For centuries, musical inspiration has come from a whole host of places, from bumblebees and the planets to fire and water. But researchers from the United States and Taiwan have taken their inspiration from a rather unusual source – the building blocks of life itself, proteins.
Found all over the human body, proteins, coded for by our genes, are heavily researched molecules in science. Yet the question of what they "sound" like had remained a mystery. Chi-Hua Yu from National Cheng Kung University, Taiwan, and Markus J. Buehler from Massachusetts Institute of Technology (MIT) have not only provided us with this long-awaited “protein music”, but in doing so, have trained artificial intelligence (AI) to design new proteins – a process that is usually time-consuming and has outcomes that are difficult to predict.
The duo’s research, published in APL Bioengineering, stemmed from the knowledge that each of the 20 amino acids that make up proteins has a unique vibrational frequency, which could be translated to the different frequencies of musical notes, otherwise known as pitch. Using this as the basis, the researchers were able to translate different protein traits into musical concepts to build up a complete score. For example, the chain length and folds of the protein could be reflected in variations in note length, volume, melody, chords, and rhythm.
Luckily for you, some of the protein music is available on SoundCloud to listen to. One piece represents a protein found in the venom of predatory marine snails, whilst another lengthier composition expresses the recently mapped COVID-19 spike protein.
However, the musical masterpieces are only part of the story. The generated scores were then also used to train deep learning neural networks in the art of protein music composition, enabling them to design not yet invented proteins.
"These networks learn to understand the complex language folded proteins speak at multiple time scales," Markus J. Buehler explained in a statement. "And once the computer has been given a seed of a sequence, it can extrapolate and design entirely new proteins by improvising from this initial idea.”
Translating the pitch and other musical information from the algorithm’s improvised scores back into sequences of amino acids led to the creation of never-before-seen proteins. The researchers also discovered that the production rate of musical variations could be controlled by adjusting the temperature during protein design – the higher the temperature, the greater the number of variations were produced by the algorithm.
"This paves the way for making entirely new biomaterials," Buehler continued. "Or perhaps you find an enzyme in nature and want to improve how it catalyzes or come up with new variations of proteins altogether.”