Scientific Reputation

A recent editorial in PLoS Computational Biology on building and maintaining a good scientific reputation just came to my attention. Much of this is probably already known and practiced by most, but it never hurts to see it again and consider it closely.

Ascertainment bias data sets?

Hi folks,
I'm working on a simulation study looking at the effects of using ascertained SNP data for phylogenetic and phylogeographic reconstruction. I'm basing my simulations on real data sets. I have a few but am looking for more!
Have any of you seen studies recently that use single nucleotide loci that were selected based on being polymorphic is some ascertainment panel? Mostly SNP chip based data sets are what I was thinking of, but there are others too.
For example, Decker et al. 2009 used marker loci that are known to be polymorphic in cattle to reconstruct divergences up to 29 mya across bovidae.
SNP chip data are cheap and easy to get- I want to figure out how far you can push using them!
Send me an email! ejmctavish [at] utexas.edu
Thanks,
ejm


Fig. 1 from Decker et al 2009. Strict consensus cladogram (no branch lengths) of 17 most parsimonious trees based on 40,843 SNP genotypes. *, Denotes paraphyletic group.

2011 and admixture!

Hi all!
It has been a long hiatus, but I just heard a talk that I though you all would find interesting, and Evolution is coming up, so I thought I would post!
I am in Okinawa at OIST for a workshop on genomics, with a focus on linkage and recombination. Today we had talks from Gil McVean and Simon Meyers, both of which were exciting, but it was Simon's work that I found particularly applicable to questions I (we, perhaps?) are interested in. He was involved in the development of HAPMIX, described in Price et al. 2008, a program for inferring ancestry of segments of chromosomes in admixed populations. I am planning to apply it to my longhorn data set, and I'll let you know how it goes, but it appears to have some advantages over using site likelihoods in STRUCTURE to estimate ancestry of alleles. Downside might be that it could require more dense genomic data than I have.


He also presented as yet unpublished work (I think in collaboration with Daniel Falush and Garrett Hellenthal) on using inference of patterns of introgression between populations to estimate levels and times of admixture, without ever actually estimating the blocks. Not clear on exactly how it works... Because he is doing this on human populations these correlate with fascinating historical events. Using a 28 year estimate for generation time, he was able to place fairly precisely times of admixture events which were corroborated by known historical migrations. Reconstructing Spanish introgression into native North American populations in the 1700s isn't terribly exciting, but inferring European ancestry from 400 BC in the Kalash people of northern Pakistan, who have an oral history of being descended from the armies of Alexander the Great is pretty cool. I'm hoping I can apply these methods to my cattle SNP data. I have already been claiming that Moors brought African cattle to Spain, but having dates to back it up make it lot more plausible...
Anyhow- I think HAPMIX might be a really useful resource, whose existence had slipped past me, and I look forwar to this new method being published.



Hope to see some of you folks in Norman next month. I am really looking forward to the next-gen data in phylogeography symposium!