Mb Notes

This page is to record notes on usage, quirks, misunderstandings etc. I have noticed with MrBayes 3.2.5 analyses. Mostly for my own memory. YMMV (your mileage may vary) by setup, dataset, MrBayes version, etc. And the date you read this (I am starting summer 2015).

State frequencies, statefreqs vs. statefreqpr

The 3.2.5 manual says this should work:

prset applyto=(all) statefreqs=fixed(equal);

…but actually it needs to be:

prset applyto=(all) statefreqpr=fixed(equal);

Even then, I see this in the screen output:

Data not partitioned --
            Datatype  = Standard
            Coding    = Variable
            # States  = Variable, up to 10
                        State frequencies are fixed to be equal
            Rates     = Gamma
                        The distribution is approximated using 4 categories.
                        Likelihood summarized over all rate categories in each generation.
                        Shape parameter is exponentially
                        distributed with parameter (1.00).

      Active parameters: 

         Parameters
         ---------------------
         Statefreq           1
         Shape               2
         Ratemultiplier      3
         Topology            4
         Brlens              5
         Speciationrate      6
         Extinctionrate      7
         Fossilizationrate   8
         Igrvar              9
         Igrbranchrates     10
         Clockrate          11
         ---------------------

         1 --  Parameter  = Alpha_symdir
               Type       = Symmetric diricihlet/beta distribution alpha_i parameter
               Prior      = Symmetric dirichlet with fixed(-1.00) variance parameter

The top bit suggests frequencies have been set to equal, the bottom suggests there is still a symmetric dirichlet prior on state frequencies.