So, now that we can 'see' the 3D arrangement in a manner of speaking is there any way we could feed known protein structure in to neural networks by imaging them in quantity and take some of the sting out of protein folding by identifying likely candidate ways to do the folding ?

Or is that too big of a leap ?

It's a kind of the opposite approach: instead of slowly collecting information from extremely static scene they will blast the sample with very powerful X-ray laser, taking snapshot before shattering it to pieces.

http://www.xfel.eu/en/research-examples/structure-of-biomole...

Not to piss on a parade, but I don't see this as major breakthrough. It's a technical achievement with an established technology, of primary benefit to material scientists and nanotechnologists, both of whom already have imaging techniques at this level, but lack easy, cheap and instant ones, and face most of their hurdles at the design and synthesis side, not imaging/proof of construct

From a biochemist's point of view, if it was anywhere near real-time, THAT would be awesome.

What he said. This is amazing.

I guess the fact this book comes with pictures probably gives it an edge there.