Article content
New benchmark confirms LENSai’s ability to accurately predict binding on 17 previously unseen antibody-antigen complexes, achieving near-crystallography precision without prior training data.
Article content
AUSTIN, Texas — ImmunoPrecise Antibodies Ltd. (NASDAQ: IPA) (“IPA” or the “Company”), an AI-powered biotherapeutics company, today announced a new validation study supporting the generalizability of its proprietary epitope mapping platform, LENSai, powered by IPA’s patented HYFT® technology. The newly released benchmark shows that the platform consistently delivers high predictive performance, even on complexes not used during training.
Article content
Article content
Article content
“It’s generally assumed that AI can only make accurate predictions if it has seen similar data before,” said Dr. Jennifer Bath, CEO of ImmunoPrecise. “But this benchmark proves otherwise: LENSai accurately mapped antibody binding sites on entirely new antibody – protein complexes-none of which were used in training. Not the antibodies. Not the targets. Not the complexes. And the predictions aligned with wet-lab results. This is a major breakthrough in generalizing AI for therapeutic discovery, made possible by our proprietary technology, which captures functional meaning instead of memorizing shapes. It shows that AI doesn’t always need massive data to be powerful and accurate – it just needs the right kind.”
Article content
Top Stories
Article content
LENSai Epitope Mapping uses artificial intelligence to pinpoint where antibodies are most likely to attach to disease-related proteins – helping scientists design better treatments faster. Unlike traditional methods that take months and require lab work, LENSai delivers results in hours – using just the digital sequences – cutting timelines, eliminating the need to produce expensive materials, reducing guesswork, and unlocking faster paths to new treatments.
Article content
In a new benchmark study, LENSai was tested on 30 antibody-protein pairs, 17 of which the platform had never seen before. Despite having no prior exposure to these molecules, LENSai achieved prediction scores nearly identical to those from its original training data. This score, known as AUC (Area Under the Curve), is a widely accepted measure of accuracy in computational biology.
Article content
The consistent performance on entirely new, unseen complexes confirms that LENSai’s artificial intelligence can reliably analyze and predict antibody binding – even for molecules outside its training set. This breakthrough demonstrates LENSai’s power to generalize across diverse biological structures, making it a valuable tool for accelerating real-world drug discovery.
Article content
Why This Benchmark Matters
Article content
In the new study, LENSai delivered high accuracy results on 17 antibody-protein complexes the platform had never seen before as it did on familiar training examples – proving true generalization, not memorization. Because no new wet-lab work or x-ray structures were required, researchers gain speed, reproducibility, and major cost savings, while freeing scarce lab resources for confirmatory or downstream assays.
Article content
Article content
What It Means for Partners and Investors
Article content
With LENSai already embedded in collaborations across big pharma and biotech, ImmunoPrecise is scaling access through secure APIs and custom partnerships. The platform helps researchers compress discovery timelines, reduce risk, and unlock previously unreachable targets – positioning the company and its investors at the forefront of AI-driven antibody therapeutics.
Article content
For more technical detail and full benchmark results, explore two complementary case studies that illustrate the power and flexibility of LENSai Epitope Mapping.
Article content
The first highlights performance on a
Article content
“seen” target
Article content
, where the system was trained on related data. The second – featured in this press release – demonstrates LENS
Article content
ai
Article content
’s breakthrough ability to accurately map binding sites on a completely
Article content
“unseen” target
Article content
, with no prior exposure to the antibody, the antigen, or their structure.
Article content
Article content
These examples underscore how LENSai performs both in well-characterized systems and in novel, previously untrained scenarios—validating its generalizability and real-world readiness.
Article content
About ImmunoPrecise Antibodies Ltd.
Article content
ImmunoPrecise (NASDAQ: IPA) is a global leader in AI-powered biotherapeutic discovery and development. Its proprietary HYFT technology and LENSai™ platform enable first-principles-based drug design, delivering validated therapeutic candidates across modalities and therapeutic areas. IPA partners with 19 of the top 20 pharmaceutical companies and is advancing next-generation biologics through data-driven, human-relevant models.
Article content
Forward-Looking Statements
Article content
This press release contains forward-looking statements within the meaning of applicable United States and Canadian securities laws. Forward-looking statements are often identified by words such as “expects,” “intends,” “plans,” “anticipates,” “believes,” or similar expressions, or by statements that certain actions, events, or results “may,” “will,” “could,” or “might” occur or be achieved. These statements include, but are not limited to, statements regarding the performance, scalability, and broader application of the LENSai™ and HYFT® platforms; the generalizability of the Company’s AI models to novel therapeutic targets; the role of AI in accelerating antibody discovery; and the Company’s future scientific, commercial, and strategic developments.
