Phylogenies from DArTs: A stochastic Dollo proces with censored data
Seminar Room 1, Newton Institute
AbstractDiversity Array Technologies (DArT) are a relatively new kind of DNA marker system that seem like they could be usefully applied to phylogenetics (a few papers have already explored this). Like marker systems such AFLP and RFLP, the method produces presence absence data, but unlike these methods it is very unlikely for shared presences to occur by chance. The basic idea is as follows. One or a small number of genomes are selected to form the genomic representation. Two enzymes are used to cut the DNA from these genomes at certain recognition sites (a rare 6bp recognition site and a more frequent 4bp recognition site). Fragments of DNA whose ends have been cut by two rare recognition sites are amplified. These fragments, which are said to form the genomic representation, are arranged on a microchip. Other genomes can then been checked to see which fragments within the genomic representation they have copies of in their own sequence. For each other genome that is compared to the genomic representation this results in a binary sequence that indicates presence (1) or absence (0) of each of the fragments. The first obvious advantage of this approach is that it creates a representation of the whole genome rather than just a few genes. This alleviates the problem of picking a small set of genes that may not be representative of the evolutionary history of the species. The second advantage is that in comparison to an individual site, long fragments of DNA are very unlikely to be similar due to chance. So if two species share a fragment it is vastly more likely that they share it due to common ancestry rather than due to a chance similarity. To use these data for phylogenetics it would be useful to develop a likelihood equivalent of Dollo parsimony (in which characters can be lost multiple times but gained only once), such models have already been explored in the context of language evolution and gene content evolution. However, another complicating issue is the censoring effect created by only being able to see those fragments that were in the original genomic representation, i.e. fragments that are shared by a group of species but that are not present in the original species used to make the genomic representation are missing from the data.
If it doesn't, something may have gone wrong with our embedded player.
We'll get it fixed as soon as possible.