Enquiry on <twoEpoch> function

[Jiehong Wei]

Dear all,

I am running 2-epoch model to estimate mutation rate following expansion dating method of Crandall et al. (2014). The 2-epoch model assumes two demographic process, constant population in the past and a present lognormal population expansion. However, I found that the starting tree could not be generated from 2-epoch model. I am now using a constant population size model to generate starting tree, and estimate the coalescent likelihood using 2epoch model. I notice that the starting tree block might be associated with the treemodel block, as the block have referred startingTree.

      

  <treeModel id="treeModel">
                <coalescentTree idref="startingTree"/>
                <rootHeight>
                        <parameter id="treeModel.rootHeight"/>
                </rootHeight>
                <nodeHeights internalNodes="true">
                        <parameter id="treeModel.internalNodeHeights"/>
                </nodeHeights>
                <nodeHeights internalNodes="true" rootNode="true">
                        <parameter id="treeModel.allInternalNodeHeights"/>
                </nodeHeights>
        </treeModel>

Although the mcmc run completed smoothly, the ESS value of coalescence likelihood is only 5. I wonder if this was caused by estimating 2epoch coalescent likelihood from a constant population size tree. Did I have the correct setting?