AlphaFold

Most people have heard about chatGPT but not that many have heard about AlphaFold. But it is AlphaFold that is the most significant achievement of artificial intelligence (AI) by far that I have seen, and I have been following AI for more than 50 years. 

AlphaFold is considered by many to have solved the protein folding problem, or at least to have made very significant advances. How strings of amino acids fold to produce stable proteins has been an important question in biology for over 50 years. 

In September 2023, the Lasker Foundation announced that the winners of the Basic Medical Research Award were Demis Hassabis and John Jumper for the invention of Google DeepMind’s AlphaFold. In the past 20 years, 32 Lasker winners have received a Nobel prize. So an AI is in line for a Nobel for the first time.

Here’s a good video with dramatic animation that explains protein folding.
AlphaFold Solves Protein Folding

This is a more detailed explanation of protein folding, its implications and asks the question
Has Protein Folding Been Solved?

Watch this video for an exciting mini-documentary look behind the scenes of how the team at DeepMind raced to get results to submit to the CASP competition.
AlphaFold: The making of a scientific breakthrough

This article shows how AlphaFold is being used in several different research projects.
Case Studies Using AlphaFold

DeepMind published the method that AlphaFold uses, put the code in an open source repository and set up a database of 200 million structure predictions, amounting to nearly all the known human proteins.The program and the database have been shared with the scientific community and more than 300,000 investigators have used these resources. The catalog has been expanded to almost every known protein in organisms whose genomes have been sequenced, including viruses that pose epidemic threats and the World Health Organization’s high-priority pathogens.
Highly accurate protein structure prediction with AlphaFold

In only two years, the impact of DeepMind’s published manuscript has vaulted over almost all of the 100,000 research articles that have been published in Nature since 1900. It ranks 50^th, having been cited in more than 7000 papers from top journals.
Lasker Foundation

You can run AlphaFold on your own computer. Using Google Colaboratory, a cloud based service, you don’t need to have a very powerful machine yourself. Here’s an example of a researcher running AlphaFold to analyze a protein.
How to predict a large protein structure with AlphaFold