Interpretation of results "Gamma_family_likelihoods.txt"

[EmilianoMora]

Hi CAFE5 team,

Thank you for developing this tool! I am running some analyses I am struggling to understand if there is meaningful (biologically?) I can extract from the 'Gamma_family_likelihoods.txt' result file.

Lets say I am using four gamma categories (k=4) and the results look like this below:

#FamilyID Gamma Cat Mean Likelihood of Category Likelihood of Family Posterior Probability Significant
OG0000022 0.213631 1.38969e-93 3.43299e-41 4.04803e-53 N/S
OG0000022 0.59063 1.01484e-64 3.43299e-41 2.95614e-24 N/S
OG0000022 1.09484 3.13438e-51 3.43299e-41 9.13018e-11 N/S
OG0000022 2.1009 3.43299e-41 3.43299e-41 1 *

OG0000030 0.213631 5.45608e-140 6.3383e-65 8.60811e-76 N/S
OG0000030 0.59063 9.75198e-101 6.3383e-65 1.53858e-36 N/S
OG0000030 1.09484 3.77645e-81 6.3383e-65 5.95815e-17 N/S
OG0000030 2.1009 6.3383e-65 6.3383e-65 1 *

OG0000037 0.213631 1.93174e-78 1.40023e-37 1.37958e-41 N/S
OG0000037 0.59063 1.44059e-55 1.40023e-37 1.02882e-18 N/S
OG0000037 1.09484 2.8451e-45 1.40023e-37 2.03188e-08 N/S
OG0000037 2.1009 1.40023e-37 1.40023e-37 1 *

In all three OGs it seems like the likelihood of the Gamma category 2.1009 is significantly different from the 'Likelihood of Family'. But what does that mean?

In the Mendes et al paper (2020) it says that "CAFE 5 then uses an empirical Bayes approach to estimate the posterior probability of a family belonging to a rate category, which in turn enables down-stream analyses of ‘slow’ or ‘fast’ families." ... Is this result file related to that statement? And if so, is a higher Gamma category (e.g., 2.1009) associated with a faster evolution in this family when compared families that are significant for a lower Gamma category (e.g., 0.213631)?

Thank you in advance!

Cheers,
Emiliano

[hahnlab-user]

Hello,

Sorry for not answering this sooner--it slipped through the cracks.

I think that the formatting is making it a bit hard to see which number lines up with which of the six columns. The second column represents "Gamma cat mean," which is the mean rate for that gamma category (=2.1 for the fourth category). Since categories are fit across all families, these mean rates are the same for each family (with mean rates increasing across the four families).

And in each of these families, the likelihood of the family is essentially the same as the likelihood of this fourth rate category because they really do not fit in any of the other categories.

Hope that helps,
Matt