But nobody thinks there are no constraints operating on functional proteins, so the mere reality of constraint is an uninteresting factoid that does not establish how much, or that no function exists elsewhere in sequence space. Even in 400 million years, an infinitesimal fraction of that space has been sampled.
So we are still left with Gpuccio’s 400 million-year-rule being completely unable to establish any basis for calculating the FI in a known protein. And his “explanation” for why we must wait 400 million years specifically, is completely useless as it says nothing about what the rest of sequence space is like. All it says is it is likely that most codons in the protein have mutated at least once. Which doesn’t tell us the difference between deleterious or nonfunctional, and says nothing about whether there are other functional sequences further away which are very dissimilar to the one that mutates.
As I explained in the other thread, even for a 150 aa protein we are left with having sampled an almost infinitesmial fraction of sequence space. For a huge protein like Prpf8 it is much worse by many orders of magnitude.