We call it ‘The Big C’, but cancer is not just one disease.
Breast cancer, for example, consists of over one hundred different diseases. Every cancer, even those from the same site on the body, behaves differently. That means that treating cancer—and curing it—is a long way from one-size-fits-all.
Which is why researchers are excited about the possibilities of personalized medicine. For cancer, this means that instead of a traditional chemotherapy protocol, with its devastating side effects, a patient’s tumors are subjected to DNA analysis.
“Nowadays, we perform DNA analysis on patients’ tumors, and the patients are then grouped based on the specific cancer mutations that are found,” says Zoe Cournia, a researcher in pharmacology at the Biomedical Research Foundation at the Academy of Athens (BRFAA).
These DNA mutations can alter the behavior of proteins in our body and cause cancer. For example, perhaps the first five patients tested have the same cancerous mutation, the next ten have another mutation, and the next fifty have yet another. Once you identify the different cancer mutations in these groups, you then know which proteins are malfunctioning and can create drugs specifically to target each of these mutant proteins.
Cournia uses computational biomolecular modeling to study protein structure and function, particularly mutated proteins implicated in cancer.
“I investigate the mutated proteins cancer patients have in their bodies and design small molecules as candidate drugs to stop the function of the mutant proteins,” says Cournia. “This technique has fewer side effects compared to conventional chemotherapy because it specifically targets the mutated proteins rather than all fast-multiplying cells.”
She continues, “With targeted therapy we are able to create a drug that only goes to the cancerous cells and stops the action of the mutant protein, but doesn’t affect the healthy cells.”
For Cournia, doing this work is very important to her. She always knew she wanted to be a chemist, but it wasn’t until she was working on her PhD that she became interested in computer-aided drug design.
“I soon realized I wanted to do research in an area that would have an impact on human life,” says Cournia. “Something that has the capability of producing products that can be used by people to improve their lives.”
Faster drug design
Computation speeds up the process of drug design, allowing researchers to identify likely drug candidates much faster than if they were testing them in the lab.
This sample of clinical triple-negative breast cancer is stained for bone morphogenetic protein-11 (red); the Golgi marker GM130 (green); glycosylated proteins (white); and nuclei (blue), Cancer cells exhibit a wide variety of molecular structures and characteristics, posing difficult challenges for researchers trying to find better ways of managing the disease. Courtesy National Cancer Institute/University of Virginia Cancer Center.“If I were to do this research experimentally, I would screen 400,000 compounds and only two would be active,” says Cournia. “That’s a success rate of only .00004 percent. And you are using all those experimental assays, which are very expensive and time-consuming, to come to those two molecules.”
But if researchers perform the initial analysis on a computer, they can narrow the field significantly. They will then perform lab tests on perhaps only twenty compounds instead of 400,000, reaching those two successful molecules with significantly less time and expense.
The systems that Cournia studies encompass hundreds of thousands of atoms. To increase the time and length scales of the biomolecular processes in order to effectively study these large systems requires thousands of computer cores.
“For one of our simulations, it would take 490 years to perform on a desktop what we can do in one month with HPC [high-performance computing],” says Cournia. “So it is absolutely crucial to have HPC resources to bring these products into the market faster.”
Cournia also looks forward to the advent of super-fast exascale computers.
“At the moment, we are using classical molecular dynamics simulations which have a lot of approximations,” says Cournia. “But if we could use quantum chemistry codes at exascale, we will be able to remove these approximations and go to very exact solutions.”
She suggests that using the current influx of big data to train artificial intelligence algorithms will also improve accuracy. However, she’s not letting resource availability slow her down.
Cournia and colleagues at BRFAA were recently granted a US patent application for a molecule that is very potent against pancreatic cancer, which is over expressing the myc protein. They hope to be able to progress to clinical trials very soon. Another forthcoming patent application will be for a class of molecules that target a mutant protein found in twenty percent of breast cancer patients.
Spreading the word
In 2016 Cournia was the first winner of the PRACE Lovelace Award for HPC. Along with the honor of having her work recognized internationally, Cournia says the award helped her to share the importance of high-performance computing with both the general public and other scientific disciplines.
“I was able to communicate to young women and girls that science is a brilliant profession with lots of opportunities. I was able to show that everyone is capable of doing science—it’s not gender dependent,” says Cournia.
“But also that science is something that we need. It is taxpayer’s money that we are using after all. So it’s important to show that the money isn’t wasted, that it’s given to produce products that have the potential to save millions of lives.”