I wonder if anyone has tried making a non-trivial protein synthetically (that does not occur in nature) and checked the structure against AI models. Why? Becaus

I wonder if anyone has tried making a non-trivial protein synthetically (that does not occur in nature) and checked the structure against AI models. Why? Because the model may "accidentally" utilise a particular folding mechanism seen in Nature that would not be found in a synthetic approach. To what end? To discover those unappreciated mechanisms, but also, in future the opposite may be possible "I want this structure: show me the amino acid sequence to get it". For instance a completely nonbiogenic Covid spike.