AlphaFold facts for kids

AlphaFold is a special artificial intelligence (AI) computer program. It was made by a company called DeepMind, which is part of Alphabet (the same company that owns Google!). AlphaFold's main job is to guess the shapes of proteins. Think of it like a super-smart detective trying to figure out how a long chain of beads (amino acids) folds up into a tiny, complex 3D shape.

AlphaFold uses a type of AI called deep learning. This means it learns from lots of examples, much like how your brain learns from experience.

There have been three main versions of AlphaFold.

• AlphaFold 1 came out in 2018. A team using it won first place in a big science competition called Critical Assessment of Structure Prediction (CASP). This competition tests how well computer programs can predict protein shapes. AlphaFold 1 was especially good at guessing the shapes of very difficult proteins, even when there weren't similar examples to learn from.

• AlphaFold 2 was released in 2020. It also won first place in the CASP14 competition. This version was much, much more accurate than any other program. For about two-thirds of the proteins, it scored over 90 out of 100 on a test called the global distance test (GDT). A score of 100 means the predicted shape was a perfect match to the real shape found in labs.

Scientists were amazed by AlphaFold 2's results. They called them "astounding" and "transformational." Even though it wasn't perfect for every protein, many experts agreed it was accurate enough to help understand tiny changes in proteins. In July 2021, the scientific paper about AlphaFold 2 was published. The program's code was also made open source, meaning anyone could use it. A huge database of predicted protein shapes for many different living things was also created. Other similar programs have also been developed, like ESMFold and RoseTTAFold.

• AlphaFold 3 was announced in May 2024. This newest version is even better at predicting protein shapes. But it can do more! It can also predict the shapes and how they interact with other important molecules like DNA (our genetic code), RNA, and even smaller molecules called ligands and ions.

How AlphaFold Helps Science

AlphaFold has been very useful in understanding diseases.

Fighting Viruses

For example, AlphaFold was used to predict the shapes of proteins from SARS-CoV-2. This is the virus that causes COVID-19. In early 2020, scientists didn't know the exact shapes of many of these virus proteins.

Researchers at the Francis Crick Institute in the UK used AlphaFold to guess these shapes. They then shared their findings with other scientists around the world. They also checked AlphaFold's predictions against real protein shapes that had been found in labs, like the spike protein of the virus.

Even for proteins that weren't being studied for new medicines, AlphaFold's predictions helped scientists learn more about how the SARS-CoV-2 virus works. For instance, AlphaFold 2 accurately predicted the shape of a protein called ORF3a. This protein is thought to help the virus escape from human cells after it has made copies of itself. It might also play a role in causing the body's immune system to overreact to the infection.

Images for kids

• 

  Amino-acid chains, known as polypeptides, fold to form a protein.

• 

  AlphaFold 2 performance, experiments, and architecture

See also

In Spanish: AlphaFold para niños