search

popgenmethods / smcpp

SMC++ infers population history from whole-genome sequence data.
GNU General Public License v3.0
151 stars 34 forks source link

Incoorporating aDNA #7

Closed harlydurbin closed 7 years ago

harlydurbin commented 7 years ago

I have a question about running estimate and plot with ancient samples alongside extant populations... In the paper, how did you go about plotting your ancient individual without extending the population estimate to present day? We have a 14kya sample we're trying to incorporate and would like to use to run 'split' in order to look at speciation/divergence times. Besides altering the mutation rate, are there any other considerations we should take into account?

terhorst commented 7 years ago

There is an option --offsets to plot which you can use to shift population(s) back in time, like for aDNA analysis. Unfortunately it is not yet able to handle a two-population model where one of the populations is ancient and the other is not. Until I have time to add that feature, you can use the -c option of plot to generate a CSV containing the raw data points of the plot, and then shift your aDNA lineage back in time manually.

When analyzing the aDNA sample in our paper, it helped to regularize a bit more strongly than for the modern populations. Also, if (as is usually the case) you have only a single sample from your aDNA population, I think the estimates from split could be noisier compared to analyzing hundreds of samples from each subpopulation. (We're currently working on an extension to SMC++ that is more suited to analyzing pairs of populations by allowing for migration between them, but it's a ways off.) Thanks.