Protein structure prediction

Protein structure prediction is a way to figure out the shape of a protein just by looking at its amino acid sequence. This means scientists can predict how the protein folds into its special shape without having to see it under a microscope.

This is really important because proteins do many jobs in our bodies, and knowing their shapes helps us understand how they work. It also helps scientists make new medicines and design special enzymes for different uses.

Every two years, scientists test how well their methods work in an event called CASP. There is also a project called CAMEO3D that checks web tools for predicting protein shapes all the time.

Protein structure and terminology

Proteins are made from chains of small building blocks called amino acids, linked together by special bonds known as peptide bonds. These chains can twist and turn in many ways, creating different shapes. The way the chain twists is controlled by two important angles at each link, called φ and ψ.

The peptide bonds have special properties that let them connect with each other through tiny forces called hydrogen bonds. Proteins usually contain 20 different types of amino acids, and each type has a unique part called a side chain that helps decide the protein's shape. Some amino acids, like glycine, are very flexible because their side chain is tiny, while others, like cysteine, can link with each other to make the protein more stable.

Proteins often fold into common patterns called secondary structures. One common pattern is the α-helix, a spiral shape where every fourth amino acid forms a hydrogen bond to hold the spiral together. Another pattern is the β-sheet, where strands of the protein chain lie next to each other and form hydrogen bonds across the sheet. These patterns help the protein fold into its final three-dimensional shape. Some parts of proteins don’t have a regular shape but still help connect different parts of the protein. These are called deltas.

Main article: α-helix

Main article: β sheet

Protein classification

Proteins can be grouped based on how they look and their building blocks. Scientists look at how proteins fold into shapes and how their building blocks line up to sort them into groups.

There are special words scientists use to talk about these groups, like "fold" for a certain shape, or "family" for proteins that do similar jobs and look alike. These groups help us understand how proteins work and how they change over time.

Secondary structure

Main article: List of protein secondary structure prediction programs

Secondary structure prediction is a way to guess the shape of proteins using just their building blocks, called amino acids. Scientists try to predict which parts of the protein will twist into spirals called alpha helices, which parts will lay flat like ribbons called beta strands, and which parts will bend called turns.

Today’s best methods can guess these shapes correctly about 80% of the time. They use smart computer programs that learn from many known protein shapes. These guesses help scientists understand how proteins fold and work together.

Tertiary structure

Main articles: homology modeling and fold recognition

Predicting how proteins fold into their 3D shapes is very important because we can learn a lot from it. Scientists get lots of information about proteins from reading their DNA, like from the Human Genome Project. But figuring out the exact shape they take is hard because it takes a lot of time and money to do it in a lab.

There are two main ways to predict protein shapes. One way looks at proteins we already know and guesses that new proteins will look similar if their DNA codes are close. The other way tries to build the shape from scratch, but this needs a lot of computer power.

Before making a guess, scientists often break the protein into smaller parts called domains. Then they try to fit these parts together to make the whole shape. Some methods use computer programs to try many different shapes quickly, while others look at patterns in the DNA to help guess the shape. Even with modern computers, this is still a big challenge, but it helps us understand how proteins work.

Quaternary structure

Main article: Protein–protein interaction prediction

When two or more proteins work together, scientists can use special methods to guess how they fit together. If we know the shape of each protein, we can predict how they form a group. Studying how changes in the proteins affect their connection helps us understand their combined shape better.

Software

There are many tools that help scientists guess how proteins will look in three dimensions just from their building blocks, called amino acids. These tools use different methods, like comparing proteins to others whose shapes we already know, or using smart computer programs to figure out the shape.

One important tool is called AlphaFold, which was created using advanced computer learning. It can quickly predict the shape of proteins very accurately. Scientists use these predictions to learn more about what proteins do in our bodies. For example, knowing the shape of a protein can tell us if it helps build strong tissues, like collagen, which makes up things like skin and bones.

Every two years, scientists meet in a competition called CASP to test how well these tools work. This helps everyone learn which methods are the best for predicting protein shapes.