Another way (influenced by some of Elchanan Mossel's work, though it also
relates to Cecile's idea of looking for saturation) is to look at
information about state at the root. If the data are very informative about
it, one state will have most of the relative likelihood. If the rate * time
is high
Hi Simon,
One option could be to look at the expected number of substitutions over
time t, and find the smallest time t for which you expect at least 1
substitution. The idea here is that the first substitution is the one
that most disrupts the signal.
Technically, if Q is your rate matrix and
Hi all,
What is the best way to convert an instantaneous transition rate (such as that
given by geiger's `fitDiscrete` method) into a measure of stability over time?
So, I have a set of traits with a small number of states. I want to fit these
onto a set of trees with branches proportional to
