Musical proteins

Scientists announced they can map out the structures of important proteins - things like the hormone insulin, or even spider silk - using music. This gives them a new dimension for studying how proteins work, and how we can use biotechnology to tweak proteins to do other important jobs, as Ankita Anirban explains...

Ankita - We think of eggs as providing us with a good source of protein but, in fact, proteins are everywhere. They're the basic building blocks of all living things. They're found in plants, animals, and even food. Jelly is, in fact, an edible protein. Proteins come in many shapes and sizes; they can be ordered structures like a corkscrew or more randomly shaped like noodles cooking in water. And for larger objects they can literally look like brick and mortar structures. Now Marcus Buehler and his team at MIT have found a way to translate these protein structures into music. But first, I asked him what is a protein actually made of?

Marcus - Proteins are made from building blocks called amino acids. There are 20 unique amino acids; you can imagine them being like bead on a string. Each bead that you add to the string is one of those 20 amino acids, and which amino acid is actually added onto the string is decided, defined by the DNA that is the blueprint for how this particular protein is made.

Ankita  - But what does this have to do with music?

Marcus - So the way we understand how these amino acids actually sound like, and if we do a very careful analysis we found that these molecules aren't static, they continuously vibrate. These vibrations can be calculated based on quantum mechanics and what we found is that each molecule has a unique frequency spectrum, and once we make that audible, each molecule has a very particular way it sounds like.

Ankita - So that's apparently how a lisosyn molecule sounds; that's a protein which you find in egg whites. So what’s actually the point in doing this?

Marcus - By listening to the proteins we can begin to link the structure of proteins. How they fold, the kind of function they have with how they sound. And we can begin to edit the protein and actually change the sound and maybe design new proteins or understand how mutations in proteins affect their functionality. Many diseases originate through mutations of proteins and in the audible space we can listen to different proteins, a healthy protein, the diseased protein and we can begin to understand what are the changes that actually cause this breakdown of functionality and maybe cause disease.

Ankita - Sounds like there may be very useful applications that come from this work.

Marcus - Proteins are everywhere in nature and a lot of times they actually act as materials. And these give us an opportunity to create materials that are not only made from renewable resources but they also are resilient. They usually have very strong mechanical properties and they can degrade automatically. With this new approach of being able to understand at a very different level how we can design proteins, we might be able to create new materials that have more sustainable properties, even better functionalities that are sustainable, resilient and can be recycled in a very simple way.

Ankita - Given that there are 20 different amino acids and we can arrange them in any way, surely there are millions of possibilities for new kinds of proteins? How do we know which are actually useful?

Marcus - We not only have translated these proteins into sound but we've also used artificial intelligence to help us understand the language that nature speaks. We train a neural network; we then let the neural network create new sounds and we translate the new sounds back into the protein sequences. And then we try to make these proteins and try to understand what they actually do and how they function and, possibly, hopefully finding proteins that actually perform even better than the natural proteins that nature has created.