Did Google’s DeepMind Just Revolutionize Medicine?

I remember the day in middle school when I was taught that amino acids were “the building blocks of life”. I was fascinated by the idea that our complex shape, and the shape of other living organisms, was like a tiny Lego set, built to make us what we were. Even then, in the early 1980s, researchers had already been trying for nearly a decade to figure out how those amino acids told proteins which form to take. Since then, with ever more powerful computers and complex algorithms, researchers have applied machine learning techniques to answer the same biological question.

Google‘S (NASDAQ: GOOG)(NASDAQ: GOOGL) DeepMind just came up with an answer and it’s driving researchers crazy. For nearly 50 years, scientists have wondered how proteins know what shape to fold into, and they do so repeatedly from time to time. In a modeling competition, DeepMind researchers just broke the code by creating a model that translates amino acid chains into three-dimensional protein structures. To understand how this could impact medicine (and investment) it is important to understand what will allow scientists to make this new knowledge. Also, what are the downstream effects? Which areas of biological research can be most affected? And which companies are going to gain – or lose – the most?

The competition

There are tens of thousands of proteins in humans and billions in other species, viruses and bacteria. How these proteins fold directly determines what they do. Indeed, in molecular biology it is said that “structure is function”. The folded shape is the key to the role that proteins, such as infection-fighting antibodies or insulin play, in regulating blood sugar. This is why the Critical Assessment of Protein Structure Prediction (CASP) has been held since 1994. It is an event that challenges teams to improve the accuracy of predictions in the field of protein structure.

AlphaFold, DeepMind’s winning model, was trained on public data from 170,000 protein structures. The program required 128 high-end cloud computing cores running for several weeks to create the algorithm. In the end, two-thirds of the model’s accuracy scores represent design errors smaller than a single atom. DeepMind was head and shoulders above the other attendees at the event, which consisted mostly of academic teams, but included voices of Microsoft (NASDAQ: MSFT) and the Chinese Internet giant Tencent (OTC: TCEHY).

Because it’s important

Most of the drugs prescribed today were discovered either by accident or by trial and error experiments that take a long time. Understanding how amino acids direct proteins to twist and bend, taking their three-dimensional shape, will create a better understanding of why each protein becomes what it does and how these signals are transmitted across cell membranes. This could allow scientists to better design the drugs that will be used by cells the way they want, understand the diseases that cause misfolding, and allow drug manufacturers to identify the cause of the genetic variations that lead to the disease.

In one example at the event, the AlphaFold model provided the structure of a bacterial protein in just 30 minutes. The Max Planck Institute in Germany had been working on this very problem for more than a decade. Next, the team could begin to address the thousands of unsolved proteins in the human genome and the hundreds of millions of proteins in nature that have not been modeled. This raises the question of when we can all get drugs designed for our specific biology.

What to look for

For now, drug discovery applications will have to wait. It’s unclear when or how DeepMind will share its model and, while impressive, it did have limitations. For example, the model struggled to predict protein complexes or clusters where interactions between proteins can distort shapes. As multiple proteins are involved, the potential for interactions to be modeled becomes nearly impossible. This mathematical constraint, known as combinatorial explosion, is common in advanced modeling, but could eventually be overcome with more computing power. It will be important to address this issue, as protein-protein interactions are one of the key mechanisms targeted for discovering new drugs.

Despite the warnings, the discovery promises to add fuel to the focus of scientific research on the functioning of the human body. A better understanding of the translation of amino acids into proteins validates the potential impact of genetic modification and companies like CRISPR Therapeutics (NASDAQ: CRSP), Intellia Therapeutics (NASDAQ: NTLA), is released Medicinal (NASDAQ: EDIT). Additionally, solving this problem should ultimately lead to less trial and error in the lab and make genome sequencing even more important, to the benefit Illuminate (NASDAQ: ILMN), Thermo Fisherman Scientific (NYSE: TMO), is Agilent (NYSE: A). After all, DNA carries information to make proteins.

The benefits of DeepMind’s discovery will largely remain behind the curtain of research, appearing to most of us just as other medical discoveries have – in the form of new or better drugs to treat disease. But make no mistake about the importance. A CASP judge, a computational biologist at Columbia University, called it one of the most significant discoveries of his life. CASP’s co-founder also added: “I never thought I’d see him in my life.” I expect this to be the first salvo in a new battle against human disease. Armed with a better understanding of the building blocks of life and previously unthinkable computing power, we may soon look back on our current drug discovery process as we now look at treating infections before penicillin was available, or monitoring the pregnancy before ultrasound – both advances made in the 1950s. In seventy years, people may marvel at the effort made by drug discovery and wonder how we were able to develop drugs in such a haphazard process.