One Antibody, Three Problems: Visualizing Emicizumab’s Mechanism with AlphaFold 3

A behind-the-scenes look at compressing a bispecific antibody’s mechanism of action into a single infographic.

Infographic on the Key Mechanism of Action of Emicizumab, a Hemophilia A Treatment

The Problem

Hemophilia A is a bleeding disorder with a deceptively simple core issue: Factor VIII is missing.

Without it, the coagulation cascade breaks down at a critical junction — Factor IXa and Factor X can’t form the complex they need to stop bleeding. The result is a system that can’t complete its most basic job.

Emicizumab fixes this with an elegant workaround. It’s a bispecific antibody that physically bridges FIXa and FX, mimicking what Factor VIII used to do. One molecule. Two binding arms. One restored pathway.

The mechanism is clean. Visualizing it without losing that clarity is not.

What I Was Working With

Two key papers:

Mechanism: Factor VIIIa-mimetic cofactor activity of a bispecific antibody to factors IX/IXa and X/Xa, emicizumab (Thrombosis and Haemostasis, 2017)

Clinical efficacy: Emicizumab Prophylaxis in Patients Who Have Hemophilia A without Inhibitors (NEJM, 2018)

All graphics are unofficial portfolio visualizations. I was careful not to overstate findings or imply conclusions beyond what each paper supports.

Design Goal: One Sentence, One Image

Before opening any software, I wrote the core message:

“Emicizumab restores the coagulation pathway by physically bridging FIXa and FX — a function the missing Factor VIII can no longer perform.”

Every element in the layout had to serve that sentence. Anything that didn’t was cut.

What I Cut — and Why

The coagulation cascade is a 12-step chain reaction. Drawing all of it would have buried the point.

I reduced the “problem” side to a single 2D inset in the top-left corner: two icons, one arrow, one broken pathway. That’s it. The reader understands the gap in three seconds and moves on to the solution.

The solution — the 3D antibody bridging structure — takes the center. That’s where the visual weight belongs.

The Hardest Technical Decision

AlphaFold 3 can model the emicizumab antibody. It can also model Factor IXa and Factor X separately.

Getting all three to render together cleanly in ChimeraX was a different problem. The combined binding data didn’t load the way I needed it to. So I made a decision: render what I could accurately, and use the pLDDT confidence map — average score 93.19 — to show that the structural prediction itself is solid.

The confidence map isn’t just a data footnote. It’s the proof that this isn’t a made-up illustration. I put it in the bottom-right corner where a skeptical reader’s eye lands last.

One Style Rule That Fixed Everything

My first pass had all three protein structures rendered in solid, opaque textures. It looked like a pile of shapes. The bridging mechanism — the whole point — disappeared into the clutter.

The fix was simple: emicizumab gets a translucent, glass-like material. The target proteins stay opaque. One rule, applied consistently, and suddenly the antibody reads as the thing connecting the other two rather than competing with them.

The Layout Logic: Z-Pattern

The reader’s eye enters top-left (the problem), sweeps to the center-right (the 3D structure), and exits bottom-right (the confidence data). That’s a Z-pattern, and it maps directly onto the narrative: here’s what’s broken, here’s the fix, here’s why you can trust it.

The grid was designed around that path. Nothing interrupts it.

What I Learned

Using structural prediction data — pLDDT scores, confidence maps — as a design element rather than a footnote changes what the image can claim. It moves a visualization from “this is what I think it looks like” to “this is what the data predicts, with measured confidence.”

That’s a different kind of credibility. Worth building into every project that touches protein structure.

What I’d Refine Next

The three visual elements — 2D inset, 3D rendering, confidence map — carry different visual weights right now. The center rendering dominates more than it should.

Next version: I want to run a visual weight check across all three panels before finalizing layout. Equal emphasis, equal credibility.

I’ll also revisit the full FIXa-FX-emicizumab complex rendering once a cleaner dataset is available. The current version is accurate. It could be more complete.

Want Something Like This for Your Research?

If you’re working in hematology, immunology, or drug mechanism visualization and need a graphical abstract or mechanism infographic for your next manuscript or grant proposal, feel free to reach out via the contact link in my profile.

References: Thrombosis and Haemostasis (2017); NEJM (2018). Structural prediction: AlphaFold 3 Server. Rendering: UCSF ChimeraX. All graphics are unofficial portfolio visualizations. Not peer-reviewed.