Week 1 HW: Final Project Ideas

1. What I would like to biomanufacture and why

The Product

I propose to biomanufacture Cardiovascular-Optimized Cultivated Beef, a cultivated meat product whose nutritional composition is intentionally engineered rather than simply reproducing that of conventional beef.

The goal is to produce beef with a more favorable fatty-acid profile, particularly a higher proportion of omega-3 polyunsaturated fatty acids (PUFAs), while also incorporating β-carotene as an antioxidant and potential provitamin A source.

Why Make It?

Conventional livestock production gives us some ability to influence this composition through animal genetics, breeding, and diet, but these approaches are constrained by the biology of the whole animal. Cultivated meat presents a different possibility. Instead of raising an entire animal and accepting the resulting biochemical composition of its tissues, we can potentially engineer the cells responsible for producing the meat. This creates the possibility of treating meat as a programmable biological product. My motivation is therefore to explore whether metabolic engineering can be used not only to manufacture meat without raising an entire animal, but also to design the nutritional properties of the meat itself.

Who Is the Product For?

The eventual product would be intended for consumers who want to continue eating beef while seeking a product with an intentionally modified nutritional profile. I would envision it as a food designed to offer a potentially more favorable fatty-acid composition than conventional beef.

Why Is This Interesting as a Biomanufacturing Problem?

What makes this idea particularly interesting to me is that it goes beyond simply replacing conventional meat production with cell culture and combines cultivated meat, metabolic engineering, and automated biomanufacturing. The same general approach could eventually be extended beyond fatty acids and β-carotene to engineer other nutritional, sensory, or functional properties of meat. My long-term vision is therefore a manufacturing platform in which meat is not simply grown, but biologically programmed according to the properties we want the final food to have.

2. What living system would I use?

The Manufacturing Chassis

I would use genetically engineered bovine muscle progenitor cells as the living manufacturing system.

These cells would first be expanded in culture and then differentiated into muscle cells to produce cultivated beef.

The overall process can be represented as:

Engineered bovine cells → cell Proliferation → muscle differentiation → cultivated beef

Engineering the Cells

I would introduce two main metabolic modules into the bovine cells.

The first module would contain an omega-3 desaturase such as FAT-1 from Caenorhabditis elegans. FAT-1 can convert several omega-6 fatty acids into their corresponding omega-3 fatty acids.

Conceptually:

Omega-6 PUFA → FAT-1 → corresponding omega-3 PUFA

The purpose of this modification would be to shift the fatty-acid composition of the cells toward a higher omega-3/omega-6 ratio.

FAT-1 alone would not necessarily produce large amounts of EPA or DHA. If specific long-chain omega-3 fatty acids were required, additional elongation and desaturation enzymes could be incorporated into the pathway.

β-Carotene Production

The second module would introduce a β-carotene biosynthetic pathway using enzymes such as CrtB, CrtI, and CrtY.

In simplified form:

Cellular precursors → CrtB → phytoene → CrtI → lycopene → CrtY → β-carotene

This would allow the engineered cells to synthesize β-carotene internally rather than relying entirely on β-carotene supplied through the culture medium.

The β-carotene would potentially provide both nutritional value and antioxidant activity. This is particularly relevant because increasing the amount of unsaturated fatty acids could also increase the susceptibility of the resulting meat to lipid oxidation.

Inputs and Outputs

The major inputs to the system would be engineered bovine cells and cell-culture medium containing the nutrients required for cell growth and metabolism.

The cells would use these nutrients to proliferate, produce biomass, carry out the engineered metabolic reactions, and eventually differentiate into muscle tissue.

The desired outputs would therefore be:

• Bovine muscle biomass
• An increased proportion of omega-3 PUFAs
• An increased omega-3/omega-6 ratio
• Detectable β-carotene
• A final cultivated beef product with the desired nutritional composition

At a high level:

flowchart TD
    A["Bovine Cells + Nutrients"] --> B["Cell Proliferation"]
    B --> C1["FAT-1 Pathway<br/>(Altered fatty-acid composition)"]
    B --> C2["CrtB / CrtI / CrtY Pathway<br/>(β-Carotene production)"]
    C1 --> D["Muscle Differentiation"]
    C2 --> D
    D --> E["Cardiovascular-Optimized Cultivated Beef"]

Key Biological Assumption

A major assumption is that these additional metabolic pathways can be introduced without substantially compromising the ability of the bovine cells to grow and differentiate.

The engineered pathways will require cellular resources and metabolic precursors that would otherwise be used for normal cell functions. Excessive expression could therefore create a metabolic burden, reduce growth rate, or interfere with muscle differentiation.

This would need to be experimentally tested by comparing engineered and unmodified bovine cells for growth rate, viability, differentiation capacity, fatty-acid composition, β-carotene production, and lipid oxidation.

The initial objective would therefore not be to immediately produce large quantities of meat, but to identify an engineered cell line that can maintain healthy growth while producing the desired nutritional changes.

3. How would I use automation?

Automation During Development

I would use automation primarily during the development and optimization of the engineered bovine cell line.

A platform such as an Opentrons liquid-handling robot could automate cell seeding, media addition, reagent transfer, sampling, and preparation of multiple experimental conditions in parallel.

This would allow many combinations of genetic constructs, gene-expression levels, media compositions, and culture conditions to be tested simultaneously.

High-Throughput Screening

The main advantage of automation would be the ability to rapidly screen many engineered cell populations.

For example, different levels of FAT-1 expression could be tested to determine how strongly they alter the omega-3/omega-6 ratio while maintaining acceptable cell growth.

Similarly, different levels of β-carotene pathway expression could be tested to determine whether β-carotene production can be increased without creating excessive metabolic burden.

The resulting cell populations could be screened for:

• Cell growth and viability
• Differentiation efficiency
• Omega-3/omega-6 ratio
• β-carotene production
• Lipid oxidation
• Overall biomass production

Automation would make these experiments more reproducible and reduce the amount of manual laboratory work required.

Automation at Manufacturing Scale

The automation technology used during development would be different from the technology used for large-scale manufacturing.

At manufacturing scale, I would use automated stirred-tank bioreactors equipped with sensors and control systems.

These systems could continuously monitor and control parameters such as Temp, pH, DO etc

Automated sampling could also periodically collect small samples for measurements of fatty-acid composition, β-carotene concentration, cell growth, and product quality.

Ultimately, automation would allow the biological manufacturing process to move from manually optimized laboratory experiments toward a reproducible and scalable production system.

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

(a) How Much?

To define a meaningful manufacturing scale, I would target enough Cardiovascular-Optimized Cultivated Beef to supply the annual beef consumption of 10,000 people.

The USDA Economic Research Service projects U.S. beef availability at approximately 60.0 lb per person per year in 2026. I use this value as a proxy for annual beef consumption. USDA ERS

First, I convert this value to kilograms:

60.0 lb/person/year × 0.453592 kg/lb = 27.22 kg/person/year

For 10,000 people:

27.22 kg/person/year × 10,000 people = 272,200 kg/year

Therefore, my target manufacturing scale is approximately:

272,200 kg/year ≈ 272 tonnes/year

This calculation assumes that one kilogram of cultivated beef would replace one kilogram of conventional beef.

I would initially target a ground-meat-like cultivated beef product rather than a fully structured steak. This is important because current large-scale cultivated-meat models primarily describe the production of animal-cell biomass rather than fully structured whole cuts.

Production Capacity of One Reactor

To estimate the infrastructure required, I use a 2026 process model by Brenner et al. The model describes a 20,000 L stirred-tank reactor operating at a maximum cell density of approximately 5 × 10⁷ cells/mL and producing approximately 2,000 kg of biomass per 30.7-day batch. Brenner et al., 2026

The approximate number of batches that one reactor could complete per year is:

(365 days) / (30.7 days/batch) = 11.89 batches/year

Therefore, the theoretical annual output of one reactor would be:

11.89 batches/year × 2,000 kg/batch
= 23,780 kg/reactor/year

or approximately 23.8 tonnes per reactor per year.

The same model reports a maximum biomass concentration of approximately 105 g/L.

For a 20,000 L reactor:

20,000 L × 105 g/L = 2,100,000 g
= 2,100 kg

This is consistent with the model’s approximately 2,000 kg biomass output per batch.

Number of Reactors Required

The theoretical number of reactors required to produce 272,200 kg/year is:

(272,200 kg/year) / (23,780 kg/reactor/year) = 11.45

Therefore, at least:

12 reactors

would be required under ideal continuous operation.

However, real facilities would require time for cleaning, sterilization, maintenance, reactor turnaround, and potential downtime.

To account for this, I would use approximately 13 parallel 20,000 L production reactors as the practical estimate.

This gives some additional production capacity rather than designing the facility to operate at exactly 100% of its theoretical capacity.

(b) How Long Would It Take?

A single 20,000 L reactor would produce approximately 2,000 kg every 30.7 days.

If only one reactor were available, the number of batches required to produce the full annual target would be:

(272,200 kg) / (2,000 kg/batch) = 136.1 batches

The corresponding production time would be:

136.1 batches × 30.7 days/batch
= 4,178 days

or approximately:

11.4 years

This demonstrates why a single large reactor would not be sufficient for the proposed scale.

Instead, multiple reactors would operate in parallel. With approximately 13 reactors, production could be distributed across the reactors so that different batches are continuously entering, growing, harvesting, and undergoing downstream processing.

Under the simplified assumptions, this would provide enough capacity to produce approximately 272 tonnes of cultivated beef per year, corresponding to the annual beef consumption of 10,000 people.

Important Assumptions and Limitations

Several assumptions are required for this back-of-the-envelope calculation.

1. Reactor performance: I assume that the engineered bovine cells can eventually achieve performance comparable to the cells represented in the published 20,000 L process model.

2. Batch productivity: I use approximately 2,000 kg of biomass per 30.7-day batch as the primary manufacturing benchmark.

3. Downtime: The theoretical calculation assumes continuous operation, while the practical estimate of 13 reactors provides some additional capacity for cleaning, maintenance, and other downtime.

4. Engineered-cell performance: The most important assumption is that introducing the FAT-1 and β-carotene pathways does not substantially reduce cell growth, viability, or differentiation efficiency.

The published process model itself does not represent every step required to manufacture a commercial steak. In particular, differentiation, tissue maturation, and structured tissue formation introduce additional engineering challenges.

This scale also illustrates why automation is important. Producing hundreds of tonnes of engineered cultivated meat would require automated monitoring, feeding, sampling, quality control, and coordination across many parallel bioreactors rather than relying on manual operation.

Sources:

Brenner et al. (2026) — Decoding cultured meat manufacturing: a full process model to identify scale-up bottlenecks.
This is the primary source for the 20,000 L reactor, 105 g/L biomass concentration, ~2,000 kg/batch, and 30.7-day production cycle.
https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2026.1844185/full USDA Economic Research Service (2026) — Projected 60.0 lb/person/year beef availability.
https://ers.usda.gov/data-products/charts-of-note/114196 Risner et al. (2019) — Independent cultivated-meat bioprocess calculation estimating approximately $2.9 \times 10^{11}$ (2.9 × 10¹¹) muscle cells per kg of wet cultivated meat, useful as a sanity check on the enormous cell numbers involved.
https://www.frontiersin.org/journals/sustainable-food-systems/articles/10.3389/fsufs.2019.00044/full

5. What don’t I know yet?

There are several important technical uncertainties that would need to be resolved before this concept could become a practical manufacturing process.

1. Will the genetic modifications interfere with cell growth?

The biggest unknown is how introducing the FAT-1 and β-carotene pathways will affect bovine cell growth and differentiation.

These pathways will consume cellular resources, metabolic precursors, and energy. This could create a metabolic burden and cause the cells to grow more slowly or differentiate less efficiently.

I would compare engineered and unmodified cells for growth rate, viability, and differentiation efficiency.

2. Will the pathways produce enough of the desired molecules?

FAT-1 may shift the omega-6/omega-3 balance, but it may not produce high levels of specific long-chain omega-3 fatty acids such as EPA or DHA on its own.

Similarly, introducing CrtB, CrtI, and CrtY does not guarantee that enough β-carotene will accumulate.

I would therefore measure the omega-3/omega-6 ratio, individual fatty acids, β-carotene concentration, and lipid oxidation to determine whether the engineering produces the intended nutritional profile.

3. Will the system work at manufacturing scale?

Performance in a small culture vessel may not translate directly to a large bioreactor.

Large-scale cultivation introduces challenges such as oxygen transfer, nutrient gradients, mixing, pH control, and waste accumulation.

The engineered cells would therefore need to be tested progressively from small cultures to larger bioreactors while monitoring their growth and metabolic performance.

4. Will the product remain stable?

Increasing unsaturated fatty acids could make the final product more susceptible to lipid oxidation.

Although β-carotene is intended to provide antioxidant protection, I do not yet know whether the amount produced would be sufficient.

This would need to be tested during processing, storage, and cooking.

The Biggest Questions

The three most important questions are:

1. Can the engineered pathways improve the nutritional profile without reducing cell growth or differentiation?
2. Can the engineered cells maintain their performance during large-scale cultivation?
3. Can the final high-PUFA product remain stable against lipid oxidation?

The first major experiment would therefore compare engineered and unmodified bovine cells under identical conditions.

If the engineered cells maintain similar growth and differentiation while producing more omega-3 PUFAs and β-carotene, the concept would have a strong foundation for further scale-up.

Final Project

Week 2 HW

Part 0: Set up in Kernel

My work is located in the HTBAA 2026 workspace, in the repository Mohammed Irfan B under the folder HW_2

Part 1: Define your expression target

What molecule are we expressing?

My HW1 product beef muscle tissue enriched in omega-3 fatty acids and pre-loaded with β-carotene is not itself a single molecule we can express. Instead, we are engineering the bovine muscle progenitor cells to produce the enzymes that synthesize these two components. This gives us five target proteins, organized into two functional modules:

Module 1 — Omega-3 enrichment (1 enzyme):

• FAT-1, an ω-3 fatty acid desaturase that converts ω-6 fatty acids already present in the cell (e.g., linoleic acid) into ω-3 fatty acids. Native mammalian cells cannot perform this conversion, so this enzyme is the sole driver of the omega-3 phenotype.

Module 2 — β-carotene biosynthesis (4 enzymes, one pathway):

• CrtE (GGPP synthase), CrtB (phytoene synthase), CrtI (phytoene desaturase), and CrtY (lycopene β-cyclase), acting sequentially to convert endogenous farnesyl pyrophosphate (FPP) into β-carotene. All four are required — bovine cells lack the entire carotenoid pathway, and CrtE specifically is included to boost the geranylgeranyl pyrophosphate (GGPP) supply beyond what the cell’s native prenylation pathway (via GGPS1) provides.

What is the sequence?

What host or chassis?

Bovine (Bos taurus) muscle progenitor cells (satellite-cell derived), consistent with our HW1 cultivated meat platform. Kernel’s codon optimization tools do not include a Bos taurus table so I used GenSmart (GenScript), which does support Bos taurus.

Where does the product need to go?

• FAT-1 remains at the ER membrane.
• CrtE, CrtB, CrtI, and CrtY are cytoplasmic.

Part 2: Design Your Genetic Parts

My host is eukaryotic (bovine muscle progenitor cells), so I designed against the eukaryotic column of the parts table. My construct has two independent expression units — one for FAT-1, one for the four-gene β-carotene cassette — both carried on a single PiggyBac vector

Unit 1: FAT-1 (omega-3 desaturase)

Unit 2 : β-carotene biosynthesis

Vector backbone (shared, not part of either expression unit)

Part 3: Assemble Your Construct

I assembled my construct, by joining the FAT-1 unit and the β-carotene unit together inside a single PiggyBac vector, flanked by the 5′ and 3′ terminal repeats. Both units run in the same direction, with the FAT-1 cassette first, followed by the four-gene crtE-2A-crtB-2A-crtI-2A-crtY cassette, all sitting inside a backbone that carries the origin of replication and ampicillin resistance needed to grow the plasmid in E. coli.

Before finalizing, I checked that the reading frame stays intact all the way through the β-carotene chain — one start codon at the beginning of crtE, one stop codon at the end of crtY, and no stray stops hiding at any of the three 2A junctions in between, since a stop codon there would have cut off translation early and left the downstream enzymes unmade. The FAT-1 cassette was simpler to check, since it’s a single self-contained gene with its own start and stop.

Linear View

Circular View

Part- 4 : Optimize and quality-check

Part 4: Optimize and Quality-Check

Codon optimization

For codon optimization, I used GenScript’s GenSmart tool rather than Kernel’s built-in optimizer, because GenSmart offers Bos taurus directly as a host option, while Kernel did not have a bovine codon table available. I input the protein sequences for all five genes (fat-1, crtE, crtB, crtI, crtY), taken from their respective UniProt entries, rather than the native nucleotide sequences from C. elegans and P. ananatis. This was a deliberate choice: since my goal was bovine-optimized DNA, starting from the protein and letting GenSmart reverse-translate it directly against the Bos taurus codon table is more direct than optimizing the native bacterial/nematode DNA, which would carry over codon biases from the wrong organism as a starting point.

One consequence of this approach is that I don’t have a true “before” DNA sequence to compare against for GC content or length — there was no original bovine-context DNA sequence to begin with, only the source organism’s native gene, which isn’t a meaningful baseline for codon usage in a mammalian host. So rather than reporting a before/after GC-content shift, I’m reporting the output GC content of each optimized sequence directly:

Fat-1: 57.46%

crtE: 64.34%

crtB: 64.08%

crtI: 61.52%

crtY: 63.79%

All five stayed within or near the typical mammalian coding-sequence range (roughly 55-65% GC), which is what I’d expect from bovine-optimized output. Base pair length did not change relative to each protein’s amino acid count (length = 3 × number of residues, +3 for a stop codon where applicable), confirming the optimization only reassigned codons and did not insert or delete any coding content.

BLAST :

Fat-1

crtE

crtB

crtI

crtY

Biosecurity:

I screened my full construct (both expression units plus PiggyBac ITRs and PuroR cassette) using the SecureDNA Biosecurity Sequence Scanner. The scan returned “Flagged sequences detected”, with one hit annotated as Human cytomegalovirus at approximately positions 3,470-3,515 of the scanned sequence.

What the flag corresponds to: On inspection, the flagged region is the CMV immediate-early enhancer/promoter, which I had placed upstream of the β-carotene cassette (Unit 2). It is a non-coding regulatory element, not a viral gene encoding a virulence factor, toxin, or replication machinery. None of my coding sequences (fat-1, crtE, crtB, crtI, crtY, PuroR) were flagged. They encode a nematode desaturase, bacterial carotenoid enzymes, and a puromycin acetyltransferase.

Risk assessment: I think this is a low-risk flag. The CMV promoter is very widely used in mammalian expression vectors, and my construct contains no CMV structural or replication genes, so it can’t produce infectious virus.

Part 5: From design to build

How would I build it?

My construct is roughly 13 kb (about 10-11 kb between the ITRs plus the bacterial backbone). For the prototype I would order it as 4-6 synthetic DNA fragments (1-3 kb each) and assemble them into a PiggyBac backbone by Gibson or Golden Gate assembly. Ordering the full 13 kb as one clone is possible but costlier, slower, and more likely to fail on repeats and high-GC regions. A chip oligo pool would be overkill for a single construct. I would only use it later to screen variant libraries, such as promoter strengths to tune FAT-1 expression against metabolic burden.

What does synthesis constrain?

Three parts of my design may be hard to synthesize:

• High GC content: the crt genes are 61-64% GC, which can cause secondary structure and poor yield.
• Repeats: the three 2A peptides encode similar motifs, and reusing a promoter or polyA across both units would create larger repeats. I used different parts for each unit to avoid this.
• PiggyBac ITRs: inverted repeats that can form hairpins and are awkward to synthesize and clone.

I would run a synthesis-feasibility check and re-optimize flagged windows with synonymous codons. Fragment-based assembly also confines any difficult region to one small fragment. Because synthesis has error rates, I would sequence-verify every assembled clone and pick a correct one rather than trust the pool.

Prototype vs. scale

The two differ, and that works in my favor. The HW1 target of ~272 tonnes/year is a cell scale, not a DNA scale. Because the construct integrates into the bovine genome via PiggyBac, I build and verify it once, at microgram scale.

• Prototype: co-transfect the plasmid with PiggyBac transposase, select with puromycin, and screen clones on the Opentrons platform.
• Scale: the best clone becomes a master cell bank that seeds every 20,000 L batch, with no new DNA synthesis per batch. The scale-up risks are about the cells: stability of the integrated construct, copy number, and re-sequencing the bank to confirm no mutations accumulate.

Part 6: What don’t I know yet?

• 1. Will the four-enzyme β-carotene cassette work in bovine cells? I don’t know whether the bacterial enzymes (CrtE, CrtB, CrtI, CrtY) will all be made in the right amounts and be active in a mammalian host. The 2A peptides are never 100% efficient, so some enzymes may stay fused, and the ones downstream of a 2A site are often expressed at lower levels. Uncleaved fusions or unbalanced ratios could stall the pathway at phytoene or lycopene instead of reaching β-carotene. Bovine cells may also not supply enough FPP or GGPP precursor. I would test this by western blot, and by measuring pathway intermediates and β-carotene by HPLC. If needed, I would reorder the genes or tune expression empirically.
• 2. Will FAT-1 shift the fatty-acid profile enough to matter? FAT-1 is a nematode enzyme that has to fold and function in the bovine ER membrane. Even if it does, the shift depends on how much omega-6 substrate the cells have, so the omega-3/omega-6 ratio may change less than I hope. I would measure the fatty-acid profile by GC-MS in engineered versus unmodified cells, and compare growth and differentiation to check for metabolic burden, as in HW1.
• 3. Will expression stay stable over time? PiggyBac integrates at random sites with variable copy number, so expression will differ between clones. Silencing over many passages, or during differentiation into muscle, could switch the transgenes off. Codon optimization alone can’t answer this. I would screen multiple clones, re-measure expression at late passages and after differentiation, and sequence the master cell bank before scale-up.
• Biggest question overall: can I get β-carotene and omega-3 production at meaningful levels without slowing growth or blocking differentiation? The first experiment is the same as in HW1: compare engineered and unmodified cells under identical conditions.

Week 3 HW

A. Compare Cell-Free Reaction Compositions

1. Why do potassium glutamate and magnesium glutamate concentrations differ between compositions? Which is more sensitive?

Different roles → different optimal concentrations

• Potassium glutamate is mainly an ionic-environment reagent. K⁺ is the dominant intracellular cation and glutamate the dominant anion in E. coli, so K-glutamate is used to recreate cytoplasmic ionic strength/osmolarity and support general enzyme and ribosome activity. Its response curve is broad: Jewett & Swartz found maximum protein production at 175 mM potassium glutamate, but the optimum sits on a fairly flat plateau, and yields were “nearly insensitive to ammonium glutamate concentrations between 0 and 100 mM.” In other words, you can be off by tens of mM and lose little activity.

• Magnesium glutamate is a functional cofactor. Mg²⁺ is required for ribosome structure and translational fidelity, it forms the true substrate complex Mg-NTP, and it is needed by ATP-regenerating enzymes (e.g., pyruvate kinase). Crucially, Mg²⁺ is chelated by phosphorylated compounds — PEP, NTPs/NMPs, and the inorganic phosphate that accumulates during the reaction. So the “right” Mg-glutamate concentration depends on how much phosphorylated material a given formulation loads (a 33 mM PEP system has a very different Mg²⁺ demand than an NMP/ribose system).

Why values differ across formulations

1. The energy system sets the Mg²⁺/phosphate balance (PEP-heavy systems need more Mg²⁺ headroom).
2. The extract itself carries over salts (~3.3 mM Mg and ~14.4 mM K from S30 buffer in the Swartz-lab systems), so the optimum shifts with every lysate batch.
3. Each formulation is empirically re-optimized — Olsen et al. explicitly re-optimized magnesium and phosphate for each new lysate batch and lysate concentration.

Which is more sensitive?

Magnesium. Mg²⁺ has a narrow, bell-shaped optimum — too little stalls ribosomes and ATP regeneration; too much promotes misincorporation, nonspecific enzyme inhibition, and precipitation with phosphate/nucleotides.

This is why Mg-glutamate is the parameter labs re-titrate first when changing extract batches, DNA load, or energy system, whereas K-glutamate tolerates a much wider window (PANOx-SP uses 175 mM, RFopt ends up near ~362 mM — both work).

2. PEP + NTP vs. Ribose + NMP

In one sentence: PEP+NTP pays upfront for a fast, self-contained, oxygen-independent burst of synthesis, while ribose+NMP outsources ATP and NTP production to the extract’s own metabolism — far cheaper and longer-lived, but only if glycolysis, salvage pathways, and oxidative phosphorylation are active.

3. Literature comparison: PANOx-SP (PEP-based) vs. RFopt (ribose-based)

PANOx-SP — Jewett, M.C. & Swartz, J.R. (2004) Biotechnol. Bioeng. 86:19–26.

RFopt — Olsen, M.L. et al. (2026) Nature Communications (doi:10.1038/s41467-026-69605-8; screened >1,200 formulations via definitive screening designs).

Contrasts

The design philosophies are 22 years apart.

• PANOx-SP was hand-crafted to mimic the cytoplasm (swap acetate → glutamate, PEG → polyamines, single-variable titrations).
• RFopt was found by design-of-experiments screening — 524 formulations tested across a 2²¹ (~2.1 million) combination design space — which stripped out everything non-essential (no NTPs, tRNA, CoA, folinic acid, polyamines) and deliberately added phosphate and buffer.
• The practical trade-off mirrors Q2: PANOx-SP is oxygen-insensitive and tolerant of lysate variability; RFopt is ~35× cheaper and higher-yielding but demands good oxygen transfer and per-batch Mg/phosphate tuning.

B. Design One Reaction Composition for Each of the Three Proteins

1. sfGFP Reaction Composition

JSON

{
  "sample_id": "mohammed_irfan_bazeer_ahamed_2026-10-06_3wu7bu",
  "sample_type": "experimental",
  "reagent_list": {
    "amino_acid_mix_17": 1100,
    "amp": 125,
    "base_buffer": 2000,
    "cmp": 75,
    "cysteine": 400,
    "glucose": 125,
    "gmp": 100,
    "hepes_koh": 300,
    "kpo_dibasic_mix": 225,
    "kpo_monobasic_mix": 225,
    "lysate": 5000,
    "magnesium_glutamate": 250,
    "nicotinamide": 600,
    "nuclease_free_water": 6000,
    "ribose": 1200,
    "spermidine": 100,
    "template_sfgfp": 2000,
    "tyrosine": 100,
    "ump": 75
  },
  "metadata": {
    "ai_model": "",
    "protein_target": "sfGFP",
    "acknowledgements": "",
    "comments": ""
  }
}

Rationale

• Energy: ribose 40 mM + glucose 6.94 mM + NMPs (AMP 0.63, CMP 0.38, GMP 0.5, UMP 0.38 mM): This composition is shared by every ≥3.0 g/L model; avoids PEP/NTP costs entirely.
• Substrates: AA 3.75 mM + Cys 4.0 mM (cysteine enrichment is in all top-4 models) + Tyr 1.25 mM.
• pH control: HEPES 45 mM + phosphate 5.6/5.6 mM dual buffer — #77’s exact recipe.
• Spermidine: 1.25 mM kept as cheap insurance: not strictly required (#77 won without it), but it stabilizes ribosomes/mRNA and sits in 3 of 5 top models at ~1.2 mM — minimal cost, plausible upside, no shown downside for a one-shot bet.
• Everything else omitted: catalase, spermidine, pyruvate, adenosine, guanine, DTT/GSSG/GSH, PEG, etc.: the data shows they add cost, not yield, for sfGFP.

2. PETase

{
  "sample_id": "mohammed_irfan_bazeer_ahamed_2026-10-06_3wu7bu",
  "sample_type": "experimental",
  "reagent_list": {
    "amino_acid_mix_17": 800,
    "amp": 200,
    "base_buffer": 2000,
    "cmp": 150,
    "cysteine": 100,
    "glucose": 150,
    "gmp": 150,
    "hepes_koh": 600,
    "kpo_monobasic_mix": 400,
    "lysate": 5000,
    "magnesium_glutamate": 300,
    "nicotinamide": 800,
    "nuclease_free_water": 4900,
    "oxidized_glutathione": 225,
    "reduced_glutathione": 100,
    "ribose": 1500,
    "spermidine": 75,
    "template_sfgfp": 2000,
    "tyrosine": 400,
    "ump": 150
  },
  "metadata": {
    "ai_model": "",
    "protein_target": "PETase",
    "acknowledgements": "",
    "comments": ""
  }
}

3. Reteplase

{
  "sample_id": "mohammed_irfan_bazeer_ahamed_2026-10-06_3wu7bu",
  "sample_type": "experimental",
  "reagent_list": {
    "amino_acid_mix_17": 800,
    "amp": 200,
    "base_buffer": 2000,
    "cmp": 150,
    "dsbc_ecoli": 200,
    "glucose": 150,
    "gmp": 150,
    "hepes_koh": 600,
    "kpo_monobasic_mix": 400,
    "lysate": 5000,
    "magnesium_glutamate": 300,
    "nicotinamide": 800,
    "nuclease_free_water": 4450,
    "oxidized_glutathione": 475,
    "reduced_glutathione": 200,
    "ribose": 1500,
    "spermidine": 75,
    "template_sfgfp": 2000,
    "tyrosine": 400,
    "ump": 150
  },
  "metadata": {
    "ai_model": "",
    "protein_target": "Reteplase",
    "acknowledgements": "",
    "comments": ""
  }
}

Rationale for PETase and Reteplase

• Ribose + NMPs replaces PEP + NTPs + NAD + CoA + folinic acid: Because ribose is cheaper.
• Acidic phosphate (mono:di, 10 mM): mirrors the sfGFP-Ribose reference; counters pH drop from ribose/glucose catabolism.
• Total Mg ~10 mM, HEPES 60 mM, AA 3 mM: squarely inside the range of the best-performing GPT-5 conditions.
• Reteplase: gets a strongly oxidizing redox pair (4:1 GSSG:GSH) and no added cysteine.
• PETase: gets a milder redox pulse suited to its fewer disulfides.