Some formulations of protein folding are NP complete. That doesn't mean those models always correspond to reality. There's no evidence that nature is solving an NP hard problem when folding proteins (See the paper NP complete problems and physical reality by Scott Aaronson)