Fueling a Revolution in Cancer Drug Development with AI and HPC

By Dr. Michael J. McManus, Senior Health & Life Sciences Solution Architect at Intel

Framework for Insights

As an organic chemist by training, I don’t use words like “revolutionary” lightly. But I believe advances in high performance computing (HPC) and artificial intelligence (AI) are fueling a revolutionary era of computational biomedicine, predictive biology, and personalized healthcare.

Powered by Intel® technologies and united within Intel® Scalable System Framework (Intel® SSF), life science researchers are performing powerful simulations of biochemical processes. They’re delving into an exploding world of biological data that’s being captured from genome sequencers, lab instruments, patient medical records, and other data sources. And they’re using deep learning and other AI elements to analyze, explore, and understand their massive data sets in unprecedented ways.

Powered by Intel® technologies and united within Intel® Scalable System Framework (Intel® SSF), life science researchers are performing powerful simulations of biochemical processes.

The results are creating a new foundation for large-scale problem solving and breakthrough insights into the biological universe.  As AI, HPC, and other critical technologies move forward, I expect this revolutionary era to yield discoveries, innovations, and impacts that will rank with those of the industrial, technical, and digital revolutions that came before.

New Tools to Fight Cancer: Deep Learning, 3D Imaging, and More

At SC16 in Salt Lake City this month, Intel is highlighting this important and exciting area of high performance computing. We’re running five demonstrations that illustrate crucial activities where Intel innovations are transforming the drug development pipeline for cancer and other diseases. The demos depict life science challenges in areas ranging from structural biology to genome analytics, all running on a single Intel SSF certified cluster with the latest Intel® Xeon® and Intel® Xeon Phi™ processors, as well as Intel® Omni-Path architecture and Intel® Enterprise Edition for Lustre* software.

  • Visualizing protein structures. An important step in developing new cancer therapies is to visualize and examine the structure of the proteins that direct cancer cells’ behavior in the body. We show how deep learning and advanced molecular imaging technology are enabling researchers to examine the shape of individual proteins with unprecedented clarity.
  • Examining protein movement. Knowing how a protein moves and interacts with other molecules is crucial to developing new compounds that will “dock” or fit smoothly with a protein and perform a designed action on it. We use 3D molecular dynamics simulations to analyze, visualize, and manipulate a high resolution 3D molecular structure of a protein, simulate its movement and interaction—and gain clues to its role in disease and treatment.
  • Identifying effective compounds. Deep learning algorithms can help researchers analyze a virtual library of chemical compounds and predict the best possibilities to target a protein of interest. Scientists from Kyoto University say this solution can save millions of dollars per research project and cut years off the time needed to perform the analysis.[1]
  • Identifying patients for clinical trials. Once a new targeted treatment is ready for clinical trials, researchers want to test it on people who possess or lack the protein structure that the drug targets. We show the use of data-intensive genome analytics to help determine if a drug is likely to be effective for a particular patient. This is a promising way to deliver benefits to cancer patients while reducing the time and expense of clinical trials.
  • Monitoring a cancer treatment’s effectiveness. Radiologists and oncologists analyze MRI scans to monitor tumor growth and see how well a drug or other treatment is working. This demo produces high fidelity interactive visualization of a patient’s brain tumor with stunning clarity—without GPUs. By comparing visualizations of the tumor at different times during the treatment, doctors can spot subtle changes and confirm or change treatment options with more confidence.

Cutting Costs and Complexity: Diverse Workloads on a Single Cluster

These demos show a few ways Intel is empowering life scientists to drive the revolution in biomedical computing—in this case, to speed the development of new and more effective cancer treatments and help get them to patients more rapidly.

The demos also highlight the practical benefits Intel SSF provides for day-to-day biomedical research and other areas of high performance computing.  By running diverse workloads on a single architecture and avoiding the programming hassles of GPUs, pharma researchers and others can reduce the costs and complexity of their technology infrastructure. With superb performance and throughput, researchers can also tackle larger problems and find answers faster. And with the flexibility of Intel SSF, research organizations can deploy Intel SSF clusters at virtually any size, to match their workloads and their budget.

AI for Good: Join Us in Making a Difference

If you’re attending SC16, stop by Booth 1819 between 11/14/16 and 11/17/16 to see how AI and HPC technologies from Intel are enabling the biomedical computing revolution, including personalized medicine. During your visit, you can also help raise money to expand precision medicine for advanced cancer patients. After experiencing the demo, get your personalized Pikazo*photo taken, share it on Facebook, Twitter or Instagramwith #IntelAIforGoodPromo.  Limit one (1) social share per platform per person.  Intel will donate on your behalf $10 for each social share to help fund more availability and access to precision medicine, up to a total combined maximum donation of $25,000.

Monies raised will go to Oregon Health & Science University (OHSU) Knight Cancer Institute through Consano, a non-profit, crowd-funding website that’s part of a new movement to connect individuals directly to specific medical research projects. By harnessing the power of a crowd, Consano aggregates donations so the public can directly choose and support high quality medical research. Intel’s #IntelAIforGoodPromo donations will go to a new precision medicine trial at OHSU’s Knight Cancer Institute with the single-minded goal of bringing benefit to each advanced cancer patient enrolled.

With Intel’s donation on your behalf, this social promotion will help expand patient access to OHSU’s precision medicine research project, bringing targeted therapies to more people needing a personalized solution to their cancer.

 

 

[1]  Kyoto University Machine Learning with Intel® SSF. http://www.intel.com/content/www/us/en/analytics/machine-learning/kyoto-university-machine-learning-with-intel-ssf-video.html