and
> it seems in agreement with malacostracan (crab) being more distant
> from
> insects than branchiopoda (daphnia).
>> -- Don
>> |From dobbard from staffmail.ed.ac.uk Fri Feb 19 16:01:30 2010
> |Date: Fri, 19 Feb 2010 21:01:14 +0000
> |From: Darren Obbard <dobbard from staffmail.ed.ac.uk>
> |To: Don Gilbert <gilbertd from bio.indiana.edu>
> |Subject: Re: [Arthropod] New crustacean gene set: Porcelain crab
> | transcriptome paper
> |
> |> Among other results here, they support the pancrustacea
> hypothesis and
> |> suggest that branchiopod crustaceans (daphnia) may be more
> closely related
> |> to insects than other malacostracan crustaceans (por. crab).
> |
> |Does this mean we don't believe "Xenocarida as the long-sought sister
> |group to the Hexapoda"?
> |http://www.nature.com/nature/journal/vaop/ncurrent/pdf/> nature08742.pdf
> |
> |
> |
> |--
> |Darren Obbard
> |Institute of Evolutionary Biology
> |Ashworth Labs
> |Kings Buildings
> |University of Edinburgh, UK
>> _______________________________________________
> Arthropod mailing list
>Arthropod from net.bio.net>http://www.bio.net/biomail/listinfo/arthropod>
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
