How can computers be used to study protein?

We can combine the fundamental concepts of chemistry, physics, and biology to predict and describe how molecules move.

Chemistry tells us that everything is made up of atoms that interact with each other in interesting ways. By treating each atom as essentially a sphere, we can pretend the bonds holding them together are springs. Just like when you hold and move a spring, you can move the atoms together and pull them apart, and even bend and rotate them, but there's a built-in resistance that depends on how strong the spring is.

We can then think about all the forces that exist between atoms. The bonds (springs) themselves are one type of force, but there's even more according to general chemistry. There are forces based on charges, either "full" or "partial", that act like magnets, where like charges repel each other and opposite charges attract. Depending on what the atoms are will decide just how strong those attractions or repulsions are, and exactly what direction the atoms will feel it in.

So then, by using Newton's Laws of Motion, most specifically F = ma, we can use those forces in combination with the masses of our atoms, to figure out exactly how the atoms are going to move. The computer will then just repeat these calculations on every single atom in the system (usually tens to hundreds of thousands!) for every "timestep" in our simulation, which is usually on the scale of femtoseconds!

So where does the biology come into play? Well biology tells us what proteins specifically look like and how they act. So we're able to use our knowledge of them to figure out how to make our calculations of motion make the most sense by applying what we call "force fields". We also rely on biochemists and structural biologists to tell us what our proteins look like in the first place! Once someone else solves the structure of a protein, they can upload the structure to an online database that we can download and put through our calculations.

But what kind of information can we get from these simulations? Figuring out how a protein moves is all well and good, but it doesn't seem like the most useful thing, right?

Well, there's all kinds of things we can learn! By studying how the proteins move, we can figure out how and why they interact with different things. In "real life" proteins don't exist by themselves in a little vaccuum, they're surrounded by water, other proteins, ions, small molecules, and all sorts of other things. And all of those things are incredibly important to the protein acting the way it's supposed to! So essentially we put these proteins into the computer with water, and as many other things as we want. And then we can see how they interact with each other, and what happens if we make tiny changes.

What do I do specifically? I'm trying to figure out why specific proteins prefer interacting with certain molecules and not others. By figuring out what it is that causes this preference and then use that information to predict what other molecules that protein will like.

Image placeholder

Kira Mills

Ph.D. student using computers to study proteins!