i've been doing biological/biomedical research for about 20 years now (god, that's a scary thought...)

I don't think that biology is well modeled by "being a meat computer", in general, but as far as the analogy goes ... :

What we are debugging is 4.5 billion years of grad student code in at least a few dozen languages, with no documentation, method names that are outright lies, no separation of concerns at all, the worst spaghetti you've ever seen, god-objects everywhere, no separated state at all, spooky action at a distance, the whole spiel.

We don't have a reliable debugger, you can't trace the stack because the stack changes when you look at it, and the entire thing is running all the time and mutating in a really nasty feedback loop with you.

Oh and there is no software, there's only FPGA-like firmware that reprograms itself all the time and can decide to add or remove logic units basically whenever, or wrap itself in duct tape and baling wire.