The reviewers noted that in the MorF pipeline we used the top hit to transfer functional annotation for each _Spongilla_ protein. They wondered if there was agreement in the functional annotation within between the morphologs ("no meaningful difference in score") of each protein. While we offered anecdotal evidence that the top hits all were within the same EggNOG orthogroup, we didn't back this up with extra analysis. This was rightfully brought up, and we are setting out to rectify it here.
Unfortunately this analysis is rather annoying to set up and execute; it requires us to define "top hits" in a meaningful way, something that's not always straightforward. Second, it requires us to get the full orthogroup assignments for a (very) large number of proteins, and, to the best of our knowledge, there is no programmatic way to get that information from the EggNOG database. The best workaround we found so far is to identify the proteins we want functional information for, retrieve their sequences from UniProt, submit these sequences to emapper, retrieve the results, and keep the best hit per sequence, which is going to be the identity match of the sequence to itself.
This is a lamentably circuitous way of getting information that, in some way, is definitely included in the EggNOG databases; unfortunately, we cannot think of a better way of accessing it.
After going through this, it is finally time to assess how well MorF does!
