GlycoWorks
What I built: GlycoTwin, an open mechanistic digital twin connecting a bioreactor knob to the antibody's Fc N-glycan — a quality attribute governing potency, half-life, and immunogenicity.
Core trick
A nine-state CHO fed-batch ODE feeds an ER-to-Golgi enzyme model, so process parameters propagate to Fc galactosylation using only literature-derived constants.
Tags
Description
What I built: GlycoTwin, an open mechanistic digital twin connecting a bioreactor knob to the antibody's Fc N-glycan — a quality attribute governing potency, half-life, and immunogenicity. A 9-state CHO fed-batch model feeds an ER→Golgi enzyme train, so any process parameter propagates to galactosylation, afucosylation, sialylation, and high-mannose. Every parameter comes from public literature; nothing is fit to proprietary data. I tested one question: does Fc-galactosylation degrade at scale because CO₂ acidifies the Golgi — and if not, what controls it?
What I found: The assumption is wrong. As CO₂ stripping falls from bench to 2000 L and pCO₂ climbs ~60→74 mmHg, galactosylation barely moves (~52→53%), because the CO₂ swing shifts Golgi-lumen pH only 6.43→6.38 — straddling the galactosyltransferase optimum (~6.4) on the flat top of its activity curve. Oxygen has zero effect. The real lever is the clone: overexpressing B4GALT1 recovers ~35 points; galactose feed and Mn²⁺ ~10 each; CO₂ stripping ~0. The model also reproduces the sub-5% sensitivity ceiling from a 2026 meta-analysis it never saw during calibration.
Why it matters: Engineer the clone and media, not the CO₂ — productivity, not glycan quality, is the scale-up risk. Being mechanistic, the twin gives a bench-testable reason, not a black-box fit, and spans the CMC workflow: batch monitor, ICH-Q8 design space, Monte-Carlo capability, live control, and a governed agent loop. Fully open, reproducible from one script.
README excerpt
# GlycoTwin: a mechanistic digital twin that overturns how we control antibody glycosylation at manufacturing scale GlycoTwin connects the **bioreactor knob** an engineer can touch to the **N-glycan quality attribute** set inside the Golgi. A 9-state fed-batch CHO bioreactor ODE feeds an ER→Golgi enzyme-train, so moving a process parameter propagates all the way to the secreted IgG1 Fc glycan (galactosylation, afucosylation, sialylation, high-mannose). Glycosylation is driven by its two real control levers — **Golgi enzymes** and **nucleotide-sugar donors** — and both are explicit in the model. **▶ Live app (no install):** https://chadakarn23cs.github.io/Hackathon-Claude/ — opens the full tool in the browser; every tab, including Agent Optimize, runs client-side. ## The problem Glycosylation is a **critical quality attribute** of therapeutic antibodies: the Fc N-glycan sets effector potency (ADCC/CDC), half-life, and immunogenicity, so regulators treat it as a release-defining CQA. Yet it is set inside the Golgi by a branched enzyme network fed by nucleotide-sugar donors — many sequential reactions, several competing enzymes — which makes it **hard to predict and hard to control** from the process parameters an engineer can actually touch. Two questions recur through development and tech-transfer, and both are usually answered by expensive empirical runs: *which lever actually moves a given glycan attribute?* and *will that attribute hold when the process scales from a f