stephaneguindon
commented
1 month ago Dear Sishuo, I did not design the EX2 model and the best would be to go back to the original article describing it. But from what I can see, two rate matrices are provided (one built from the analysis of buried sites and one from exposed), so that the ratio of 2.44 you are mentioning here is already built into them (i.e., this ratio is not an explicit parameter which value is fixed to 2.44). But you are correct that, besides edge lengths, one only estimates a single parameter here, Pc. As for EX3, I am not quite sure I understand what this model corresponds to (a mixture of "buried" + "exposed" + "what else?") so I'm afraid I can't really help here. Best regards, -Stéphane-
evolbeginner
Hi Stéphan,
First I'd like to thank you for your great efforts in developing and maintaining phyml. Could i ask a simple question about the EX2 mixture model? In brief, i'd like to confirm with you if there is a single parameter to estimate for EX2.
I see EX2 is implemented in both phyml and iqtree. In iqtree, it's given in the following format:
In atgc, it's given at http://www.atgc-montpellier.fr/phyml/structure/data/EX.htm
As i understand, the ratio of the relative rate of the above 2 mixtures is fixed as
r_2/r_1 = 1.6413466609931/0.672020808818762 = 1.360/0.557 = 2.44where r_1 and r_2 indicate the relative rate of Bur and Exp respectively. If I am not mistaken, then we have$$ w_1 + w_2 = 1 $$
$$ w_1 \cdot r_1 + w_2 \cdot r_2 = 1 $$
$$ \frac{r_2}{r_1} = 2.44 $$
So, 4 unknowns with 3 equations, thus one to estimate. The above is consistent with the # of parameters = branch_# + 1. Similarly, for EX3 there are 6-4=2 free params.
Would you pls mind letting me know if the above makes sense? Any help is highly appreciated. Many thanks!
best, sishuo