Crazy busy week
Oh man, this week has been too busy even for my taste.
I really was supposed to be working on a project that I am going to London to finish in a week, but a genome project got in the way.
We are involved in a few genome sequencing and analysis project at BiRC where we use our CoalHMM methodology to infer speciation times and ancestral effective population sizes.
The method uses changes in genealogy topology to infer this, but essentially ignores the variation in coalescence times along the genomes and just uses the mean sequence divergence.
Back when I was writing about how the coalescence process leads to such variation in divergence, see:
- On segment lengths, going back in time, in the coalescence process. Part 1: The ancestry of a single species
- On segment lengths, going back in time, in the coalescence process. Part 2: The ancestry of two species
I also developed an approximation to the coalescence with recombination process that lets us estimate the distribution of coalescence times and fragment lengths. I'll write about how it works later, I promise, but it takes a bit explaining so I need to find a day where I'm not over worked, so it will be some upcoming vacation or something...
Anyway, because we can compute these distributions we can also construct a hidden Markov model that uses them in its transition matrix, and over Easter, Julien implemented this HMM. Our initial simulation-based validation is very promising, so now we are analysing a few genomes.
The method is relatively fast, so we can analyse a genome in about a CPU week on our Linux grid. Still, because we are adding a new method and completely new results very late in a project, we need to move fast to get the results into the paper. Since I am away in the UK at the end of the month, and Julien is away for a week of well deserved holiday and then job hunting in France, we need some convinsing results this week already. Thus a very busy week.
That is why the blog has been very quiet lately. No excuse, of course, but an explanation.
Ok, off I go again. I need to finish making slides for a talk I'm giving this afternoon (on genomic analysis of abes and the issues mentioned above...)