Machine learning is accelerating real-life antibody generation in a virtual lab.
An innovative tool using computational biology can improve the human body’s capability to rapidly design effective and potent antibodies against the antigen that triggered them. Naturally created antibodies identify invading pathogens as foreign entities and flag them for destruction and removal; however, many of these antibodies do not prevent disease and only bind weakly to pathogens or toxins.
To improve the body’s immune reaction, high-affinity antibodies can be cloned and manufactured to provide short-term immunity in pre-exposure situations or to be used therapeutically during illness to help suppress tissue damage, allow the host defenses to catch up, and eventually overcome the infection.
The Defense Threat Reduction Agency’s (DTRA) Chemical and Biological Technologies Department in its role as the Joint Science and Technologies Office (JSTO) for Chemical and Biological Defense invested with researchers at the Massachusetts Institute of Technology (MIT) to design a technology capable of predicting antibody-antigen interactions and improving antibodies, which are an invaluable tool in DTRA JSTO’s arsenal against biological weapons.
Antibody therapies are sometimes the first treatments available for new biological threats, as was the case in the beginning of the COVID-19 pandemic. However, antibody therapies developed from the blood of immune humans or animals take several months to produce, test, and formulate before beginning clinical trials. Typically, highly active human or animal antibodies are isolated and improved through laboratory experimentation to enhance efficacy by increasing binding affinity and functional activity. However, these efforts take a long time to develop, and in a nonemergency setting, such development often takes several years before clinical trials can start.
The best antibodies fit their targets tightly like a lock and key: the better the fit, the better the binding affinity, the more effective the response. Previously, an antibody’s affinity was improved through antibody maturation experiments, but these assay-based laboratory experiments are costly, labor-intensive, and often do not adequately improve the binding affinity of a candidate antibody in a single round of experimentation. To overcome this constraint, researchers use computational biology for computer-based experiments to improve antibody binding affinity through mathematical modeling. Although these computational tools delivered improvements over previous methods, several limitations remained including reliance on pre-existing template features, sequences, or structures; systems capable of either designing or docking antibodies, but not both; and large computational times and costs.
As a remedy, the MIT researchers created an innovative tool called Hierarchical Structure Refinement Network (HSRN) to improve the capability to rapidly design effective and potent antibodies. HSRN uses a multiscale, spatial representation of antibody-target interactions at the atomic level that enables it to:
Date Taken: | 04.11.2023 |
Date Posted: | 04.11.2023 17:05 |
Story ID: | 442432 |
Location: | FT. BELVOIR, VIRGINIA, US |
Web Views: | 504 |
Downloads: | 0 |
This work, Docking by Design, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.