Fighting Cancer With Supercomputers

Chemotherapy and Drug Design

New drugs can cost billions of dollars to develop and take decades to reach the marketplace. Supercomputers can speed up the process by finding new uses for approved drugs.

TACC's Lonestar5 supercomputer virtually tested more than 1,400 small molecule drugs approved by the Food and Drug Administration to see if they could be used to treat cancer. Shuxing Zhang, associate professor of experimental therapeutics at MD Anderson Cancer Center, and graduate student Zhi Tan found that the parasite-fighting drug mebendazole could effectively bind to and inhibit the activity of TNIK, an enzyme that plays a key role in cell signaling related to colon cancer.

"Such advantages render the possibility of quickly translating the discovery into a clinical setting for cancer treatment in the near future," Zhang said.

---

Immunotherapy

Immunotherapy supercharges the body's natural defenses to fight cancer, but not every immune therapy works the same way on every patient.

Researchers from Wake Forest School of Medicine and Zhejiang University in China turned to TACC's Stampede1 to help develop a new mathematical model that represents the interactions between prostate tumors and common immunotherapies.

By doing millions of simulations to predict tumor responses to treatments, the researchers found that the depletion of T cells and the neutralization of the signaling protein Interleukin 2 can have a stronger effect when combined with androgen deprivation therapy and vaccines.

Said lead researcher Xiaobo Zhou: "TACC provides an important assistance for discovering clinically meaningful and actionable knowledge across highly heterogeneous biomedical big data sets."

---

Proton Therapy

Proton therapy causes less damage to surrounding tissues than the commonly used X-ray radiation therapy for tumor irradiation. But the proton beam needs pinpoint accuracy and precise calibration.

Mayo Clinic researcher Wei Liu used Lonestar5 and Stampede1 to develop a model for treatment planning that is more accurate and better at sparing organs than radiation therapy.

"It's very computationally expensive to generate a plan in a reasonable timeframe," Liu said. "Without a supercomputer, we can do nothing."

---

Cancer Diagnostics

Manual breast exams and mammograms are currently the most effective and widely-used techniques for early detection of breast cancer. Unfortunately, manual breast exams only provide information about the site where the force is applied, and mammograms (breast X-rays) expose patients to radiation and produce many false positives.

Researchers at the Rose-Hulman Institute of Technology developed an electro-mechanical device that gently indents breast tissue in various locations and records the tissue surface response. This data is then converted into detailed 3D maps which can be used to identify suspicious, stiffer sites for further testing.

They used Stampede1 to map the distribution of stiffness in a given tissue to find which tissue stiffness maps best match the response they see in testing.

"This system has the potential to significantly increase the early detection of breast cancer with no unnecessary radiation, essentially no risk, and with little additional cost," said Lorraine Olson, lead researcher.

---

Surgery

Surgical removal of cancer cells is risky. Removing too little of a tumor can lead to a relapse; too much — especially in a critical area like the brain — can harm the patient.

A pioneering project performed minimally invasive laser treatment on a canine tumor without a surgeon. The project team from UT Austin and MD Anderson Cancer Center used TACC's advanced computing resources to develop an interactive system that plans, predicts, and dynamically alters the course of a laser treatment for cancer patients.

"The more data and images that can be acquired, the more confidence researchers and surgeons can have in planning surgical simulations," said David Fuentes of the MD Anderson Cancer Center.

---

Genomics

The human genome consists of three billion base pairs, so identifying a single mutation by sight simply isn't possible. Computers, on the other hand, are great at finding patterns in massive datasets and have been a boon to cancer researchers.

Researchers from UT Austin and the National Cancer Institute used Stampede1 to mine massive amounts of data from the Cancer Genome Atlas to identify genetic variants and patient subtypes.

They found a specific mutation in the protein FOXP1 that's associated with an aggressive type of lymphoma.

"This knowledge can be helpful in the development of more targeted therapies that seek to eliminate cancer at its origin," said UT Austin geneticist Vishy Iyer.

---

Patient-Specific Treatments

Scientists used TACC's Lonestar5 supercomputer to develop mathematical models of cancer that predict how the disease will progress in a specific patient.

A tumor's response to treatments such as chemotherapy is characterized by an equation that captures its behavior. Scientists at the UT Austin Center for Computational Oncology combined these equations with data from patients in a study to simulate tumor development.

"If you have a model that can recapitulate how tumors grow and respond to therapy, then it becomes a classic engineering optimization problem. ‘I have this much drug and this much time. What's the best way to give it to minimize the number of tumor cells for the longest amount of time?'" said Thomas Yankeelov of Dell Med at UT Austin.

Yankeelov's group is currently conducting a clinical study in Austin, Texas to predict, after one treatment, how an individual's cancer will progress. They will use these predictions to plan the course of treatment.

---

Artificial Intelligence

An emerging way that researchers are using high performance computing for cancer research is through the application of machine and deep learning.

Researchers from Tufts University and the University of Maryland, Baltimore County used TACC's systems to uncover the complex cellular communication networks that underlie cancer and cellular mutations, and to design methods to disrupt them.

"Getting a true understanding, given the complexity of the information, without some assistance from machine learning, is probably hopeless," said Michael Levin, of Tufts Univesity. "I think it's inevitable that we use machine learning to enrich scientific and biomedical discovery."

---