“Form Follows Function”: Rethinking Therapeutic Antibody Design

The central gap is that antibody development has historically been very good at generating blockers, but less able to systematically design antibodies around more complex biological functions.

Traditional antibody approaches often start with a target structure or binding event. That can be powerful, but it does not always capture what matters most therapeutically: what happens after binding. Biology is dynamic; proteins move, signal, interact with cofactors, and behave differently in living cellular systems. If we only optimize for whether an antibody binds, we may miss the mechanism or functional outcome needed for a medicine to work.

We are focused on designing antibodies around intended biological activity from the outset. That includes agonism, biased signaling, multi-target engagement, multifunctional activity, and selective control of biology.

“The goal is to expand what antibody medicines can be designed to do.” — Angela Hwang.

That matters because many high-value areas of biology (G protein-coupled receptors (GPCRs), ion channels, and complex signaling pathways) have been difficult for traditional antibody approaches to address. As a result, small molecules and peptides have dominated these areas. But antibodies offer important advantages, including specificity, durability, long half-life, and manufacturability. Functional antibody design can bring those advantages to diseases where treatment requires more than stopping biology.

Metaphore’s platform changes the starting point for antibody design. Traditional workflows often begin with affinity or structure. They ask where an antibody binds and how tightly it binds, then try to engineer the molecule from there.

“We start with function, asking ‘What biological outcome do we want the antibody to drive?’ Then, we generate data in living systems to understand the features that produce that outcome.” — Angela Hwang.

The platform combines large-scale live-cell functional screening, functional deep mutational scanning, and AI/machine learning. These tools allow us to measure how protein interactions behave in a more physiologically relevant context and connect those interactions to functional readouts. We then use computational models trained on proprietary functional data to design antibodies with intended activity and drug-like properties. This is important because many desired antibody functions cannot be inferred from a static structure alone.

In practical terms, the platform helps us move from discovering what binds to designing what works. It also allows us to bring the advantages of antibodies into target classes and indications where small molecules and peptides have historically been the default, but where an antibody’s specificity, half-life, and durability could offer meaningful therapeutic advantages.

Throughput is only valuable if the data remain biologically meaningful. The goal is not to generate more data for its own sake. The goal is to generate the right data, at sufficient scale, in systems that preserve the functional biology we care about.

That is why the platform is built around live-cell systems and functional readouts. We use high-throughput and automation, but we apply them to questions of mechanism and biological outcome, not binding alone. Automation allows us to run experiments at scale with greater consistency, while functional readouts help ensure that the data remain connected to the biology we are trying to control.

The balance comes from integrating experimental biology and computation in an iterative loop. Large-scale functional datasets help our models identify patterns that drive activity, bias, selectivity, or multi-target engagement. Molecules are then designed, built, and tested experimentally—the results feed the next cycle. Each cycle improves the platform’s ability to prioritize molecules with both the desired function and the properties needed to become medicines.

Maintaining biological relevance also requires discipline in how programs are selected and advanced. We focus on areas where the biology is well-grounded, the therapeutic hypothesis is clear, and functional antibody design can create meaningful differentiation.

Some of the greatest opportunities are in diseases where the biology is validated, but the right antibody mechanism has not been achievable with traditional approaches.

Metabolic disease is one example. Many metabolic pathways are coordinated across receptors, tissues, and signaling pathways. There may be opportunities for antibodies that can activate or coordinate multiple pathways with durability and precision.

Immunology is another important area. Immune diseases often involve complex signaling networks, cell states, and feedback loops. To drive better outcomes for patients, we may need medicines that can tune immune activity or engage multiple nodes in a pathway, rather than broadly suppressing the immune system.

There are also important opportunities in gastrointestinal disease, pain, vascular disease, GPCR biology, and ion channels, where antibody medicines have historically been limited.

More physiologically relevant models matter because they allow us to observe biology closer to how it behaves in living systems. They can help reveal mechanisms that static structures or binding assays miss. For antibody therapeutics, that means a better chance of designing medicines around the functional outcome patients need, not just the target we want to bind.