Is convergent evolution evidence against common descent?

Comparing whole genomes is a pretty major task, the sort of thing that takes research groups a year or more of work to do properly and several years of computing time.It might be better to identify some specific genes which we think might be relevant and try and find sequences for them amongst the selected groups. I fear this will prove difficult outside of well studied organisms or organisms whose genomes are being, or have been, sequenced.There are any number of well conserved genes involved in development which might effect morphology in the sort of ways we see in some convergent forms, the problem is pinpointing which ones.Most of the sequence data we have been looking at is totally divorced from the question of convergent evolution and more to do with the extenet to which molecular genetic phylogenies for mitochondrial sequences match up to the morphologically derived phylogenies, which is a relatively worthwhile exercise but seems to be a bit straying from the actual topic.One problem with your selection of organisms is that there is only moderately complete genomic data on 3 of them. You might be able to do whole genome comparisons of their mitochondria, the data is there for all of those species I think, but that sort of analysis requires considerably more sophistication than what we have been doing. You cant just plug 16kb sequences into ClustalW and have much faith in the results being meaningful.Having just typed that I was then seized by a compulsion to do just that, so treat the meaningful character of these results with caution. I wont post the sequences for this alignment, for obvious reasons, but the Genbank accessions are as follows :-Kangaroo - Y10524
Human - NC_001807
Mouse - NC_006914
Turtle (not snapping I'm afraid) - NC_000886
Opossum -NC_001610 You may notice that the pairwise alignments even of the same sequence only produce values of around 50%.The tree produced is like this :- Which looks like it conforms to our expectations, but as I say, treat this analysis with a degree of scepticism.TTFN,WK

Some research puts the estimate as low as 95%, if one takes indels into account (Britten, 2002). Another paper shows 98.6% sequence identity in the MHC/HLA locus between human's and chimps drops to as low as 86.7% when indels are taken into account (Anzai, et al., 2003)TTFN,WK

I suspect the problem is based on the method the CLustalW program uses to do its faster but less accurate method. I have rerun the data using the slower more accurate method and there were no self sequence comparisons and the similarity scores were considerably higher. I then bootstrapped the alignment and mapped the resulting tree getting a pretty robust result. Perhaps I could have taken some time to adjust the parameters for the quick alignment, different parameters produce more reliable results for longer sequences, and the 16,000 odd base pairs in the mitochondrial genome is orders of magnitude beyond the length of amino acid sequences we have been looking at previously.TTFN,WK

I used the web based ClustalW program to do the initial alignment. I then used a local version of ClustalX, which you can download, to perform the bootstrap analysis and then Treeview to to produce the tree with bootstrap values. If you are performing a bootstrap it helps to set the program to label the nodes rather than the branches, the plotting programs don't seem to see labelled branches for some reason.TTFN,WK