DeepMind's AlphaFold won the Nobel Prize for solving protein folding, a problem that stumped biologists for 50 years. The AI predicted structures for 200 million proteins and made them publicly available, accelerating drug discovery and disease research across the world.
Score of 90
About 30 minutes
From ~1 million to ~200 million structures
Dear fellow scholars, this is Two Minute Papers with Dr. Zsolnai. Today you will see history in the making. We are going to have a look at AlphaFold from DeepMind, perhaps one of the most important papers of the last few years. This is a true gift to humanity. And if even that is not good enough, it gets better. Because today you will see how, how this gift has now become 200 times bigger. So what is AlphaFold? AlphaFold is an AI that is capable of solving protein structure prediction, which we will refer to as protein folding. Okay, but what is a protein and why does it need folding? A protein is a string of amino acids. These are the building blocks of life. This is what goes in, which in reality has a 3D structure and that is protein folding. Letters go in, a 3D object comes out. This is hard, really hard. For instance, DeepMind's earlier AI programs were able to play hard games like chess Go and Starcraft 2. And none of these AIs were a match to protein folding. Yet AlphaFold has become the world's best solution at the CASP event. I've heard DeepMind CEO Demis Hassabis call it the Olympics of protein folding. If you look at how teams of scientists prepare for this event, you will probably agree that yes, this is indeed the Olympics of protein folding. And what is even more incredible is that it has reached a score of 90, which means two amazing things. One, in absolute terms, AlphaFold 2 is considered to be about three times better than previous solutions. The score of 90 means that we can consider protein folding as a mostly solved problem. Wow. This is not something that I thought I would be able to say in my lifetime. Now we said that this is a gift to humanity. Why is that? Where is the gift part here? What makes this a gift? Well, a little after publishing the paper, DeepMind made these 3D structure predictions available for free for everyone. For instance, they have made their human protein predictions public. Beyond that, they also made their predictions public for yeast, important pathogens, crop species and more. I also said that AlphaFold 2 is not the end, but a start of something great. And I promised to tell you how it improves over time. So here are two incredible revelations. One, we noted that AlphaFold will likely help us fight diseases and develop new vaccines. And hold on to your papers because this is not just talk, this is now a reality today. Look, this group was already able to accelerate their research work a great deal due to alphafold in fighting antibiotic resistant bacteria. They are studying how to fight deadly bacteria that modify their own membrane structure. So that the antibiotic can't get in. That is a hugely important project for all of us. They noted that they have been working vigorously on this project for 10 years and now that they have access to alphafold, they could get a prediction done in 30 minutes. And this is something that was not possible to do for 10 years before that. That is absolutely incredible. This is history in the making. I love to be able to live through the times like this. So thank you so much for sharing this amazing moment with me. So this was one project and what is this here? Oh yes, these are some more of these alphafold predictions that were already referenced from other scientific research papers. This is making big waves. And two, this gift to humanity has now become bigger. So much bigger. Initially, the protein prediction database they published contained 1 million entries. That is a lot of proteins. And now if you have been holding on to your papers, squeeze that paper because they have grown it from 1 million to 200 million structures. That is 200x within less than a year.
Wow.
So just imagine how, how many more of these amazing medicine research projects it will be able to accelerate. I cannot even fathom how many lives this AI project is going to save in the future. What a time to be alive. Here is the distribution of the data across different categories and these small circles represent what was available before and the big ones show what is available from now on. Absolutely amazing. It seems clearer and clearer that we are now entering the age of AI driven art creation, but it might be possible that we will also soon enter the age of AI driven medicine. These tools can help us so much. How cool is that? So does this get your mind going?
What would you use this for?
Let me know in the comments below.
This episode has been supported by Cohere AI. Cohere builds large language models and makes them available through an API so businesses can add advanced language understanding to their system or app quickly. With just one line of code, you can use your own data, whether it's text from customer service requests, legal contracts or social media posts, to create your own custom models to understand text or even even generated. For instance, it can be used to automatically determine whether your messages are about your business hours, returns or shipping. Or it can be used to generate a list of possible sentences you can use for your product descriptions.
Make sure to go to Cohere AI.
Papers or click the link in the video description and give it a try.
Today, it's super easy to use. Thanks for watching and for your generous support and I'll see you next time.
