Hello,

So I got the development version of MrBayes to try the now available LG model. I also have the analysis on the same analysis with the next best model (Blosum62) in the mrbayes-3.2.1 version. I ran both for 5000000 gen, sample frequency 250. I was wondering how would I judge the quality of the run? The LG model finished with an average standard deviation of split frequencies 0.008475. The Blosum62 finished with an average standard deviation of split frequencies: 0.007811. But is judging just on these enough? Prottest qualified LG as a better model.

Here are analysis from Tracer from the Blosum62 model:

For the LG model:

Thank you for your time,

Adrian