How would I assess evolutionary estimates between duplicated vs. non-duplicated genes in PAML?

Question

[codoncity]

Hello,

I have been investigating a four-fold gene duplication in a species. All closely related species only possess a single copy of this same gene. How would I check to see if rates of selection for duplicate copies differ when compared with the single copies found in other species?

I feel like partitioning my tree into two groups (e.g. four duplicate sequences vs. all singles) and then comparing them with a LRT stat wouldn't be valid since PAML would apply different rates of evolution to each group. So I'm a bit stumped.

Anyone have any experience with this?

Cheers.