Protein folding

Proteins are fascinating. They are long and complex molecules that perform numerous tasks in the body like building tissue/muscles and fighting disease. Their purpose is dictated by their structure – which folds like Origami into complex and irregular shapes.

Understanding protein structure would enable researchers to synthesize proteins that can do all sorts of tasks. We could design better medicines, create virus-resistant crops, or break down waste.

But, protein’s complexity made it very hard for researchers to understand a protein’s shape. Determining a protein’s structure sometimes was the fruit of an entire researcher’s P.hD. If they got lucky. So, after decades of work on protein structure, we only had 180,000 protein structures in the public domain.

Then, Deepmind announced that their AI was able to product accurate predictions of DNA structure. They then released 350,000 protein structures across 20 organisms – for free. And, the kicker was that these structures include predictions for 98% of all human proteins.

That. is. insane.

This is such a giant step that it will take a few years for researchers to make sense of this data and figure out practical uses. But, make no mistake, this is a massive leap.

In ten years, we’re going to look back at two leaps over these two years – with the coming of mRNA and the advent of protein folding – as leaps that changed medicine.

I loved this note from Deepmind’s CEO – “From the beginning, this is what we set out to do: to make breakthroughs in AI, test that on games like Go and Atari, [and] apply that to real-world problems, to see if we can accelerate scientific breakthroughs and use those to benefit humanity.”

Well played.

H/T: The Verge, The Wired