What can be the reason of getting negative branches lengths after BEAST analysis?

[Pavel Rinkman]

I thought that BEAST does not allow negative lengths of branches on prior distribution level so I got very surprised when I have looked at my consensus tree in FigTree.

What could be the reason of this and how to wrestle with this phenomenon?

I am analysing about 1500 taxa of HAV (subtype: IA, region: VP1) under the next parameters: GTR+Г, Strict Clock, Coalescent constant population. Tree was generated by treeannotator through ~/beast/bin/treeannotator -heights median -burnin 20 -limit 0.5 VP1_test_bt_ExpPop.trees VP1_test_bt_ExpPop.tree

​

[Remco Bouckaert]

TreeAnnotator only guarantees positive branch lengths when annotating heights using the common ancestors approach (treeannotator -heights ca), which is the default. However, you used the median heights to annotate the tree, which sometimes causes negative branch lengths for clades that have few samples in the tree set. 

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/beast-users/7e426561-1986-4876-b4e2-734d870cd45dn%40googlegroups.com.

[Pavel Rinkman]

Thank you! It works. I found the solution a week ago but I hope your answer will be useful for potential readers of the group.

понедельник, 4 апреля 2022 г. в 02:44:24 UTC+3, higg...@gmail.com: