Dear Jose,

In the GUI, base frequencies are estimated for the partition (not the whole dataset). You can force them to be the same for all partitions, by constraining frequency parameters from different partitions.

To estimate base frequencies by ML in the GUI, you need to parameterize the equilibrium frequencies vector for the model. For example, this will define a nucleotide model with frequencies estimated by ML:

Code:

/* the following is simply a way to parameterize
    base frequencies so that they are all in [0,1]
    and add up to 1, i.e. with 3 free parameters */

global P_1 = 0.5; P_1 :< 1;
global P_2 = 1/3; P_2 :< 1;
global P_3 = 1/4; P_3 :< 1;

global pi_A := P_1;
global pi_C := (1-P_1) * P_2;
global pi_G := (1-P_1) * (1-P_2) * P_3;
global pi_T := (1-P_1) * (1-P_2) * (1-P_3);

estimatedFreqs = {{pi_A}{pi_C}{pi_G}{pi_T}}; /* the vector of frequencies */

define the rate matrix here...

Model myModel = (myRateMatrix, estimatedFreqs);

Tree ...

LikelihoodFunction ....

 

HTH,
Sergei