stamatak
commented
3 years ago Two different trees having the same likelihood can not happen theoretically as the likelihood model is generally identifiable, with the exception of the circumstances under which phylogenetic terraces can occur ... so I would say that we can safely assume that our model here is identifiable,
alexis
On 12.11.20 14:37, Sarah Lutteropp wrote:
Regarding the red comment in this formula:
Screenshot from 2020-11-12 13-31-08 https://user-images.githubusercontent.com/1059869/98940482-582a1700-24eb-11eb-8130-017d56455be3.png
Do I understand correctly that the re-estimation of reticulation probabilities should be changed to this (see below)? Screenshot from 2020-11-12 13-32-56 https://user-images.githubusercontent.com/1059869/98940654-9aebef00-24eb-11eb-9b06-0c689aea1ff9.png
The change in words: Instead of having a binary value stating whether a reticulation is taken in a single best displayed tree for a given site, have a value stating in which fraction of best displayed trees for a given site the reticulation has been taken. The rationale for this is that the rare case could happen that there are multiple displayed trees with exactly the same loglikelihood.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lutteropp/NetRAX/issues/4, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAGXB6XBWRLHQV36VHPLOFLSPPJH3ANCNFSM4TTHN2NQ.
-- Alexandros (Alexis) Stamatakis
Research Group Leader, Heidelberg Institute for Theoretical Studies Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
www.exelixis-lab.org
lutteropp
Regarding the red comment in this formula:
Do I understand correctly that the re-estimation of reticulation probabilities should be changed to this (see below)?
The change in words: Instead of having a binary value stating whether a reticulation is taken in a single best displayed tree for a given site, have a value stating in which fraction of best displayed trees for a given site the reticulation has been taken. The rationale for this is that a rare case could happen where there are multiple displayed trees with exactly the same loglikelihood.