Running simulation to choose best model

[timeforthemoonnight]

I ran the analysis with separated lambdas and now i want to see which model fits better (separeted or global), and i understand i must do it by running the simulation command. I have done the previous steps for it, like estimating the global lambda with the error model, but when i try to run the simulation i get the following error:

ommand line: cafe5 -i hog_gene_counts.tsv -t SpeciesTree_rooted_ultra.txt -eBase_error_model.txt -y SpeciesTree_separated.txt -s800 -l 0.39675575354218 -o simulation

Filtering families not present at the root from: 19861 to 8780

No root family size distribution specified, using uniform distribution
Simulating with 1 model(s)
Simulating 100 families for model Base
Failed to create a family that would exist at the root

Trying to simulate leaf family size that was not included in error model

What am i doing wrong ?
I tried a few different values in the -s parameter, but it does not change the anything
I really appreciate if anyone could help me, thanks !!