The 2024 Nobel Prize In Chemistry : Why You Should Care About Proteins

Scientists won the Nobel Prize
12 followers
2 Video ViewsยทMay 27, 2025
At the time of making this video, the 2024 Nobel Prize in Chemistry was announced,
and it was awarded to Demis Hassabis and John Jumper for their innovations with alpha-fold
and predicting protein folding. It was also awarded to David Baker for the computational
design of proteins. If you're someone who has not come across proteins before,
then you're in for a treat today, because proteins are just about some of the most
awesome, beautiful, and functional molecular structures out there, and they can serve a
myriad of incredible applications, such as cancer therapy and drug design, and even removing that
dirty stain from your favorite t-shirt. So what is protein folding all about,
and why should you care about it? Welcome to Atomic Stories.
Before we explain the concepts behind the Nobel Prize, it's important to review some
There are 21 amino acids that are found in nature, and their chemistry allows them to
bind together to make small strands called peptides. When these peptides start to get
really long, they tend to either assemble themselves into higher-order structures,
or they get some help from other proteins. This process is called protein folding,
and it not only determines the final shape of this entire collection of amino acids,
but it ultimately determines the final function of the protein.
But why are proteins so cool? What makes them so special, and why should you care about them?
Well, you can think of proteins as macromolecular machines, literally. I mean, look at the way this
protein rotates. Does it not remind you of something like a gearbox or a drill?
But instead of changing gears, or making holes, this protein on screen produces ATP.
Isn't that special? And it's not just ATP synthase that does cool stuff.
Take a look at kinesin. Does it not remind you of someone taking a stroll on the street?
So you see, proteins are super cool and intricate structures, but it's one thing to observe their
function, and it's another to predict how they will fold in order to have said function.
This is the heart of the 2024 Nobel Prize in Chemistry.
Developed by Demis Hassabis, John Jumper, and the team at Google DeepMind,
AlphaFold2 is an AI model that can accurately predict the structure of almost any known protein.
It uses a type of neural network called a transformer, which is great at finding patterns
in data. AlphaFold2 analyzes the relationships between amino acids in a sequence, and leverages
vast amounts of protein data to learn how different sequences fold into their unique structures.
The transformer architecture used by AlphaFold2 is key to its success. Transformers are particularly
powerful because they can process all parts of protein sequences simultaneously, rather than
one part at a time. This allows AlphaFold2 to understand complex dependencies between amino
acids, even if they are far apart in the sequence. Another critical aspect of AlphaFold2's approach
is the incorporation of evolutionary data. Proteins evolve over time, and similar sequences
are often found across different species. By comparing a target sequence to known,
similar sequences, AlphaFold2 can make informed predictions about the likely structure.
This combination of deep learning, transformers, and evolutionary analysis is what makes AlphaFold2
so effective. But there's more. Another laureate, David Baker, took the challenge even further.
Not just predicting, but designing new proteins from scratch. He and his team designed computational
tools to build proteins with new, custom functions, opening doors to creating proteins
that can break down pollutants, fight diseases, or even help in creating new materials. Essentially,
they're designing proteins that do things nature never intended, but that humanity desperately
needs. Despite this, one major limitation is the availability of data. These systems rely heavily
on existing experimental data to make accurate predictions. For many proteins, especially those
from less studied organisms or newly discovered proteomes, there is often insufficient data
available. Without a large and diverse dataset, these AI models struggle to predict structures
accurately. Not only is data limited, but so-called bad data can be completely omitted,
because many researchers do not report failed attempts, meaning that AI models cannot learn
from trials that did not work. This limitation highlights the importance of ongoing experimental
research to provide high-quality data that can feed into and improve these predictive models.
Well, now you know that protein folding can be predicted, but why should you even care about
proteins in the first place? Well, if you've ever consumed any modern medicine in your life,
chances are that some of those drugs were made with the help of proteins. One of the most
important factors when synthesizing drugs is their purity and stereochemistry. Stereochemistry sounds
like a very difficult word and topic, and it is if you get into the details, but in general,
you can think of stereochemistry as the way atoms are organized in three-dimensional space.
This is super important, and let me show you an example of why.
There's a famous drug which showcases the power of stereochemistry, and its name is thalidomide.
Thalidomide was a drug that treated morning sickness in pregnant women, and here are two
of its enantiomers. One points towards the screen, and the other points to the back of the screen.
The structures are completely identical, the only difference being their three-dimensional
orientation. Now what if I were to tell you that one structure could treat morning sickness,
but the other one causes birth defects? This is how powerful stereochemistry can be.
This drug used to be present as a mixture of both isomers, and due to this, it historically
led to what is known as the thalidomide tragedy, where its use caused devastating birth defects.
Proteins are super useful for making drugs with the correct stereochemistry,
and this is due to the nature of their shape. By having new predictive tools at our hand,
we can better engineer proteins, making them not only produce the correct drugs,
but also produce them with high efficiency and speed. And this computational design does not
only work for drugs, but for proteins involved in other applications, such as breaking down plastic,
or removing oil stains from your clothes. That's the end of the video! Hopefully you
appreciate proteins and the molecular world a little bit more after watching,
and thanks for sticking until the end. Please do let us know what your thoughts are about this
video, it means a lot to us to receive constructive criticism. If you enjoyed the video, please
consider sharing it with your friends and family, and if you want to watch more of our videos,
feel free to subscribe. Until next time, bye bye!
