Q&A: Microsoft’s groundbreaking AI tools make it easy to find out how to build proteins

Microsoft’s AI for Good Lab We have created Seq2Symm, an open source AI tool that helps scientists determine the 3D shape of a particular protein, including those found in viruses.

SEQ2-SYMM uses AI to predict the 3D shape and structure of proteins from one-dimensional sequences. This tool will help researchers better understand diseases, develop drugs and vaccines, and create more sustainable materials.

Juan La Vista FerezCVP and Chief Data Scientist and Project Senior Research Scientist and Chief Researcher Meghana Kshirsagar, MobiHealthNews We will discuss SEQ2SYMM and how it will affect healthcare and more.

MobiHealthNews: Can you tell me about seq2symmm?

Juan La Vista Ferez: In general, proteins, and protein symmetry in particular, has been found to be extremely important. Proteins are important from areas like drug discovery to energy. Much of what we do as living organisms is dependent on proteins. Therefore, having an understanding of proteins and designing these proteins is useful for many researchers, and the fundamental aspect of it is to understand symmetry.

Until now, up until this discovery, there were ways to try to predict symmetry, but it could not be made very fast. So the overall idea of ​​these models… the main contribution is the fact that now that aspects can be done faster. If you can do that faster, it will help researchers work faster. This can facilitate research discovery.

Meghana Kshiragar: Juan said that the main contribution of this study is to understand the structure of proteins of certain types of proteins, including many repeating units, which are called homo-oligomers. These are extremely important as they appear in many living things. So, for example, they appear in viruses.

So these are [see picture above]for example, a viral capsid. They are these spherical structures and are present in almost all viruses. And what the virus does is that they place DNA in this capsule-like structure, which is put into our cells when the virus enters our bodies. This then falls apart, causing viral DNA to come out and increase. Currently, this is made of these repeating units, which are what are called homo-oligomers.

So, since there are 180 copies of the same thing and they are forming this wonderful sphere repeatedly, this is extremely essential for the virus to function well. This is a very important part of how it works.

Looking at a pandemic like Covid, what researchers first had from this virus was what they called virus sequences. In other words, they only have one-dimensional information. It’s just that you have someone’s name, like someone’s description, but you don’t have 3D information about them.

Our method is that we need this one-dimensional information and can predict this 3D information. It can be said that it forms this shape and has many copies of these.

Therefore, we can imagine a large number of situations that do not have this 3D information of the molecule or protein of interest. I know only this one-dimensional information.

But then going to 3D is very important and what we do here is to predict what the number of copies and shapes will look like. This is one specific application that allows you to use methods.

MHN: So it is a predictive model.

Kshirasagar: Yes, that’s a predictive model.

Ferez: We predict, but this is an example, a virus is an example. Again, this is protein dependent for everything that is living organism. Therefore, this involves a vast range of issues, not just viruses, but also from understanding Alzheimer’s disease to creating new drugs. Therefore, due to the dependence that better understand proteins, the types of effects and effects this has are incredible.

MHN: Are there any particular areas that are most promising? Like you said, maybe cancer or Alzheimer’s.

Kshirasagar: So certainly there are applications in the study of Alzheimer’s disease and viruses. These are the biggest applications from a health perspective. And of course, there are many applications such as sustainability.

MHN: So it’s not just medical care. This is something you said that can be used in all living creatures.

Kshirasagar: yes.

Ferez: Exactly, and this is from the materials… This is one of the reasons why we decided to invest in a better understanding of protein folding, and we have been working with Baker Lab, Gregory Bowman and our team for at least three to four years.

These are extremely difficult, very important, and sometimes not the easiest projects to explain.

Many people don’t understand why we care so much about protein. Obviously, these are fundamental aspects of life and material, and they basically touch on everything.

MHN: And you made it an open source model.

Ferez: This is open research and is completely open source. Anyone can use it to conduct further research. Our impact provides these tools so that other researchers can take advantage of it. We expect others to have an impact, so we are experiencing it through this.

This will affect evolving diseases, how to target drugs, and how vaccines and new therapies are designed. Therefore, it has a broad impact.

Kshirasagar: As fans said, proteins form the fundamental components of not only all life on Earth, but many materials, and affecting its space, leads to a very broad and useful tool.