SwiftPharma, a leading innovator in the research & development of recombinant proteins derived from a proprietary plant-based production system, launches a revolutionary AI-based computational antibody drug discovery platform in which artificial intelligence (AI), big data, machine learning, and phage display techniques are combined to predict antibody-antigen binding and provide antibody drug candidates with high specificity and affinity, at unprecedented speed.
“Our platform for “Optimized Discovery of Immunotherapeutic Novel Antibodies through Artificial Intelligence”, termed ODIN.AI, plugs into almost every aspect of our discovery process consolidating three highly efficient steps”, Jeroen HOFENK, SwiftPharma’s founder and Chief Science Officer, said.
First, an AI-algorithm is developed that draws from an extensive proprietary dataset, amalgamating information from publicly available sources on antibody usage, scientific journals, collaborations with partners, and patent databases. This algorithm showcases an exceptional ability to generate, screen, optimize, and evaluate hundreds of millions of therapeutic antibodies.
HOFENK elaborates, "Our approach employs a model rooted in natural language processing. However, instead of analyzing written language, it operates on antibody sequences. Essentially, the processor generates every computationally conceivable variation of an antibody, empowering SwiftPharma to construct antibody libraries that encompass the entire spectrum of human sequence diversity."
In step two, prediction of antibody-antigen binding solely based on protein primary structures are elucidated. Moreover, it leverages our advancements in plant-based cell-free protein synthesis technology to facilitate the high-throughput generation of a proprietary antibody cluster. These antibody sequences are meticulously designed, a process that encompasses enhancing various facets including affinity, solubility, cross-reactivity, manufacturability, immunogenicity, specificity, and stability into one multi-factorial design space.
“Our libraries comprise numerous antibody sequences, ranging from hundreds of thousands to even millions. The process involves taking one or more engineered epitopes and incubating them with this extensive library. These engineered epitopes are not only cost-effective but also rapidly produced using our plant-based cell-free expression process. This enables us to selectively identify antibodies that bind to these epitopes. In our pursuit of multi-specific antibodies, we acknowledge the challenge. Perhaps, out of a billion antibody sequences, only 10 may bind to a specific epitope. Using traditional methods, uncovering these 10 antibodies within such a vast pool would be exceptionally arduous. However, by concentrating on these engineered epitopes, we can efficiently isolate these elusive candidates," HOFENK continues.
Third, SwiftPharma has pioneered a plant cell display system for these antibody libraries, a pivotal advancement in the scaling-up of antibody production within plants. To achieve this, SwiftPharma integrates whole plants into the early stages of the recombinant antibody discovery process.
"In particular, we've engineered plant cells to emulate B cells by displaying IgG antibodies. We then employ single-cell sorting techniques to identify those cells that express candidate antibodies most optimally. This approach greatly expedites enhancements in our already efficient downstream process and enables us to promptly assess the feasibility of an antibody. Consequently, we avoid committing significant time and resources to a path only to realize six months down the line that an antibody may not meet our standards", HOFENK elaborated.
To maintain a streamlined and agile platform, SwiftPharma actively recruits "bilingual" scientists who possess expertise in both biology and machine learning or data science. According to Johan VAN HAVERMAET, Co-Founder and Chief Operating Officer of SwiftPharma, this integrated approach sets SwiftPharma apart and contributes to its steep success.
"Our distinctive strength lies in the fusion of both realms within a single individual. This type of interdisciplinary scientist is paramount for companies leveraging machine learning. It's intriguing to observe that many first-generation AI-enabled firms still grapple with bridging the gap between these two disciplines. Big Pharma, in particular, faces challenges in this regard", VAN HAVERMAET explained.
VAN HAVERMAET continued, "By intentionally designing our team in this manner, we prevent the formation of silos. Explaining the nuances of a biological experiment to someone grounded in binary code can be challenging unless they have a background that integrates both worlds. Fortunately, scientists with this profile are now available, even if they were scarce a few years ago."
He emphasized that this philosophy is ingrained in SwiftPharma's workspace, with one part dedicated to computational work in dry labs and the other focused on wet lab biochemistry and molecular biology. The meticulous design ensures that each lab bench represents a crucial step in the drug discovery process, seamlessly feeding into the next stage.
SwiftPharma is a Belgium-based privately held biotechnology company dedicated to developing next generation, animal-free, and affordable recombinant proteins by transforming plants into mini bioreactors. The company is on a mission to bring a paradigm shift to multiple industries by serving as a bridge between high unmet industry needs and the significant advances being made in sustainable biomanufacturing. For more information, visit www.swiftpharma.eu