Biologics Brief: E4: From Discovery to Drug: What Antibody Developability Really Means

Published: August 24, 2026 | Category: Insights, Biologics Discovery

---

Finding a promising antibody is the beginning of a different story not the end of one. The question shifts from “does this antibody do something interesting?” to “can this antibody become a drug?” That is the world of developability, and it is where a lot of compelling discovery stories get complicated.

In Episode 4 of The Biologics Brief, Mosaic’s Chief Strategy Officer Tracey Mullen sits down with Stacy Capehart, Director of Data Sciences at Mosaic, to talk through what developability actually means, the most common ways good antibodies fail, which assays reveal each type of risk, and why timing is everything.

---

What Developability Actually Means

Developability is the collection of properties that determine whether an antibody can realistically move from discovery into development. A therapeutic antibody has to do more than bind its target, it has to be manufacturable, stable, soluble, formulatable, and reasonably well behaved in biological systems.

That includes expression level and purification profile, thermal stability, aggregation propensity, self-association and viscosity behavior, hydrophobicity, polyreactivity, and sequence liabilities that can affect stability or function over time. Discovery asks whether you found something that works biologically. Developability asks whether you found something that can survive the rest of the process and actually become a drug. Those are related questions, but they are not the same question.

A good binding curve does not mean you have a developable antibody. It means you have an antibody that is worth learning more about.

---

Why Good Antibodies Still Fail

Discovery teams that surface antibodies with the right biology have done exactly what they are supposed to do. But biology is not the only constraint a therapeutic has to meet. An antibody has to be produced at scale with a reproducible purification profile, remain stable during storage, tolerate the concentration required for dosing, and be formulated in a way that is compatible with the intended route of administration. It also has to behave in vivo, which is a very different context from a primary binding screen.

Some liabilities only become visible under stress conditions or at concentrations that never appear in early assays. An antibody might look fine in a binding assay but show self-association at higher concentrations. It might express well at small scale but present purification challenges later. It might be potent but carry a deamidation site that affects binding over time. It might appear specific against the intended target but show broader polyreactivity.

Developability is not one thing. It is a broader risk picture, and no single assay captures all of it.

---

The Major Failure Points

Aggregation occurs when antibody molecules associate with each other and form larger species, driven by partial unfolding, exposed hydrophobic patches, charge interactions, or local instability. Aggregation can affect product quality, manufacturing consistency, stability, and immunogenicity risk, and it can show up during expression, purification, storage, or when you first try to concentrate the molecule. The severity and conditions matter: aggregation under extreme stress at irrelevant concentrations is a different conversation than aggregation under routine handling at the dose concentration you need.

Viscosity and self-association become critical when high-concentration formulations are required, particularly for subcutaneous delivery. An antibody that behaves well at low concentration can become very difficult to formulate at the concentration needed for dosing. That is not a small issue when the route of administration is central to the product profile. Assays like AC-SINS can provide an early signal of self-association risk before enough material exists for formulation-relevant experiments, not the whole answer, but a useful early screen.

Polyreactivity, binding to multiple unrelated targets or surfaces, can complicate assay interpretation, create off-target binding concerns, affect pharmacokinetics, and increase the risk of undesirable tissue binding. Importantly, confirming that an antibody binds the intended target, not a closely related family member, is not the same as understanding non-specific binding behavior. Polyreactivity can hide in plain sight if the workflow is only designed to find positive binders, and potency does not cancel out non-specific binding.

Sequence liabilities such as deamidation, oxidation, isomerization, clipping, unpaired cysteines in CDRs, and predicted glycosylation sites can create heterogeneity, instability, or manufacturing concerns. Critically, the location of a liability determines its impact. A deamidation site in a framework region that does not affect function may be manageable. The same liability in a CDR that affects antigen binding over time is a product quality issue. Seeing a liability motif is not enough, biological context is required to understand the actual risk.

---

The Assay Panel: What Each Test Actually Measures

Developability assessment requires multiple orthogonal assays because each one captures a different dimension of risk. A clean result in one assay does not guarantee success, and a red flag in one assay is not automatically fatal. Multiple red flags across orthogonal assays, however, become much more concerning in combination.

At Mosaic, the core developability panel includes sequence liability analysis, size exclusion chromatography for percent monomer content and aggregation profile, AC-SINS for self-association propensity, melting temperature (Tm) for conformational stability, temperature of aggregation (Tagg) for aggregation behavior under thermal stress, and hydrophobic interaction chromatography (HIC) for hydrophobicity. Tm and Tagg are related but distinct, a molecule can unfold without immediately aggregating, and aggregation behavior depends on more than just the unfolding transition. Both are interpreted alongside the full panel, not in isolation.

Material requirements also matter. Earlier in discovery, high-throughput assays with limited material are appropriate. Later in the funnel, more detailed characterization requiring slightly more material becomes justified. Developability assessment should scale with the stage of the campaign, the goal is not to run every assay on every hit, but to generate decision-relevant data at each stage.

---

Why Timing Is Everything

The most common pattern is that teams delay developability assessment until after lead selection, after investing heavily in one or two candidates. That is one of the biggest avoidable mistakes in antibody development.

When developability data arrives after lead selection, it becomes a pass/fail gate on a molecule the team has already committed to. There is pressure to rationalize the problems. If the lead has a serious liability, the team may have to revisit backups, repeat characterization, engineer the molecule, and delay downstream work. A rescue mission is expensive, scientifically, operationally, and in terms of program momentum.

When developability data arrives while a panel of candidates still exists, the same information feels completely different. It becomes a ranking tool rather than a verdict. It helps the team choose better before they have narrowed too aggressively.

This is especially important when multiple candidates have similar biological profiles. If the top candidates are close on potency and specificity, developability can be what actually determines the best lead. The most potent antibody is not always the best development candidate. The best lead is the molecule with the strongest overall profile for the intended use.

---

What to Ask Before a Campaign Begins

If your team wants to avoid late developability surprises, here are the questions worth asking upfront:

• When does developability assessment begin, after lead selection, or at an earlier checkpoint while a panel still exists?
• Which assays are being used and why, and are they appropriate for the stage of the program and the material available?
• How does developability data feed into lead ranking, is it part of the decision framework, or a separate report generated after the fact?
• How are liabilities handled, is there a plan for engineering, formulation, backup selection, or risk mitigation?
• Is the workflow aligned with the intended target product profile, a program targeting high-concentration subcutaneous dosing requires a different risk assessment strategy than one targeting IV dosing at lower concentration?
• How broad is the candidate panel when developability first comes in, because the value of developability data is highest when options still exist?

Developability data does not make decisions. It informs trade-offs. The goal is not to run every assay immediately or to treat every liability as a stop sign. The goal is to not be surprised by liabilities that could have been visible earlier, and to choose molecules based not only on what looks exciting in discovery, but on what has the best overall chance of becoming a drug.

---

View and listen to more episodes on our podcast: The Biologics Brief.

---