Week 1 HW: Automated Biomanufacturing of an Anti-Dengue Monoclonal Antibody

Automated Biomanufacturing of an Anti-Dengue Monoclonal Antibody

1. What would I like to biomanufacture, and why?

I would like to biomanufacture a recombinant human monoclonal antibody (mAb) against dengue virus (DENV). For this assignment, I will assume that a suitable broadly neutralizing antibody has already been discovered and validated. My focus is not antibody discovery, but the engineering problem of turning such a biological molecule into a reproducible manufactured product.

Dengue is a useful application because it remains a major global health problem. The World Health Organization estimates that approximately 100–400 million dengue infections occur each year, and there is currently no specific antiviral treatment for dengue. [1]

This project is also personally interesting to me because my research background is in virology, immunology, and dengue biology. Most of my experience has been on the discovery side: understanding viruses, immune responses, and antibody-mediated immunity. I am therefore interested in the next part of the pipeline: how a biological discovery becomes something that can actually be manufactured at meaningful scale.

The target product would be a purified recombinant human IgG antibody intended as a hypothetical passive immunization or therapeutic candidate against dengue.

For the purposes of this homework, I assume that:

• the antibody sequence has already been identified;
• its biological activity has already been demonstrated;
• a stable production cell line can be generated; and
• the main problem is developing and scaling the manufacturing process.

The final output I would want is purified antibody drug substance of sufficient quality and quantity to support treatment of a defined population.

2. What living system would I use?

I would use Chinese hamster ovary (CHO) cells as the manufacturing chassis.

CHO cells are widely used for recombinant therapeutic protein production, including monoclonal antibodies. Fed-batch CHO culture is a standard industrial approach, and modern CHO processes can achieve antibody titers above 5 g/L. [2,3]

At a high level, the process would work as follows:

Inputs:

• a CHO cell line carrying genes encoding the antibody heavy and light chains;
• chemically defined culture medium;
• nutrient feeds such as glucose and amino acids;
• oxygen and other gases;
• controlled temperature, pH, and dissolved oxygen.

Biological process:

The CHO cells transcribe and translate the antibody genes. The heavy and light chains are assembled into IgG molecules inside the cells, undergo mammalian post-translational processing, and are secreted into the culture medium.

The process would begin with expansion of the producer cells through a seed train, followed by inoculation into a larger fed-batch bioreactor. During production, additional nutrients would be supplied while temperature, pH, dissolved oxygen, cell density, and metabolite concentrations are monitored.

Outputs:

At the end of the culture, the antibody-containing medium would be harvested. Cells and debris would first be removed, followed by purification steps such as Protein A affinity chromatography and polishing operations.

A simplified process is therefore:

CHO producer cells → seed expansion → fed-batch bioreactor → antibody secretion → harvest → purification → purified antibody

The manufacturing chassis does more than simply produce protein. Mammalian cells are also useful because they can perform protein folding, assembly, and post-translational processing required for many therapeutic antibodies.

3. How would I use automation?

I would use automation differently during process development and during large-scale manufacturing.

Automation during process development

One challenge in developing a CHO process is the large number of parameters that could affect antibody production.

For example, I might initially want to test:

• 4 culture media;
• 4 feeding strategies;
• 3 temperature conditions; and
• 4 initial cell densities.

This would already create:

4 × 4 × 3 × 4 = 192 experimental conditions

Testing these conditions manually would require substantial repetitive work and would introduce variation between experiments.

I would therefore use an automated liquid-handling platform, such as an Opentrons system, together with parallel small-scale cell cultures.

Automation could perform tasks such as:

• preparing media and feed combinations;
• dispensing cells and reagents;
• carrying out scheduled feeding;
• collecting samples;
• preparing samples for antibody quantification;
• transferring measurement data into a structured database.

This would allow many conditions to be tested systematically and reproducibly.

The purpose of automation here is not to make the CHO cells grow faster. Instead, it increases the number of conditions that can be explored in parallel and reduces manual variability.

Petzold and Mukhopadhyay describe this broader role for high-throughput automation, self-driving laboratories, cloud laboratories, and integrated data systems in exploring large biomanufacturing design spaces. [4]

Automation during manufacturing

The automation required for manufacturing at scale would be different.

A large bioreactor would use automated process-control systems for:

• temperature;
• pH;
• dissolved oxygen;
• agitation;
• gas flow;
• nutrient feeding;
• sampling; and
• process data recording.

Downstream purification could similarly use automated chromatography systems, pumps, sensors, and fraction collection.

Therefore, I see two distinct roles for automation:

Development automation helps identify a good process.

Manufacturing automation helps execute that process reproducibly at large scale.

4. How much would I need to make, and how long would it take?

4a. How much?

For a meaningful but manageable manufacturing target, I will estimate the amount required for 1,000 treatment courses.

Because there is no established anti-dengue monoclonal antibody dose for this hypothetical product, I need to make an explicit engineering assumption.

I will assume:

• average adult body mass = 70 kg
• antibody dose = 10 mg/kg
• one administration per treatment course

Therefore:

Dose per person

70 kg × 10 mg/kg = 700 mg

700 mg = 0.7 g

For 1,000 people:

1,000 people × 0.7 g/person = 700 g antibody

I therefore need approximately:

0.7 kg of purified antibody

Next, I estimate the required culture volume.

I will assume:

• CHO antibody titer = 5 g/L
• overall downstream recovery = 70%

The 5 g/L titer is consistent with reported multi-gram-per-liter fed-batch CHO production. [2]

The 70% recovery is an engineering assumption representing losses during harvest and purification.

Therefore, purified antibody obtained per liter of culture is:

5 g/L × 0.70 = 3.5 g/L

The required culture volume is:

700 g ÷ 3.5 g/L = 200 L

Therefore, approximately 200 L of productive CHO culture would theoretically be sufficient.

In practice, I would not design a process with zero margin. If I used a 250 L working culture volume:

250 L × 5 g/L = 1,250 g antibody before downstream processing

After 70% recovery:

1,250 g × 0.70 = 875 g purified antibody

This would provide approximately:

875 g ÷ 0.7 g/person = 1,250 treatment courses

A 250 L culture would therefore provide roughly a 25% manufacturing margin above the 1,000-course target.

4b. How long?

I will use a simplified manufacturing timeline.

My assumptions are:

• seed expansion: approximately 10 days
• production fed-batch culture: approximately 14 days
• harvest and downstream purification: approximately 3 days

Total process time:

10 + 14 + 3 = 27 days

Therefore, a rough estimate is that one manufacturing campaign could generate approximately 0.9 kg of purified antibody in about:

27 days, or approximately four weeks

This is the biological and processing time rather than the full commercial release time.

A real therapeutic manufacturing process would require additional activities such as:

• quality-control testing;
• sterility testing;
• product characterization;
• batch review;
• facility turnaround; and
• regulatory release.

These could substantially increase the calendar time before the product could actually be distributed.

An important observation is that automation does not eliminate the approximately two weeks required for the cells to grow and produce antibody. Its main value is reducing manual work, improving reproducibility, controlling the process continuously, and accelerating process development.

5. What do I not know yet?

Three assumptions would have particularly large effects on whether this concept is feasible.

1. What dose would actually be required?

The assumed dose of 10 mg/kg is only an engineering assumption for this exercise.

The actual required dose would depend on:

• antibody potency;
• pharmacokinetics;
• tissue distribution;
• duration of protection;
• timing relative to infection; and
• whether one or multiple administrations are required.

This variable has a direct linear effect on manufacturing demand.

For example, if the required dose were 100 mg/kg rather than 10 mg/kg, the required manufacturing capacity would increase by tenfold.

2. Can the CHO process achieve high titer while maintaining product quality?

I assumed a titer of 5 g/L.

However, the actual titer would depend on the producer clone, culture medium, feeding strategy, process conditions, and stability of antibody expression.

High antibody concentration alone would also not be sufficient. I would need to measure critical quality attributes such as:

• aggregation;
• glycosylation;
• charge variants;
• structural integrity; and
• biological activity.

Therefore, the process-development goal would be to optimize both quantity and quality, not simply maximize titer.

3. Is 70% downstream recovery realistic for this particular molecule?

My calculation assumes that 70% of the antibody present at harvest reaches the final purified product.

Actual recovery would depend on losses during:

• clarification;
• Protein A capture;
• polishing;
• filtration;
• concentration; and
• formulation.

I would therefore need experimental mass-balance measurements across the downstream process.

Overall engineering model

The central manufacturing relationship can be summarized as:

Required product = number of people × dose per person

and:

Required culture volume = required product ÷ (culture titer × downstream recovery)

For my assumed case:

1,000 people × 0.7 g/person = 700 g

and:

700 g ÷ (5 g/L × 0.70) = 200 L

This simple calculation shows how decisions made in biology and medicine propagate directly into manufacturing requirements.

A more potent antibody requiring a smaller dose would reduce manufacturing demand. A higher-producing cell line would reduce bioreactor volume. Better downstream recovery would reduce material losses.

This is what I find most interesting about the problem: discovery, biological engineering, automation, and manufacturing scale cannot ultimately be considered independently.

References

1. World Health Organization. Dengue. WHO Fact Sheet, 2025.

2. Dahodwala H, Lee KH. The fickle CHO: a review of the causes, implications, and potential alleviation of the CHO cell line instability problem. Current Opinion in Biotechnology. 2019;60:128–137. DOI: 10.1016/j.copbio.2019.01.011.

3. Reinhart D, Damjanovic L, Kaisermayer C, Sommeregger W, Gili A, Gasselhuber B, Castan A, Mayrhofer P, Grünwald-Gruber C, Kunert R. Bioprocessing of recombinant CHO-K1, CHO-DG44, and CHO-S: CHO expression hosts favor either mAb production or biomass synthesis. Biotechnology Journal. 2019.

4. Petzold CJ, Mukhopadhyay A. From bench to biofactory: high-throughput technologies and automated workflows to accelerate biomanufacturing. Current Opinion in Biotechnology. 2025;94:103320. DOI: 10.1016/j.copbio.2025.103320.

5. Asin-Garcia E, Fawcett JD, Batianis C, Martins dos Santos VAP. A snapshot of biomanufacturing and the need for enabling research infrastructure. Trends in Biotechnology. 2025;43(5):1000–1014. DOI: 10.1016/j.tibtech.2024.10.014.

AI Use

AI client: OpenAI ChatGPT, GPT-5.6 Sol, accessed September 22, 2026.

ChatGPT was used to help structure my initial idea of manufacturing an anti-DENV monoclonal antibody, suggest reasonable assumptions for a back-of-the-envelope estimate (e.g., CHO production yield and antibody dosing), and check the calculations and wording.

Example prompt:

Help me organize an idea for manufacturing an anti-DENV monoclonal antibody and identify reasonable assumptions for CHO yield, antibody dose, production scale, and manufacturing time.

The project concept was based on my own research background and interests, and I reviewed and edited the final analysis.

Final Project