Mind Matters Natural and Artificial Intelligence News and Analysis
Human prion (3d model). Prion is an infectious agent that can fo
Human prion (3d model). Prion is an infectious agent that can fo

AlphaFold Scores Huge Breakthrough in Analyzing Causes of Disease

In a world so deeply designed and complexly organized, we need a quick and practical way of knowing what was going on in cells and viruses. AI can help

Alphabet’s DeepMind team has just scored a breakthrough in finding treatments for diseases. Their latest AlphaFold system won a grand challenge in analyzing the “folds” of proteins.

Proteins—large and often very complex chains of amino acids—do the work in our cells. But, like all bodies, they are three-dimensional. We can’t understand them until we can analyze the folds (the third dimension) that are unique to each type among hundreds of thousands. Knowing what a given protein actually does (or doesn’t) is critical to developing many new medical treatments.

How hard is the problem?

In his acceptance speech for the 1972 Nobel Prize in Chemistry, Christian Anfinsen famously postulated that, in theory, a protein’s amino acid sequence should fully determine its structure. This hypothesis sparked a five decade quest to be able to computationally predict a protein’s 3D structure based solely on its 1D amino acid sequence as a complementary alternative to these expensive and time consuming experimental methods. A major challenge, however, is that the number of ways a protein could theoretically fold before settling into its final 3D structure is astronomical. In 1969 Cyrus Levinthal noted that it would take longer than the age of the known universe to enumerate all possible configurations of a typical protein by brute force calculation – Levinthal estimated 10^300 possible conformations for a typical protein. Yet in nature, proteins fold spontaneously, some within milliseconds – a dichotomy sometimes referred to as Levinthal’s paradox.

The AlphaFold Team, “The ‘protein folding problem’” at DeepMind

That’s why it has been called as 50-year challenge. The “gold standard” technique, X-ray crystallography, is slow and expensive and diseases don’t wait for health care professionals to catch up with them.

Every two years since 1994, a contest (CASP) has been held, to predict a protein’s DNA sequence to determine its three-dimensional shape. AlphaFold did so “within about an atom’s width of accuracy in two-thirds of cases and was highly accurate in most of the remaining one-third of cases” (Slashdot), beating out a hundred competitors.

The more we know about proteins, the more easily we can develop medicines to target the ones that are performing badly and causing disease:

Janet Thornton, an expert in protein structure and former director of the European Molecular Biology Laboratory’s European Bioinformatics Institute, said that DeepMind’s breakthrough opened up the way to mapping the entire “human proteome”—the set of all proteins found within the human body. Currently, only about a quarter of human proteins have been used as targets for medicines, she said. Now, many more proteins could be targeted, creating a huge opportunity to invent new medicines.

Jeremy Kahn, “In a major scientific breakthrough, A.I. predicts the exact shape of proteins” at Fortune

Already, the AlphaFold technology has proven useful in a swift response to COVID-19:

Earlier this year, we predicted several protein structures of the SARS-CoV-2 virus, including ORF3a, whose structures were previously unknown. At CASP14, we predicted the structure of another coronavirus protein, ORF8. Impressively quick work by experimentalists has now confirmed the structures of both ORF3a and ORF8. Despite their challenging nature and having very few related sequences, we achieved a high degree of accuracy on both of our predictions when compared to their experimentally determined structures.

The AlphaFold Team, “The ‘protein folding problem’” at DeepMind

Many in the science media hope for great things:

AlphaFold is unlikely to shutter labs, such as Brohawn’s, that use experimental methods to solve protein structures. But it could mean that lower-quality and easier-to-collect experimental data would be all that’s needed to get a good structure. Some applications, such as the evolutionary analysis of proteins, are set to flourish because the tsunami of available genomic data might now be reliably translated into structures. “This is going to empower a new generation of molecular biologists to ask more advanced questions,” says Lupas. “It’s going to require more thinking and less pipetting.”

Ewen Callaway, “‘It will change everything’: DeepMind’s AI makes gigantic leap in solving protein structures” at Nature (November 30, 2020)

Of course, this development has left industry insiders wondering about their careers. One of them muses on AlphaFold’s earlier (not so decisive) triumph in the 2018 contest:

I think many of us went through the following phases: (i) fearing that the DeepMind team outsmarted us all by some brilliant fundamental insight, combined with virtuoso engineering; (ii) breathing a sigh of relief that the insights were not radically different from what most of the field was thinking; (iii) (slightly) belittling DeepMind’s contribution by noting its seeming incrementality and crediting their success to Alphabet’s resources.

Setting aside the validity of the above sentiments, the underlying concern behind them is whether protein structure prediction as an academic field has a future, or whether like many parts of machine learning, the best research will from here on out get done in industrial labs, with mere breadcrumbs left for academic groups. Truth be told, I don’t know the answer, and I think it’s possible that some version of this will come to pass. What is clear is that the protein structure field has a new, and formidable, research group. For academic scientists, especially the more junior among us, we will have to contend with whether it’s strategically sound for our careers to continue working on structure prediction.

Mohammed AlQuraishi, “AlphaFold @ CASP13: “What just happened?”” at Some Thoughts on a Mysterious Universe (December 9, 2018)

AI is already reshaping many careers, from the fashion industry to cancer research. In The Human Advantage, Jay Richards points out that the long-term trend of AI has always been to increase opportunity by empowering more people to do creative things—the ones machines don’t really do. Including things like developing AlphaFold.

Meanwhile, the AlphaFold technology that has already proven useful in the fight against COVID-19, will doubtless create new jobs tackling other diseases that were formerly hard to beat because, in a world so deeply designed and complexly organized, we didn’t really have any practical way of knowing what was going on. Now maybe we do.

You may also enjoy: If AlphaFold is a product of design, maybe our bodies are too. And the deeper we go into science, the more important our unique human contributions become.


How AI can help us fight cancer. Breast cancer is an excellent example of how AI can speed up early detection.

Mind Matters News

Breaking and noteworthy news from the exciting world of natural and artificial intelligence at MindMatters.ai.

AlphaFold Scores Huge Breakthrough in Analyzing Causes of Disease