skip to Main Content

Translating the Coronavirus to Music with Artificial Intelligence

December 16, 2020 | cloud, Emerging Technologies
Read Time: 10:00

MIT Professor and scientist Markus Buehler is attempting to understand the deadly Coronavirus (Covid-19) better by turning it into classical music. By translating the spike protein of the novel coronavirus into sound, Professor Buehler and his team visualized the virus’ vibrational properties. These patterns could help us find a way to stop the virus. Spike protein is the appendage of the virus that makes it so contagious.

We know that all living things are made of proteins and proteins are made up of smaller units called amino acids. But did you know that these proteins are all alive with music? Professor Buehler has been working for quite some time on developing artificial intelligence and deep learning models to design new proteins. Translating the protein sequences into sound is one of the ways he accomplishes this. The researchers assign a unique musical note to every amino acid in the protein. An algorithm then converts these notes in the amino acid sequence into music. By doing this, the team aims to create new biological materials for sustainable, non-toxic applications. The research involves the active use of artificial intelligence, machine learning, deep learning, etc.

When this process was applied to the coronavirus spike protein sequence it yielded different pieces of music, depending on the stage of the virus’ lifecycle. The first part of the ‘coronavirus music’ is soothing, representing the stage during which time the spiked protein eases into our unsuspecting cells. In the second stage, as the virus replicates and the protein binds to more cells, the corresponding music becomes dramatic and tumultuous.

What does this translation accomplish?

By translating the virus proteins into music, the research team hopes to be able to help develop antibodies by taking a musical counterpoint to the virus’ melody and rhythm and then use artificial intelligence & machine learning to find an antibody that matches it. The overall coronavirus melody is found to be about 1 hour 50 minutes long. This piece could unpack the vibrational properties of the spike protein through molecular-based sound spectra, which could unravel the key to stopping the spread of the deadly coronavirus.

What are the benefits of translating proteins into sound?

Human brains are amazing when it comes to processing sounds. In just a few seconds, our ears can figure out the characteristics of the sound – the pitch, the timbre, the volume, the melody, the rhythm, even the chords. If something similar had to be done with an image, one would need a high-powered microscope to pick up such minute details of the media. And even then, not all the components could be seen at once. However, this is not the case with sound, making sound a lot more elegant way to access the information that is stored in the proteins.

How is the sound produced?

Sound is generally produced by a vibrating material, say a guitar string, or a percussion skin like in a tabla, or a vibrating screen like in a harmonium. These different sounds when arranged in characteristic hierarchical patterns produce music. Artificial intelligence and machine learning can combine these fundamental concepts of music and construct new musical forms using molecular vibrations and neural networks. The research team at MIT has been working on methods to turn protein structures into audible representations, and in turn, translate them into new biological materials.

What has the ‘coronavirus melody’ told us?

The spike protein of the novel coronavirus or SARS-CoV-2 as it is known comprises three protein chains that are folded together into an intriguing compact pattern. While our eyes are not capable of seeing this, our ears are perfectly capable of hearing it. The physical protein structure along with the entangled chain structure has been reproduced into interwoven music pieces forming a multi-layered composition. The composition covers everything about the spike protein – the amino acid sequences, the secondary structure patterns, and the intricate three-dimensional folds. As a result, the music derived is a sort of counterpoint music – notes being played against notes, since there is so much happening at the same time. It is like a symphony, where multiple musical instruments are played at the same time to produce melodious music. So, just like a symphony, the coronavirus musical patterns reflect the protein’s intersecting geometry that is realized by materializing the DNA code.

We already know that the Covid virus has an uncanny ability to deceive and exploit the host for its multiplication. The coronavirus genome is known to hijack the host cell’s protein manufacturing machinery and then force it to replicate the virus’ proteins, with which it can create new viruses. This sounds cruel and horrifying, but the music this process produces is hard from being as unpleasant or terrifying. And that is one of the crucial things – this particular music tricks our ears just as the coronavirus tricks our cells. It is like a fox in sheep-skin, an invader disguised as a friendly visitor. And now, thanks to this melody, we can see the coronavirus in a completely new light, a completely new angle. The solution to the coronavirus crisis in essence could lay in the urgent need to learn and appreciate, the language of proteins.

This process, though, could take time. Translating the proteins into sound doesn’t yield a miracle cure directly. Instead, they help understand and design the proteins that could help build a cure. The genome is such a sensitive entity that a slight mutation can limit or enhance the pathogenic power of the COVID-19 virus. Another major advantage of this process of sonification is that it can help compare the biochemical processes of the coronavirus spike protein with the previously known coronaviruses like the SARS virus or the MERS virus.

Understanding the vibrational patterns of the virus protein produced using artificial intelligence and machine learning can prove to be a critical tool for drug design. These vibrations would change with changing factors like temperature. It could also help us unravel why the SARS-CoV-2 has such great affinity for human cells as compared to other viruses.

A compositional approach would also be helpful to design drugs for the coronavirus infection. In this approach, the experts would search for a new protein that would match the melody and rhythm of an antibody that would be capable of binding to the spike protein. This antibody would then interfere with the virus’ ability to infect the human cells.

How can music help design proteins? 

Simply put, music is an algorithmic reflection of structure. Different notes are put together to form music. However, in a real sense, this process is done by our brains rather than a computer, though there is still a mental algorithm we are running. There are repetitions and sequences involved. This is exactly how things exist inside proteins as well. Nature puts together different amino acids in specific sequences to create specific proteins. By using deep learning, machine learning, and AI, we can hear this conception exactly as nature composed it. Proteins are a disguised form of nature’s music. So, once we unravel this music, we can compare it to known or imagined ideas or use artificial intelligence to speak the language of protein design to build new protein structures.

How can I be a part of such exciting projects?

The first step towards becoming a part of such exciting and interesting projects is to acquire the right skills and experience. Projects like these require up-to-date knowledge and lots of practical experience in machine learning, deep learning, and artificial intelligence. To begin this journey, you could enroll in intensive machine learning training which would help you learn the skills and acquire the necessary knowledge to begin a career in the field. The machine learning training should involve hands-on exercises to give you practical exposure, so you know not just the theory but also the practical application of the machine learning concepts learned. Make sure you get a globally recognized machine learning certification after the training. Getting certified in machine learning could help you move ahead in your career not just with your existing role but also conceptualize & deliver interesting projects. Keep yourself updated by interacting with machine learning professionals would also be helpful. Once you are certified in machine learning you could try to come up with your pet projects and try to build your project portfolio. Highlight your learnings and thoughts on LinkedIn and be more active in the machine learning community.

Where can I get the best online machine learning training?

Cognixia – the world’s leading digital talent transformation company offers thorough machine learning training along with the fundamentals of Python for application in ML. Our Machine Learning certification course covers all the important concepts and skills that would help you begin a career in machine learning. This machine learning course covers:

  • How to install and import libraries
  • Methods of handling different data types
  • Data visualization
  • Distinguishing between artificial intelligence, machine learning, and deep learning
  • Working with data in real-time
  • Implementation of machine learning algorithms
  • Implementation of deep learning algorithms
  • Types of time series data
  • Performing text and sentiment analysis
  • Business analytics

Aspiring machine learning professionals would find this course immensely beneficial.

Back To Top

  • This field is for validation purposes and should be left unchanged.