An AI system can predict the structures of life’s molecules with stunning accuracy, helping to solve one of biology’s biggest problems

<spanklasse=Christoph Burgstedt / Shutterstock” src=”–/YXBwaWQ9aGlnaGxhbmRlcjt3PTk2MDtoPTQ3OA–/ faae3dc543a245148cd” data-src= “–/YXBwaWQ9aGlnaGxhbmRlcjt3PTk2MDtoPTQ3OA–/ 543a245148cd”/>

Unveiled to the world on May 9, AlphaFold 3 is the latest version of an algorithm designed to predict the structures of proteins – vital molecules used by all life – based on the ‘instruction code’ in their building blocks.

Predicting protein structures and the way they interact with other molecules is one of the biggest problems in biology. Yet AI developer Google DeepMind has found a way to solve this problem in recent years. This new version of the AI ​​system offers improved functionality and accuracy compared to its predecessors.

Like the next release in a video game franchise, structural biologists – and most recently – chemists have been waiting impatiently to see what it can do. DNA is commonly seen as the instruction book for a living organism, but in our cells, proteins are the molecules that actually do most of the work.

They are proteins that allow our cells to perceive the outside world, integrate information from different signals, create new molecules within the cell, decide to grow or stop growing.

They are also proteins that enable the body to distinguish between foreign invaders (bacteria, viruses) and itself. And it’s proteins that are the target of most of the drugs you or I use to treat disease.

Protein Lego

Why is protein structure important? Proteins are large molecules consisting of thousands of atoms in a very specific order. The order of these atoms, and the way they are arranged in 3D space, is crucial for a protein to perform its biological function.

This same 3D arrangement also determines the way a drug molecule binds to its protein target and treats disease.

Imagine you have a Lego set in which the bricks are not cube-based, but can be any shape. To join two stones in this set, each stone must fit snugly against the other, with no gaps. But this is not enough: the two stones must also have the right combination of bumps and holes so that the stones stay in place.

Designing a new drug molecule is a bit like playing with this new Lego set. Someone has already built a huge model (the protein target found in our cells), and the drug discovery chemist’s job is to use their toolbox to put together a handful of bricks that adhere to a given part of the protein and – in biological terms – stop it from performing its normal function.

So what does AlphaFold do? Based on knowing exactly which atoms are in any protein, how these atoms have evolved differently in different species, and what other protein structures look like, AlphaFold is very good at predicting the 3D structure of any protein.

AlphaFold 3, the most recent version, has expanded the possibilities to model nucleic acids, for example pieces of DNA. It can also predict the shapes of proteins modified with chemical groups that can turn the protein on or off, or with sugar molecules. This gives scientists more than just a bigger, more colorful Lego set to play with. It means they can develop more detailed models for reading and correcting the genetic code and cellular control mechanisms.

This is important in understanding disease processes at the molecular level and in developing drugs that target proteins whose biological role regulates which genes are turned on or off. The new version of AlphaFold also predicts antibodies with greater accuracy than previous versions.

Antibodies themselves are important proteins in biology and are an essential part of the immune system. They are also used as biologics such as trastuzumab, for breast cancer, and infliximab, for diseases such as inflammatory bowel disease and rheumatoid arthritis.

The latest version of AlphaFold can predict the structure of proteins bound to drug-like small molecules. Drug discovery chemists can already predict the way a potential drug will bind to its protein target once the target’s 3D structure has been identified through experimentation. The downside is that this process can take months or even years.

Predicting how potential drugs and protein targets bind together is used to help decide which potential drugs to synthesize and test in the laboratory. Not only can AlphaFold 3 predict drug binding in the absence of an experimentally identified protein structure, but when tested it also outperformed existing software predictions, even when the drug’s target structure and binding site were known.

These new capabilities make AlphaFold 3 an exciting addition to the repertoire of tools used to discover new therapeutic drugs. More accurate predictions will allow better decisions to be made about which potential drugs to test in the laboratory (and which are unlikely to be effective).

Time and money

This saves both time and money. AlphaFold 3 also offers the ability to make predictions about drug binding to modified forms of the protein target that are biologically relevant but currently difficult – or impossible – to make using existing software. Examples include proteins modified by chemical groups such as phosphates or sugars.

Of course, as with any new potential drug, extensive experimental testing for safety and efficacy is always required – including in human volunteers – before it is approved as a registered medicine.

AlphaFold 3 has some limitations. Like its predecessors, it is poor at predicting the behavior of protein regions that do not have a fixed or ordered structure. It is poor at predicting multiple conformations of a protein (which can change shape due to drug binding or as part of its normal biology) and cannot predict protein dynamics.

It can also make some slightly embarrassing chemical mistakes, such as placing atoms on top of each other (physically impossible), and replacing some details of a structure with its mirror images (biologically or chemically impossible).

A more substantial limitation is that the code will not be available – at least for the time being – so it will have to be used on the DeepMind server on a purely non-commercial basis. While this will not deter many academic users, it will limit the enthusiasm of expert modelers, biotechnologists and many drug discovery applications.

Nevertheless, the release of AlphaFold 3 certainly seems to stimulate a new wave of creativity in both drug discovery and structural biology on a larger scale – and we’re already looking forward to AlphaFold 4.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

The conversation

The conversation

The authors do not work for, consult with, own shares in, or receive funding from any company or organization that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

Leave a Comment