With super-resolution methods, it became possible to circumvent Abbe's resolution limit for light microscopes. At the same time, other parts of the electromagnetic spectrum were being used, with radio waves for NMR and X-rays for crystallography. In addition to photons, electrons and neutrons were also made usable for "seeing". Although the different methods measure different properties of the molecules, the information obtained is surprisingly similar. However, "seeing" has become indirect because diffraction or reconstruction from constraints replaces direct imaging.
In principle, protein structure is determined by the amino acid sequence. Most recently, AlphaFold was presented as a method for predicting structures far exceeding other competing methods. It is based on deep neural networks that could be trained using the huge data set of experimental, related protein structures. AlphaFold marks a great leap forward in structural biology and biology in general. However, a multitude of unsolved problems await, in particular those related to pharmaceutical research and drug design. The importance of structural information there cannot be overestimated and has led to the establishment of special departments in Big Pharma and the founding of focused companies (see http://www.proteros.com; http://www.astx.com)