The ambitious project, led by investigators at Vanderbilt University Medical Center, aims to use artificial intelligence technologies to generate antibody therapies against antigen targets of interest.
VUMC is awarded up to $30 million by the Advanced Research Projects Agency (ARPA-H) to build large-scale antibody antigen atlas, develop AI-based algorithms to engineer antigen-specific antibodies, and apply AI technology to identify and develop potential therapeutic antibodies.
ARPA-H is an agency within the US Department of Health and Human Services that supports transformative, high-risk, high-remuneration research to promote biomedical and health breakthroughs and benefit everyone.
Over the past decades, monoclonal antibodies have begun to play an important therapeutic role in a wide range of disease settings, but are only damaging the surface. Monoclonal antibody discovery can affect many different diseases that currently do not have therapeutic agents. ”
Ivelin Georgiev, PhD, Professor of Pathology, Microbiology and Immunology, Director of Vanderbilt Computational Microbiology and Immunology, and Project Chief Investigator
Traditional methods for antibody discovery are limited by inefficiency, high costs and failure rates, logistics hurdles, long turnaround times and limited scalability, Georgiev said.
“What we’re proposing is to address all of these major bottlenecks in the traditional antibody discovery process and make it a more democratic process. Here we can understand what the antigenic target is and generate monoclonal antibody therapy for that target in a very effective and efficient way.”
Antibodies are part of the immune system. They are proteins produced by leukocytes (B cells) that bind and inactivate antigens – target viruses, bacteria, and even our own cells. Antibodies are effective in preventing and treating viruses, cancer, autoimmune disorders, and other diseases.
To identify candidate therapeutic antibodies, researchers generally screen and test thousands of antibodies against antigenic targets, looking for “haystack needles” that bind and neutralize the target. Traditional discovery processes require specific types of biological samples. For example, to find antibodies against pathogens of infectious diseases, blood samples from people exposed to the pathogen and animal models are required. And if the pathogen mutates, the therapeutic antibodies may become ineffective.
“The computational approach no longer relies on access to biological samples and multiple screening cycles,” says Georgiev. “Before a variant occurs, you can simulate it and generate antibodies in advance.”
Georgiev and his colleagues are engaged in three tasks towards developing a computational approach for antibody discovery.
Generating unprecedented sized antibody antigen atlas and developing diversity of AI-based algorithms to extract information from antibody antigen atlas and generating AI-based algorithms to extract antigen-antigen-specific antibody proof empirical studies to apply AI technology to identify antibody candidates for antibody candidates
In the first task, researchers use a technique called Libra Seq (linking B-cell receptors to sequencing antigen specificity), which simultaneously enables high-throughput mapping of antibody antigen interactions between many antigens and B-cells.
“It takes a lot of data for the calculation method to work,” says Georgiev. “The size of available data for antibodies and antigens is lower than in other areas. This is one of the limiting factors when it comes to the development of AI approaches.
“When training algorithms for data that are currently present, most of them are for SARS-COV-2, influenza and HIV. These target algorithms may be accurate, but they are less likely to succeed in extrapolating to new targets.
Investigators aim to include hundreds of thousands, and potentially more than one million, antibody antigen pairs, compared to the approximately 15,000 pairs currently available from currently published data, providing unparalleled resources to researchers around the world.
The team has already advanced the second task of building a computational model, and it will be improved when you register it with the Antibody Antigen Atlas. In the third task, AI technology is applied to develop antibodies against cancer antigens and bacteria, viruses and autoimmune targets. They select one candidate antibody for preclinical development, including IND (New Drugs in Clinical Study) applications.
“Our project will provide a platform that can be used for a variety of different diseases, not just the specific goals that we are interested in,” Georgiev said. “Our team has spent years trying to discover antibodies to various indications, and it’s a very inefficient process with a lot of failures. If we can help change that, it will be a big deal not only for us, but for people with diseases where antibody therapy can make a difference.
“It’ll be difficult. It’s not an easy problem, but I think we have a good foundation for it, and we’ll do our best to make it work.”
The project collaborators are as follows: MD, MD, PhD, Sarah Croessmann, PhD, Eric Skaar, PhD, MPH, Maria Hadjifrangiskou, PhD, Jeremy Goettel, PhD, Vumc. Dr. Ted Ross, Ph.D. and Ph.D. from Cleveland Clinic. and PhD, Maria del Pilar Quintana Baron, University of Copenhagen. The Brock Family Center for Applied Innovation, a catalyst for moving translation research into the market, engages and supports the Georgiev team.
Vanderbilt University and VUMC share important resources for their projects. Vantage (Vanderbilt Technologies in Advanced Genomics), Accre (Advanced Computing Center for Research and Education), and FCSR (Flow Cytometry Shares Resources).
sauce:
Vanderbilt University Medical Center
