Before anyone had seen a protein, people had already drawn one. The drawings came first because they had to: a representation is a bet about which parts of a thing matter, and you can only place that bet by drawing it and seeing what the drawing lets you do.
Kendrew’s 1958 model of myoglobin was a lumpy sausage built from the first low-resolution map. It looked like nothing in particular, and that was the finding: the chain folded in a way no one had expected, and no symmetry was going to rescue it.
Each later representation was a new bet. Ball-and-stick models made bonds the subject. Space-filling models made surfaces the subject, and bonds disappeared. Ribbon diagrams, introduced by Jane Richardson in 1981, made the fold the subject, and almost everything else went quiet.
Try it. The same chain, drawn four ways:
Drag the slider: atoms, bonds, ribbon, surface. Each drawing answers a different question.
Notice what each drawing makes easy to ask, and what it makes impossible to see. That asymmetry is not a flaw of the drawing. It is the whole reason to have one.
Models shape the work that shapes the models
Once a representation is standard it becomes the thing new work has to be expressed in, which is how a drawing ends up steering what gets measured next.