Enzymes are the engines of life − machine learning tools could help scientists design new ones to tackle disease and climate change

Enzymes are molecular machines that carry out chemical reactions that maintain all life, and they are the ability that has acquired Sayed scholars.

Consider muscle movement. Your body releases a molecule called acetylchlin to create your muscle cells to contracting. If acetylchlin holds for a long time, your muscles – including myocardial cells can be paralyzed – that is. This is where the enzyme Asetel Culturester This enzyme comes can collapse thousands of acetylchlene molecules per second to ensure muscle contraction, avoiding paralysis and life continued. Without this enzyme, it will take a month for a molecule of acetyl colin 10 billion times slower.

You can imagine the reason for the importance of enzymes in particular for scientists looking to solve modern problems. What if there is a way to destroy plastic, capture carbon dioxide, or destroy cancer cells as quickly as to break the acetyl Colinstrase acetyl Collin? If the world needs to take action quickly, the enzymes are a convincing candidate for this job – if only researchers can design them to deal with these challenges on the demand.

Enzyme design, unfortunately, is very difficult. It is similar to working with an atomic LEGO group, but the instructions are lost and will not stick together unless it is completely assembled. Recently published research from our team indicates that automated learning can serve as the architect in this LEGO group, which helps scientists Building these complex molecular structures accurately.

What is the enzyme?

Let’s take a closer look at what is an enzyme.

Enzymes are proteins Large molecules that do behind the scenes that keep all living organisms alive. These proteins consist of Amino acids,, A group of basic building blocks that can be sewn together to form long chains that are complex in specific forms.

The specific structure of the protein is the key to its function in the same way in which the forms of daily beings are located. For example, like a lot of spoon designed to hold the liquid in a way that the knife can not simply, the enzymes participating in moving your muscles are not perfectly suitable for optical representation in plants.

In order for the enzyme to work, it adopts a shape that completely matches the molecule that treats it, very similar to a Lock. The unique grooves are found in the enzyme – the lock – which interacts with the target molecule – the key – in an area of ​​the enzyme known as the active site.

The active location of the enzyme wraps the amino acids to interact with the target molecule when entering it. This makes it easier for the molecule to undergo a chemical reaction to convert it into a different group, which makes the process faster. After completing the chemical reaction, the new molecule is launched and the enzyme is ready for another processing.

How to design an enzyme?

Scientists have spent contracts in an attempt to design their own enzymes to make new molecules, materials or treatments. But making enzymes that look and go as soon as possible in nature is very difficult.

Enzymes have irregular complex shapes consisting of hundreds of amino acids. Each of these building blocks should be placed perfectly, otherwise the enzyme will slow down or close completely. The difference between the rider of speed and slow enzyme can be less distance than one atom width.

Initially, scientists focused on Modify the amino acid sequence of existing enzymes To improve their speed or stability. Early successes with this approach improved the stability of enzymes in the first place, allowing them to stimulate chemical reactions in a higher temperature. But this approach was less useful to improve enzymes. To this day, the design of the new enzymes by modifying individual amino acids is not an effective way to improve natural enzymes.

The researchers found that the use of a process is called Development directedWhere the amino acid sequence of the enzyme is changed randomly so that it can perform the required function, proved to be more useful. For example, studies have shown that directed development can improve the speed of chemical reaction, heat, and even generate enzymes with characteristics that are not seen in nature. However, this approach is usually an intense work: you should check many mutants to find an approach that does what you want. In some cases, if there is no good enzyme to start, this method may fail at all.

Both these two rituals are limited due to their dependence on natural enzymes. This means that restricting your design on the forms of natural proteins is likely to limit the types of chemistry that enzymes can facilitate. Remember, you cannot take the soup with a knife.

Is it possible to make enzymes from scratch, instead of modifying the nature recipe? Yes with computers.

Enzyme design with computers

The first attempts to design mathematical enzymes still depend largely on natural enzymes as a starting point, focusing on setting Enzyme active sites in natural proteins.

This approach is similar to trying to find a suit in the savings store: it is unlikely to find an ideal attack because the engineering of the active enzyme site (your body in this analogy) is very specific, so a random protein with strictly is unlikely to accommodate the fixed structure (a suit with random measurements) completely. The enzymes caused by these efforts were much larger than those in nature, which required more improvement with the development directed to reach common speeds between natural enzymes.

Recent developments in deep learning have changed the scene of enzyme design using computers. Enzymes can now be created in the same way as artificial intelligence models such as Chatgpt and Dall-E text or pictures, and do not need to use the original protein structures to support your active site.

Our team showed that when we ask the artificial intelligence model, He called RFDIFFUTIONWith the structure and the amino acid sequence of an active site, it can create the rest of the enzyme structure that will completely support it. This is equivalent to paying Chatgpt to write a full short story based on a claim that says only to include the line “Unfortunately, the eggs have never appeared.”

We used the artificial intelligence model specifically to create enzymes called Hydrolis SereneA group of proteins that have possible applications in medicine and plastic recycling. After designing the enzymes, we mixed them with their intended molecular goal to see if they can stimulate their collapse. Foreoustly, many of the designs we have tested BreakIt is better than the previously designed enzymes for the same reaction.

To find out the accuracy of our mathematical designs, we used a method called X -ray crystals to determine the shapes of these enzymes. We found that many of them were A perfect match To what we designed digitally.

The results of our findings were distinguished by the design of the enzyme, with a highlight of how artificial intelligence helps scientists start addressing complex problems. Automatic learning tools can help more researchers reach the enzyme design and take advantage of the full potential of enzymes to solve the problems of the modern era.

This article has been republished from ConversationAn independent non -profit news organization brings you facts and a trustworthy analysis to help you understand our complex world. Written by: Sam Belkand Washington University

Read more:

Sam Belok receives funding from the Washington Research Foundation and the future Schmidt program.