So if I understand this correctly the most promising angle to neurodegenerative diseses is proteine folding.

So if I understand this correctly the most promising angle to neurodegenerative diseses is proteine folding. Something goes wrong with that. Imagine tying your kids shoe-laces: It is the same knot, but you do it the other way round as wou would on yourself. Something simmilar happens in a patients brain while folding proteines.

So what we need to stop that is some kind of enzyme / katalysator that either breaks and refolds these or prevents the wrong kind of folding in the first way. Such an enzyme would be a pharmacological intervention that could hypothetically stop or at least slow donw the progress of these diseases. If diagnosed early e.g. by genetic screening, these diseases could become managable. Even within our lifetime.

In both of these fields - the topology of proteine folding and functional mapping of the human genome - machine learning has drasticly accelerated research within the last years/months, and it seems not entirely out of scope anymore to get to a point where actual treatments might get developed and applied to real patients.

This is so exiting. What a time to be alive! Let's do this!