Welcome to the Theoretical Biophysics Group Open House. My name is Robert Brunner, and I am one of the research programmers in the group, and this is Melih and Zhu, two of the scientists here. Our goal today is to explain to you a bit about the work we do here, and hopefully entertain you a little.

The Theoretical Biophysics Group studies how biological molecules behave by simulating their behavior with computers. Knowing how these molecules work improves our understanding of what goes on inside the human body. You may have heard about the recent publication of the human genome. The human genome shows us how to make many of the protein molecules that make up people, but it doesn't show how those parts work together. What we do is to put these parts together inside a computer, and watch how the parts interact.

You might guess that this work requires a lot of computing power. We do advanced graphics with our visualization program VMD, which you'll see in a moment. Behind the scenes, we do complex computations with our simulation program, NAMD, which harness the work of dozens or even hundreds of computers, both in our group and at supercomputer centers across the country. Our work also involves a lot of interaction with other facilities and researchers across the country, so we are working on a project called BioCoRE, which is a collection of tools that use the Internet to assist scientists collaborating over long distances.

Now, Melih will use VMD to explain the interaction of a protein with a loop of DNA. After that, Zhu will talk about docking of molecules using an interactive tool called IMD. Finally, I'll show you some of the components of BioCoRE.

Lac Repressor as a Simple Genetic Switch

In this presentation we demonstrate the working of the lac operon in the bacterium E. Coli as a genetic switch that is turned on when lactose [sugar] is avaliable to the cell and needs turning off at the absence of lactose. The switch is controlled by a protein known as the lac repressor which binds the lac operon portion of the DNA at two points as shapes it into a loop. This bends the binding site for the RNA polymerase and therefore renders the gene for the lactose digestion mechanism unreadable.

The shape of the DNA-loop is missing from x-ray data on the lac repressor protein, which must be reconstructed by the efforts of the modeler. Kirchoff's late 19th century elasticity theory is used to reconstruct the loop, using a set of differential equations describing the elastic behavior of the DNA loop.

We also demonstrate how the CAP protein modifies the conformation of the loop. And illustrate by showing a box of water molecules around the system how a multiscale simulation is achieved by combining a full atom molecular dynamics study on the system with an elastic rod model enabling the use of a smaller water box. We conclude by reemphasizing that when sugar enters the system again, the lac repressor releases the DNA loop rendering it readable again.

This system provides us with a simple example of a genetic switch which should help us better understand genetic switches in higher organisms, such as humans.

Molecular Docking with IMD

Now I'm going to illustrate how to use our visualization and simulation tools to study biological process.

In order to study the system, we first have to visualize the biological molecules. The picture we see now is produced by our visualization software VMD. Here on the left is a lactose, the whole thing on the right is the lac repressor, the protein we just saw in last presentation. In order to study the dynamics of this system, we're doing molecular simulations on it, using our molecular dynamics software, NAMD. Now the simulation is running on our PC cluster, and the coordinates of every simulation step are sent to this visualization machine via network, then the visualization program displays the trajectories frame by frame, in real time, so we see the atoms are moving. Thus, our visualization program, simulation software, and network communication allow us to do interactive molecular dynamics study.

This example also represents an important process in drug design, namely, the docking of a ligand to a protein. In our case, this part is the binding pocket of the protein. It binds this lactose, and also undergoes some conformational change during the binding process. But this process happens in a time scale that's much longer than the simulation time people can afford in normal simulations, we cannot just wait and see it happen. So in order to study this process, we have to apply artificial force to accelerate it. But since the binding pathway is not clear beforehand, it's not easy to know what force to apply,

That's why we use this haptic device. With this device, we can apply force by our hand; in the meanwhile, we can feel immediate feedback from the simulation. In this way, we can manipulate the molecule interactively. Now I grab the lactose and exert force to pull it toward the binding pocket, and I'm feeling the feedback force from the running simulation. In this way I can use my sense of touch to tell which direction is easy to go. (Of course, the stereo feature of the visualization program makes this job much easier)

Now I stop the simulation, and all the trajectories of the whole system have already been recorded, so we can play back to see all that happened. (Our visualization program can draw the molecules in various views, and can give us informative pictures, for example, ...) If we draw the binding pocket in surface representation, we can see the conformational change more clearly. We can also analyze the trajectories and the force, using our theoretical model, to calculate some important quantities of this system.

Both the visualization and the simulation software, which I just showed, were developed in our group. They are the tools we use everyday for our biophysics research.

Collaborating with BioCoRE

You've seen a couple of examples of the work that researchers do here. Now I'd like to talk about a project we are doing that help them do that research more efficiently.

A great deal of our research is collaborative. I am one of several members of the group who is building tools that harness the power of the Internet and the World-Wide Web to make such collaboration easier. BioCoRE stands for the Biological Collaborative Research Environment. It is a set of tools, most of which run through standard web browsers, which assist people in collaborating on structural biology research. You start a BioCoRE session by logging in to the BioCoRE server.

BioCoRE provides an assortment of tools that many of you are familiar with, such as chat areas and message boards. The tools I will show you today are those that are specialized for structural biology research.

One type of collaboration with supercomputing sites which supply a lot of the computational horsepower we need to perform our simulations. Although we have a lot of computer power here, including a 32-computer Athlon-Linux computational cluster, about half of our processing is done at NCSA, or the San Diego and Pittsburgh Supercomputing Centers.

Our simulations might run for days at a time, so we'd like to be able to check up on them easily, to make sure they are running properly. We can use BioCoRE to submit a simulation to a number of computer facilities, as I'll show now.

After submitting the job, we can come back at any time and reconnect to the job to view its status. We can also have it generate images of the molecule's state so far, and have that returned to us for viewing. Since this all runs in a standard web browser, we can do this whether we're sitting at our desk or attending some conference on the other side of the country.

We are also developing tools to assist in collaborations between researchers. Our field often requires close contact between experimental researchers, who use various lab techniques to determine what atoms make up particular proteins, and theoretical researchers, who use the techniques we've been showing to understand the behavior of those proteins. Usually those researchers are located in different places around the world, so communication is difficult.

Another tool we've been developing is the capability of linking VMD sessions over the internet. Think about how a researcher might want to discuss a molecule with a collaborator at some other university. Currently, they mainly communicate by e-mail, phone, and occasional visits. However, pictures and images are important in understanding protein behavior, and these methods are really not efficient or cost effective for just looking at the molecules. So let me show you what we're working on now. Here I have VMD running, as you've seen in the previous demos. I've linked this VMD session with another one running on the laptop that Melih is using. Now, when I move the molecule, it is also rotated on Melih's screen, and Melih can do the same thing. Also, either of us can change how the molecule is displayed. Now we could use a BioCoRE chat, to discuss what we are looking at. I think you can imagine how this is a big improvement over e-mail.


That concludes our formal demonstration. Thank you for attending our open house. I hope you now have a little bit of an idea what we do, and how it might improve everyones lives by understanding how living things work at the atomic level. If you have any questions, please ask, and if a few of you would like to try out the haptic device, we have a few minutes for that. Again, thanks for coming.