What You Should Know:
– The Lieber Institute for Brain Development (LIBD) is expanding its technology capabilities on Amazon Web Services (AWS), migrating its IT infrastructure to fully utilize AWS generative AI and compute services to advance its research.
– A recipient of the 2024 AWS IMAGINE Grant, the Institute plans to develop a new tool called GRAPE that combines generative and predictive AI to find new, more effective treatments for brain disorders such as schizophrenia.
Harnessing Big Data to Unravel Brain Complexity
Building on AWS will enable the Institute to store its massive collection of genomic and other data in the cloud, giving its scientists access to expansive computing power and advanced AI capabilities. Cloud storage will also make it easier for Lieber Institute scientists to collaborate and share data with researchers worldwide.
The scale of the data is immense—the human genome contains three billion letters, and the human brain contains 170 billion cells. Analyzing this requires leading-edge tools. Lieber Institute scientists are collaborating with AWS solution architects to create new, custom AI applications that make deep learning accessible to all the Institute’s scientists, regardless of their coding expertise.
“AWS’s AI capabilities give the Institute the speed, security, and scale the organization needs to drive research innovation that will radically change outcomes for people affected by brain health issues,” said Jeff Kratz, Vice President of Nonprofit and Public Sector Industries at AWS.
GRAPE: A New Frontier in Generative AI for Drug Design
A cornerstone of this collaboration is the development of a new generative AI tool called Generative Reinforcement Alignment of Predicted Expression, or GRAPE. The project is funded in part by the AWS IMAGINE Grant, which awarded the Lieber Institute up to $200,000 in unrestricted funding and $100,000 in AWS Promotional Credits.
GRAPE aims to address the limitations of existing drugs for conditions like schizophrenia, which often target only a handful of the hundreds of risk genes involved. The new AI tool will design novel molecular structures for potential drugs based on the known gene expression patterns of these complex disorders. Uniquely, GRAPE will combine generative AI to design new drugs with predictive AI to evaluate their effectiveness, aiming to treat the root cause of the disease.
“To make the most of AI, we need powerful supercomputing resources, and we also need new techniques that enable us to stretch these resources as far as possible,” says Dr. Michael Nagle, a Staff Scientist at the Institute. “GRAPE represents an opportunity for us to maximize our impact for as many patients or as many conditions as possible.”
The collaboration is already yielding significant efficiency gains. Dr. Frank Piscotta, a Staff Scientist at the Lieber Institute, uses a technique called “cell painting” to identify new drug targets. This process generates very large datasets, and analysis that once took almost a week can now be completed in about half an hour on AWS.
