DeepMind's AlphaFold 2: Solving Protein Folding and Its Profound Impact on AI and Life Sciences

This podcast episode celebrates DeepMind's AlphaFold 2 breakthrough, which has "solved" the 50-year-old grand challenge of protein folding by achieving prediction performance comparable to much slower and more expensive experimental methods like X-ray crystallography. The speaker positions this as one of the biggest advancements in structural biology of the past two decades and a monumental leap in artificial intelligence, comparing its significance to milestones like ImageNet/AlexNet, AlphaZero, and GPT-3, but with a more direct real-world impact.

The discussion delves into the biological significance of proteins as the workhorses of living organisms, explaining that their 3D structure, determined by the amino acid sequence, dictates their function. A key nuance highlighted is the "weirdness" of protein folding's uniqueness and the combinatorial complexity (e.g., 10^143 possible folds compared to 10^100 chess games), which makes the problem incredibly difficult. The speaker emphasizes that protein misfolding is the underlying cause of many diseases, underscoring the critical importance of accurately predicting protein structures.

Practically, AlphaFold 2 is expected to revolutionize structural biology by vastly increasing the availability of 3D protein structures, which are currently limited by the high cost ($120,000 per protein) and time (one year per protein) of experimental determination. This acceleration will enable scientists to uncover unknown functions of genes, understand and treat diseases caused by misfolded proteins, and design novel proteins for therapeutic applications, agriculture (e.g., insecticidal or frost-protective proteins), tissue regeneration, health supplements, and advanced biomaterials.

Looking ahead, the breakthrough has profound broader implications, including the potential for multiple Nobel Prizes for derivative work, possibly even for discoveries heavily reliant on machine learning systems. It represents a significant step towards advancing end-to-end learning for increasingly complex biological problems, such as multi-protein interactions and physics-based simulations of entire biological systems like cells or organs. Ultimately, AlphaFold 2 pushes AI beyond game-playing into real-world scientific discovery, inspiring future advancements in both biological and artificial life.

"deepmind has announced that its second iteration of the alphavote system has quote unquote solved the 50 year old grand challenge problem of protein folding"

"solved here means that these computational methods were able to achieve prediction performance similar to much slower much more expensive experimental methods like x-ray crystallography"

"this is one of the biggest advancements in structural biology of the past one or two decades and in my field of artificial intelligence i think a strong case could be made that this is one of the biggest advancements in recent history of the field"

"the good old argument over beers about uh which is the biggest breakthrough comes down to the importance you place on how much real world direct impact a breakthrough has"

"my prediction is that there will be at least one potentially several nobel prizes that will result in derivative work launched directly with these computational methods"

"the 3d structure determines the function of the protein so one of the correlators of that is that the underlying cause of many diseases is the misfolding of proteins"

"the protein folding problem just in the number of possible combinations is much much harder than the game of chess but it's also much weirder"

"if you involve attention if you evolve transformers you're going to get a big boost and the other lesson is that if you make as much of the problem learnable as possible you're often going to see quite significant benefits"

"because the protein structure gives us the protein function figuring out the structure for maybe millions of proteins might allow us to learn unknown functions of genes encoded in dna"

"taking a step even further is this is physics biophysics so being able to accurately do physics-based simulation of biological systems"