search

tgvaughan / MultiTypeTree

BEAST 2 package which provides support for multi-type trees: phylogenetic trees on structured populations.
http://tgvaughan.github.io/MultiTypeTree
GNU General Public License v3.0
21 stars 16 forks source link

Time-calibrated tree #20

Open amymariecampbell opened 2 years ago

amymariecampbell commented 2 years ago

Hi there,

I've been running a structured coalescent using the MultiTypeTree template, tip locations and dates etc.. All ESS values > 200, but when I make my final tree, the x axis looks strange (from 0.99ish to 1), and the node positioning isn't in time-order, if that makes sense.

How can I calibrate this so the node positioning corresponds to time, and have the real dates on the x axis? Not sure if I'm missing a step.

Thanks!

tgvaughan commented 2 years ago

Dear Amy,

That does indeed sound strange. You're not missing a step, although it sounds the tree portion of the state has either (a) not reached equilibrium (perhaps this posterior is difficult for the algorithm to sample) or (b) that there is some explicit or implicit constraint on the ages of the tree events.

How did you include the time information in the analysis in the first place?

Tim

amymariecampbell commented 1 year ago

Hi there,

I added in tip dates into the analysis. In some previous analysis where I was testing the workflow on ~10 samples, the nodes were in time order, and the heights were numbers that realistically corresponded to going back in years (e.g. 1-14) but for some reason it hasn't worked this time and I'm getting small decimal numbers. Posterior ESS is 239 currently.

Thank you Tim!

tgvaughan commented 1 year ago

Hi Amy, did the tip dates help?

amymariecampbell commented 1 year ago

No unfortunately not- though I've just realised the heights also look wrong (small decimals) in the untyped start tree I am providing in the xml- so perhaps this could be the source of the problem instead? I will go and amend this and see if it helps!