You are here
Using HPC to Find Solutions to Heart Problems
You could say that Sergei Noskov is in the movie business. With the help of his research team, the Assistant Professor at the University of Calgary’s Institute of Biocomplexity and Informatics and Department of Biological Sciences is creating movies, or computer simulations, of molecular models of proteins. These movies are used to study the ways proteins interact with various drugs on a molecular level.
“In our body we have special proteins called ion channels and these channels are part of machinery regulating the concentration of ions inside and outside of the cell and control the cell signalings,” says Noskov. “We look at one molecule with a known structure, such as a potassium channel or membrane transporter, and, then, we try to understand first of all what makes it selective for a specific kind of ion, then, how particular ions change functions of this protein.”
The goal of these experiments is to learn about cardiovascular conditions, how they develop and if they can be treated on a molecular level. Occasionally, certain drugs will block the ion channels flowing to the heart and that can cause an irregular heartbeat, also referred to as drug-induced cardiac arrhythmia. For example, a patient takes a drug that is supposed to decrease the harmful effects arthritis will have on his knee, but instead, that drug gets blocked in the cardiac channel. That patient would most likely end up with cardiac arrhythmia.
“This is a side effect for many drugs. They tend to block those channels due to a yet unknown mechanism, so we’re trying to understand how it is happening and more importantly how to prevent this inadvertent blockade,” says Noskov.
Once Noskov and his research team understand the cardiac channel blocking mechanism, they can work towards creating better drugs that will potentially prevent blocking. Most of this research is done using powerful computing resources provided by WestGrid.
“What we do is basically put the protein and the membrane in water and we describe the movement of every atom, because we know the initial position and we know the way the atoms interact with each other, so we basically produce a kind of movie,” says Noskov.
From that movie, the researchers can study a lot of things because it displays the exact position of an atom at every moment of time. It is a simulation of thousands, if not millions, of atoms moving simultaneously.
So just how much computational power does it take to run simulations of thousands, if not millions, of atoms moving simultaneously?
“You can’t run the jobs we’re interested in on desktop computers or even small clusters of computers, so the only solution for us was WestGrid and its high performance computing where you can use 256 or 512 processors at once,” says Noskov.
WestGrid resources are used for Noskov’s big experiments. His team has its own local computer cluster, but it is only used for routine computation, such as pilot simulations. When Noskov’s team wants to run simulation experiments that require a great deal of high performance computing power, they turn to WestGrid.
“WestGrid provides high performance computing power, plenty of memory and fast interconnect,” says Noskov. “We strongly rely on its resources.”
The protein simulations are run on WestGrid infrastructure at the Universities of Calgary, British Columbia and Alberta. The output and trajectories of the protein movies are then downloaded onto local machines and analyzed.
“It’s simple, because everything can be done online,” says Noskov. “We can log into WestGrid infrastructure from here in our labs and then work virtually.”
The output and trajectories not used immediately are stored for later analysis. Because of the success of the protein simulation experiment, Noskov believes they will continue with simulation experiments to solve other problems.
“Right now, we are able to handle a chunk of the membrane and the protein, but we’re expanding our analysis, zooming out to models of cyber cells or zooming in to include electronic effects of the whole story. This will create a better understanding of how all parts work together from atoms and electrons to real cells and organisms. I think this quest will not end in the next 20 or 30 years, so we’ll still be in need of large centralized resources and storage.”