Individual Final Project

Here are my preliminar three individual final project ideas!

An interactive 3D bio-art sculpture where human touch meets living bioluminiscence bloom.

A river-sensing automated robot system that helps both advance research and act as a tourist spot for the community to reunite and spend time together.

Chlorella vulgaris in silico optimization and automation for the optimal accumulation of polyhydroxybutyrate (PHB).

After some thought and consideration, I decided to go with my third idea for my final individual project, here you can find a more detailed view of it:

Modeling and Automating Polyhydroxybutyrate Bioplastic Production in Microalgae: An In Silico and Automated Approach

Author: Fabrizio Flores Huamán Course: How to Grow (Almost) Anything (HTGAA) 2026 Username: 2026a-fabrizio-flores-huaman Date: May 2026

SECTION 1: ABSTRACT

The global accumulation of petroleum-derived plastics has caused severe and often irreversible ecological damage to marine ecosystems, including the biodiverse coastal waters of Peru. Polyhydroxybutyrate (PHB), a fully biodegradable and biocompatible bacterial polyester, represents a scientifically compelling alternative to fossil fuel-derived plastics, yet its industrial scaling remains hindered by low production yields and the high cost of manually optimizing metabolic conditions. Microalgae such as Chlorella vulgaris are promising biological chassis for PHB production because they fix carbon dioxide and accumulate carbon-rich polymers under nutrient stress, but the optimization of these conditions continues to rely on slow, one-variable-at-a-time experimentation that is incompatible with industrial timelines.

This project addresses that gap by developing a high-throughput, in silico pipeline that integrates computational metabolic modeling with automated robotic liquid handling protocols. Using Flux Balance Analysis (FBA) implemented through the COBRApy Python library on the iJO1366 genome-scale metabolic model of Escherichia coli, this work identifies optimal combinations of nitrogen limitation, phosphorus limitation, and carbon source variation that maximize carbon flux toward PHB biosynthetic precursors, particularly acetyl-coenzyme A (acetyl-CoA). The E. coli iJO1366 model is used for Aim 1 because E. coli BL21 (DE3) is the practical validation chassis; the iCZ843 genome-scale model of C. vulgaris is designated for Aim 2, where it informs the algal construct design and transformation conditions. The computationally predicted conditions are then translated into modular Opentrons OT-2 Python protocols, creating a reproducible and stoppable screening pipeline.

In parallel, a synthetic PHB biosynthetic operon encoding phaA (beta-ketothiolase), phaB (acetoacetyl-CoA reductase), and phaC (polyhydroxyalkanoate synthase) is designed for expression in Escherichia coli BL21 (DE3) as a practical validation chassis, with the construct to be synthesized by Twist Bioscience and deployed at Ginkgo Bioworks. The long-term vision of this project is to empower Peruvian coastal communities to participate in decentralized, community-scale biomanufacturing of sustainable materials through AI-guided, closed-loop photobioreactor systems that connect local biodiversity with the global circular bioeconomy.

SECTION 2: PROJECT AIMS

Aim 1: Experimental Aim

The first aim of this project is to identify the optimal metabolic conditions for Polyhydroxybutyrate (PHB) accumulation in Escherichia coli BL21 (DE3) by utilizing COBRApy for Flux Balance Analysis (FBA) on the iJO1366 genome-scale metabolic model of E. coli, and to translate the predicted nutrient stress conditions into automated Python-based protocols for the Opentrons OT-2, including the design and ordering of a synthetic PHB biosynthetic operon from Twist Bioscience for validation at Ginkgo Bioworks. The iJO1366 model is the correct tool for Aim 1 because E. coli BL21 (DE3) is the experimental chassis; the iCZ843 genome-scale model of Chlorella vulgaris is reserved for Aim 2, where it informs the design of the algal construct and transformation conditions.

Resources and methods for Aim 1:

• COBRApy (Python library) with the iJO1366 E. coli genome-scale metabolic model (JSON format, fully compatible with modern COBRApy and Python 3.12)
• iCZ843 C. vulgaris genome-scale model referenced for Aim 2 design (requires MATLAB or a legacy Python environment due to SBML format constraints)
• Combinatorial FBA simulation across nitrogen, phosphorus, and carbon source gradients (glucose, acetate, glycerol)
• Opentrons OT-2 Python API (v2.15) for automated media preparation and Nile Red fluorescence screening
• Twist Bioscience whole plasmid synthesis of the phaA-phaB-phaC-sfGFP construct (shipped to Ginkgo Bioworks)
• PHERAstar FSX and Spark Plate Reader for Nile Red and sfGFP fluorescence detection
• 96-well Eppendorf deep-well plates for media preparation; 384-well Greiner black clear-bottom plates for assay readout

Aim 2: Development Aim

The next step following the computational and E. coli validation would be to design a Chlorella vulgaris-specific genetic construct, codon-optimized for algal nuclear expression and placed under the control of the NIT1 nitrate-inducible promoter, and to produce a complete electroporation-based transformation and Nile Red screening protocol ready for execution at a partner facility such as Ginkgo Bioworks or a collaborating university laboratory. This aim also includes running FBA on the iCZ843 genome-scale metabolic model of C. vulgaris to identify algal-specific nutrient stress conditions, a step that requires MATLAB or a legacy Python 2 environment due to the 2016-era SBML format of the published model and is therefore scoped as a future computational deliverable beyond the course timeline. While the physical execution of stable algal transformation requires specialized containment equipment and multi-week clonal selection timelines that exceed the scope of this course, the complete construct design, codon optimization rationale, and step-by-step transformation protocol constitute meaningful scientific deliverables that advance the project toward a genuine in vivo test in C. vulgaris.

Aim 3: Visionary Aim

The long-term vision is to establish a framework for decentralized, community-scale bioplastic production by developing AI-guided, closed-loop modular photobioreactors that integrate real-time environmental sensing with the metabolic models developed in Aims 1 and 2. This system would allow Peruvian coastal communities to culture locally sourced microalgae strains under computationally optimized conditions, transforming the extraordinary regional biodiversity of coastal lagoons, river deltas, and Andean water sources into a renewable feedstock for sustainable, compostable materials. By combining open-source metabolic modeling, low-cost automation, and participatory science education and knowledge-sharing, this aim envisions a future where bioplastic manufacturing is not confined to industrial facilities but is instead embedded within the communities most affected by plastic pollution, giving people both the tools and the understanding to build a circular bioeconomy from the ground up.

“The long-term vision is to establish a decentralized framework for bioplastic production through AI-guided, closed-loop modular photobioreactors, enabling coastal communities to transform local microalgae biodiversity into sustainable, high-value compostable materials that replace petroleum-based plastics.”

Here’s my Final Idea Aims Slide:

And here are the slides I used in the final presentation day!

SECTION 3: BACKGROUND

Literature Context

Paper 1: Flux Balance Analysis and constraint-based metabolic modeling

Orth, Thiele, and Palsson (2010) provided a foundational overview of Flux Balance Analysis in their landmark Nature Biotechnology publication, establishing the mathematical and conceptual framework that underpins the computational core of this project. FBA models cellular metabolism as a system of linear equations in which stoichiometric constraints define the boundaries of possible metabolic states, and an objective function such as biomass production or metabolite yield is optimized using linear programming. The authors demonstrated that FBA can accurately predict growth rates, gene essentiality, and the effects of environmental perturbations across multiple organisms without requiring detailed kinetic parameters. Crucially for this project, nutrient limitation can be represented as a constraint on exchange reactions, making FBA directly applicable to modeling nitrogen and phosphorus stress in microalgae and to predicting how carbon flux is redirected under suboptimal growth conditions. This approach transforms the traditionally empirical process of media optimization into a computational design problem, dramatically reducing the number of physical experiments required before identifying high-yield conditions.

Paper 2: The phaABC operon and PHB biosynthesis

Peoples and Sinskey (1989) characterized the biochemical pathway for PHB biosynthesis in Cupriavidus necator (then known as Alcaligenes eutrophus) H16, identifying and sequencing the three-enzyme cascade encoded by phaA (beta-ketothiolase), phaB (acetoacetyl-CoA reductase), and phaC (polyhydroxyalkanoate synthase) that converts two molecules of acetyl-CoA into PHB polymer granules. This foundational study established the genetic basis for heterologous PHB production and demonstrated that the pathway is functionally transferable to other organisms, a property that has since been exploited in dozens of bacterial hosts and, more recently, in green microalgae and plants. The characterization of acetyl-CoA as the central metabolic node for PHB production directly informs the choice of FBA objective in this project, since maximizing acetyl-CoA flux serves as a computationally tractable proxy for maximizing bioplastic precursor availability. Furthermore, the modularity of the three-gene operon makes it an ideal candidate for whole plasmid synthesis via Twist Bioscience, since all three coding sequences can be assembled into a single synthetic construct without requiring post-synthesis cloning.

Paper 3: Chlorella vulgaris as a multi-application biotechnology chassis

Al-Hammadi and Güngörmüşler (2024) published a comprehensive review in Biotechnology and Bioengineering documenting the breadth of C. vulgaris biotechnology applications, covering wastewater treatment, biodiesel, biohydrogen, biocement, biopolymers, and food additives, and concluding that despite its documented potential for polyhydroxyalkanoate (PHA) production from algal biomass, the field still lacks critical strain-specific data that would make industrial PHB production from C. vulgaris economically competitive. The authors note that C. vulgaris is particularly attractive as a production chassis because of its ease of cultivation, rapid growth, and capacity to use carbon dioxide and wastewater nutrients simultaneously, reducing both feedstock costs and environmental impact. The review explicitly identifies a gap between the biological potential of C. vulgaris for biopolymer synthesis and the practical optimization infrastructure needed to bridge laboratory results to industrial scale. This directly motivates the computational and automated pipeline approach taken in this project, since the gap the authors describe is precisely the one that FBA-guided combinatorial screening is designed to address.

Knowledge Gap

Despite the well-established FBA methodology, the clearly characterized PHB biosynthetic pathway, and growing recognition of C. vulgaris as a promising bioplastic chassis, these three domains have rarely been integrated into a single, closed-loop design pipeline. Most optimization studies in algal PHB production continue to rely on one-variable-at-a-time manual experimentation, generating data slowly and failing to capture the multi-dimensional interaction effects between nitrogen, phosphorus, and carbon source availability. Peruvian research groups have independently demonstrated that C. vulgaris is highly responsive to nitrogen and phosphate concentration changes: Oscanoa et al. (2021), working at the Instituto del Mar del Perú, showed that C. vulgaris efficiently removes nitrates and phosphates from domestic wastewater, demonstrating the organism’s natural sensitivity to these nutrient gradients and validating nitrogen and phosphorus as the most impactful variables for metabolic manipulation. Automated robotic screening of metabolic model predictions remains largely confined to bacterial and yeast systems, with very few examples connecting genome-scale algal models to physical high-throughput screening pipelines. This project addresses that gap by connecting the iCZ843 genome-scale model of C. vulgaris to a computationally guided automated OT-2 screening pipeline, establishing a Design-Build-Test-Learn (DBTL) cycle that is both predictive and robotically executable.

Innovation

This project is novel in its integration of three components that have not previously been combined in a single pipeline for microalgal bioplastic research: a genome-scale metabolic model (iCZ843), a modular automated combinatorial nutrient stress screening protocol (OT-2), and a synthetic PHB operon designed for both bacterial validation and future algal expression. Most published work in this area either performs FBA on C. vulgaris without connecting the predictions to physical experiments, or conducts wet-lab PHB screening without the guidance of a metabolic model. By treating the OT-2 protocol as a direct translation layer between computational predictions and physical experiments, this project establishes a reusable pipeline architecture that can be adapted to other bioproduction targets and chassis organisms beyond PHB and microalgae. Unlike previous computational approaches that apply a single generalized metabolic model to predict PHB production, this project uses the iJO1366 genome-scale model of E. coli for Aim 1 FBA to ensure predictions are matched to the actual experimental chassis, and designates the iCZ843 genome-scale model of C. vulgaris specifically for Aim 2 algal construct design, creating a clear organism-to-model correspondence that improves scientific coherence across the project aims. Additionally, the explicit design of a second construct codon-optimized for C. vulgaris under the NIT1 nitrate-inducible promoter represents a forward-looking contribution to algal synthetic biology, where inducible expression systems remain significantly underexplored relative to their bacterial counterparts.

Significance

Plastic pollution is one of the most pressing environmental crises of the 21st century, with an estimated 11 million metric tons of plastic entering the ocean annually, causing harm to marine food webs, coastal ecosystems, and the communities that depend on them, including those along the biodiverse Peruvian coast. Polyhydroxybutyrate (PHB) offers a route to replace petroleum-derived single-use plastics with a material that is fully biodegradable and biocompatible under natural conditions, degrading within months rather than persisting for centuries. Microalgae represent an ideal production organism for PHB because they grow using only sunlight, carbon dioxide, and minimal inorganic nutrients, giving them a dramatically lower carbon footprint than fermentation-based bacterial systems that require sugar feedstocks. Critically, Peru harbors native Chlorella strains with documented biotechnology potential: Mariano Astocondor et al. (2017) characterized Chlorella peruviana, a native strain isolated from the Mellisera salt lagoon in Chilca, Peru, demonstrating robust biomass and chlorophyll productivity across a range of salinities and confirming that locally sourced Peruvian Chlorella strains are viable candidates for controlled cultivation and bioproduction. Complementing this, Condori et al. (2024) demonstrated that Peruvian Chlorella strains can be cultivated using nutrients valorized from local fruit residues, producing growth and lipid yields comparable to synthetic media and illustrating a circular economy model in which local organic waste streams feed microalgae bioproduction, directly relevant to the decentralized manufacturing vision of Aim 3. The integration of FBA-guided optimization with robotic automation directly addresses the industrial bottleneck of manual media optimization, which currently makes algal PHB production economically uncompetitive compared to petrochemical alternatives. By demonstrating that computationally predicted metabolic conditions can be automatically screened at scale, this project contributes a generalizable methodology that could accelerate the path from laboratory research to industrial relevance. This is particularly meaningful for research environments in countries like Peru that possess extraordinary microalgae biodiversity but currently lack access to the high-throughput experimentation infrastructure that makes systematic optimization feasible. Ultimately, reducing the cost and time of metabolic optimization is not only a scientific problem but a social one, and making these tools more accessible is a prerequisite for truly sustainable and equitable biomanufacturing rooted in local biodiversity.

Bioethical Considerations

Paragraph 1: Ethical considerations

This project involves the design and intended use of genetically modified organisms (GMOs), specifically Escherichia coli strains expressing a heterologous PHB biosynthetic pathway, with a longer-term vision of engineering Chlorella vulgaris, a microalga present in natural freshwater and marine environments. The introduction of synthetic genetic constructs into organisms that have natural environmental relatives raises important questions about biosafety, containment, and the potential for unintended horizontal gene transfer or competitive ecological disruption, even when experiments are conducted in well-controlled laboratory settings. The design of organisms intended for eventual scale-up in semi-open systems such as outdoor photobioreactors requires additional evaluation of containment strategies and kill-switch designs that go beyond standard laboratory biosafety level 1 (BSL-1) practices. The project also intersects with questions of biodiversity sovereignty, since any future use of Peruvian microalgae strains for commercial bioplastic production must respect the principles of the Nagoya Protocol on Access and Benefit-Sharing, ensuring that local communities and the Peruvian state retain informed consent and equitable participation in any commercial or intellectual property outcomes arising from the use of their biological resources.

Paragraph 2: Responsible implementation and risk mitigation

To mitigate these risks, all experimental work described in Aim 1 is conducted under standard BSL-1 conditions appropriate for non-pathogenic E. coli strains, with all constructs including a selectable antibiotic resistance marker that requires continuous supplementation for maintenance, significantly reducing the probability of survival outside controlled conditions. The C. vulgaris construct designed for Aim 2 is intended for nuclear rather than chloroplast-targeted integration, which limits expression levels and reduces the likelihood of creating competitive growth advantages in natural algal populations. All synthetic DNA sequences submitted to Twist Bioscience will be screened through the SecureDNA biosecurity framework to ensure compliance with dual-use research standards before synthesis is initiated. For the visionary Aim 3 framework involving photobioreactors in coastal communities, the project design explicitly includes a community consultation and participatory education component, ensuring that local knowledge holders are active participants in the governance and design of any deployment rather than passive recipients of an externally imposed technology. Collaboration with biodiversity documentation initiatives such as those supported by Basecamp Research prior to strain collection and use further ensures that any commercialization pathway is grounded in scientific rigor and community rights from the earliest stages.

SECTION 4: EXPERIMENTAL DESIGN

The experimental design follows a Design-Build-Test-Learn (DBTL) cycle. The Design phase uses FBA modeling to predict optimal nutrient conditions. The Build phase involves synthetic construct design and ordering from Twist Bioscience. The Test phase uses automated OT-2 protocols and Ginkgo Bioworks equipment for Nile Red fluorescence screening. The Learn phase closes the loop by feeding experimental data back into the metabolic model to refine predictions.

SECTION 4.1: Step-by-Step Workflow

Check out this Colab Notebook for the genome scale metabolic model.

Step 1: Install COBRApy and load the E. coli iJO1366 genome-scale metabolic model

• Method: Install COBRApy and compile the multi-modal XML metabolic model structures representing the iCZ843 genome-scale architecture of Chlorella vulgaris. A native Python ElementTree text parsing layout is implemented to assemble valid COBRApy objects, ensuring full operational performance inside modern Python 3.12 environments without requiring deprecated dependency stacks.
• Automation: Python scripting environment (Google Colab; all code below is Colab-ready)
• Expected result: All three distinct trophic model profiles (Heterotrophy, Mixotrophy, Photoautotrophy) parse successfully into memory, displaying their explicit metabolic boundary dimensions and resolving true positive baseline growth rates.
• Timeline: Day 1

# ── CELL 1: Install COBRApy ────────────────────────────────────────────────────
!pip install -q cobra
print("COBRApy installed.")

# ── CELL 2: Upload the three iCZ843 XML files ─────────────────────────────────

from google.colab import files
import os

print("Select all three XML files when the picker opens.")
uploaded = files.upload()

rename_map = {
    'Heterotrophy':    'iCZ843_heterotrophy.xml',
    'Mixotrophy':      'iCZ843_mixotrophy.xml',
    'Photoautotrophy': 'iCZ843_photoautotrophy.xml',
}
for original in list(uploaded.keys()):
    for keyword, new_name in rename_map.items():
        if keyword.lower() in original.lower():
            if original != new_name:
                os.rename(original, new_name)
            print(f"Saved: {new_name} ({os.path.getsize(new_name)/1024:.0f} KB)")

# ── CELL 3: Custom parser — builds COBRApy models without SBML reader ──────────

import xml.etree.ElementTree as ET
from cobra import Model, Reaction, Metabolite
import warnings
warnings.filterwarnings('ignore')

def parse_iCZ843(xml_path, model_name):
    """Parse a Zuniga et al. 2016 iCZ843 SBML Level 2 file into a COBRApy model."""
    tree = ET.parse(xml_path)
    root = tree.getroot()
    cobra_model = Model(model_name)
    metabolites  = {}
    objective_rxn = None

    for model_elem in root:
        # Pass 1: extract all metabolites (species)
        for child in model_elem:
            if 'listOfSpecies' in child.tag:
                for sp in child:
                    sp_id = sp.get('id')
                    met = Metabolite(
                        id=sp_id,
                        name=sp.get('name', sp_id),
                        compartment=sp.get('compartment', 'c')
                    )
                    metabolites[sp_id] = met
                cobra_model.add_metabolites(list(metabolites.values()))

        # Pass 2: extract all reactions with stoichiometry and bounds
        for child in model_elem:
            if 'listOfReactions' in child.tag:
                for rxn_elem in child:
                    rxn_id   = rxn_elem.get('id')
                    rxn_name = rxn_elem.get('name', rxn_id)
                    lb, ub, obj_coeff = -1000.0, 1000.0, 0.0
                    stoich = {}

                    for sub in rxn_elem:
                        if 'listOfReactants' in sub.tag:
                            for sr in sub:
                                sid = sr.get('species')
                                if sid in metabolites:
                                    stoich[metabolites[sid]] = -float(sr.get('stoichiometry', 1))
                        elif 'listOfProducts' in sub.tag:
                            for sr in sub:
                                sid = sr.get('species')
                                if sid in metabolites:
                                    stoich[metabolites[sid]] = float(sr.get('stoichiometry', 1))
                        elif 'kineticLaw' in sub.tag:
                            for lop in sub:
                                for param in lop:
                                    pid = param.get('id')
                                    val = float(param.get('value', 0))
                                    if   pid == 'LOWER_BOUND':          lb = val
                                    elif pid == 'UPPER_BOUND':          ub = val
                                    elif pid == 'OBJECTIVE_COEFFICIENT': obj_coeff = val

                    rxn = Reaction(rxn_id, name=rxn_name, lower_bound=lb, upper_bound=ub)
                    rxn.add_metabolites(stoich)
                    cobra_model.add_reactions([rxn])
                    if obj_coeff == 1.0:
                        objective_rxn = rxn_id

    if objective_rxn:
        cobra_model.objective = objective_rxn
    return cobra_model

import pandas as pd
import numpy as np
import itertools

# Load all three models
model_hetero = parse_iCZ843('iCZ843_heterotrophy.xml',    'iCZ843_heterotrophy')
model_mixo   = parse_iCZ843('iCZ843_mixotrophy.xml',      'iCZ843_mixotrophy')
model_photo  = parse_iCZ843('iCZ843_photoautotrophy.xml', 'iCZ843_photoautotrophy')

for name, m in [('Heterotrophy',    model_hetero),
                ('Mixotrophy',      model_mixo),
                ('Photoautotrophy', model_photo)]:
    sol = m.optimize()
    print(f"{name}: {len(m.reactions)} rxns | "
          f"{len(m.metabolites)} mets | "
          f"biomass = {sol.objective_value:.4f} mmol/gDW/hr | {sol.status}")

# Key reaction IDs in iCZ843 (C. vulgaris model naming convention)
N_RXN   = 'R_EX_no3_LPAREN_e_RPAREN_'    # Nitrate exchange (primary N source in algae)
P_RXN   = 'R_EX_pi_LPAREN_e_RPAREN_'     # Phosphate exchange
GLC_RXN = 'R_EX_glc_DASH_A_LPAREN_e_RPAREN_'  # Glucose exchange
AC_RXN  = 'R_EX_ac_LPAREN_e_RPAREN_'     # Acetate exchange
BIO_H   = 'R_Biomass_Cvu_hetero_DASH_'   # Biomass objective (heterotrophic)
ACCOA   = 'M_accoa_c'                    # Cytoplasmic acetyl-CoA (PHB precursor)
print("\nModel routing: glucose/acetate -> model_hetero | CO2/light -> model_photo")

Step 2: Explore baseline metabolic fluxes

• Method: Run FBA with default exchange reaction constraints to establish a baseline biomass flux and identify the acetyl-CoA reaction node and relevant nutrient exchange reactions.
• Automation: COBRApy optimize() function
• Expected result: Baseline biomass flux value; identification of exchange reaction identifiers for nitrate (EX_no3_e), phosphate (EX_pi_e), glucose (EX_glc__D_e), and acetate (EX_ac_e).
• Timeline: Day 1 to 2

# ── CELL 4: Explore baseline metabolic fluxes (heterotrophic C. vulgaris) ──────

sol = model_hetero.optimize()
print(f"Baseline biomass flux: {sol.objective_value:.4f} mmol/gDW/hr | {sol.status}")

key_rxns = [N_RXN, P_RXN, GLC_RXN, AC_RXN]
print("\nKey exchange reactions:")
for rxn_id in key_rxns:
    try:
        rxn = model_hetero.reactions.get_by_id(rxn_id)
        print(f"  {rxn_id}: [{rxn.lower_bound:.1f}, {rxn.upper_bound:.1f}]")
    except KeyError:
        print(f"  {rxn_id}: NOT FOUND")

# Confirm acetyl-CoA metabolite is present
try:
    accoa_met = model_hetero.metabolites.get_by_id(ACCOA)
    print(f"\nAcetyl-CoA metabolite: {accoa_met.id} | compartment: {accoa_met.compartment}")
except KeyError:
    print(f"WARNING: {ACCOA} not found — search manually:")
    candidates = [m.id for m in model_hetero.metabolites if 'accoa' in m.id.lower()]
    print(candidates)

Step 3: Add a PHB demand reaction as a proxy for bioplastic potential

• Method: Add a synthetic boundary reaction that consumes acetyl-CoA to represent PHB biosynthetic flux; set this as the new FBA objective to directly optimize for bioplastic precursor availability.
• Automation: COBRApy add_boundary() and Reaction() classes
• Expected result: A working optimization model whose objective value represents acetyl-CoA flux directed toward PHB biosynthesis.
• Timeline: Day 2

# ── CELL 5: Add PHB demand reaction to C. vulgaris heterotrophic model ─────────
from cobra import Reaction

# Work on a copy so the original model stays clean
model_phb = model_hetero.copy()

# Cytoplasmic acetyl-CoA is the PHB precursor proxy
accoa = model_phb.metabolites.get_by_id(ACCOA)
print(f"Using acetyl-CoA: {accoa.id} | compartment: {accoa.compartment}")

# Add PHB demand reaction
phb_demand = Reaction('PHB_demand')
phb_demand.name = 'PHB precursor demand (acetyl-CoA sink)'
phb_demand.subsystem = 'PHB biosynthesis'
phb_demand.add_metabolites({accoa: -1.0})
phb_demand.lower_bound = 0
phb_demand.upper_bound = 1000
model_phb.add_reactions([phb_demand])

# Confirm baseline PHB proxy flux without constraints
model_phb.objective = 'PHB_demand'
sol = model_phb.optimize()
print(f"\nBaseline PHB proxy flux: {sol.objective_value:.5f} mmol/gDW/hr | {sol.status}")

Step 4: Simulate nitrogen limitation gradient (adaptive two-stage FBA)

• Method: First measure the empirical nitrate uptake flux the C. vulgaris cell actually uses at baseline, then build a fine-grained gradient relative to that real biological value using numpy linspace. For each nitrogen level, a two-stage FBA runs: first find the condition-specific maximum biomass, then lock a 1% biomass floor and maximize PHB precursor flux. This approach captures biologically realistic stress levels rather than arbitrary percentages.
• Automation: COBRApy context managers; numpy linspace; pandas for results.
• Expected result: PHB precursor flux values that vary with nitrogen availability, showing reduced flux under severe nitrogen starvation and confirming that the two-stage optimization captures meaningful metabolic shifts.
• Timeline: Day 2 to 3

# ── REVISED CELL 6: Adaptive Step Gradient for Nitrate Starvation ────────────
# 1. Detect empirical baseline uptake
model_phb.objective = BIO_H
baseline_sol = model_phb.optimize()
real_n_max = baseline_sol.fluxes[N_RXN] # Captures exact metabolic inflection point

print(f"Empirical Nitrate Baseline Uptake: {real_n_max:.5f} mmol/gDW/hr\n")

# 2. Build adaptive dynamic range constraint array
# Standardize high-volume arbitrary bounds to log historical baseline steps
custom_bounds = [-10.0, -5.0, -1.0]

# Generate millimetric high-resolution steps from 100% down to 0% of true uptake
fine_steps = np.linspace(1.0, 0.0, num=8)
for step in fine_steps:
    custom_bounds.append(real_n_max * step)

# Sort from high-volume capacity to starvation threshold (negative flux dynamics)
custom_bounds = sorted(list(set(custom_bounds)))

# 3. Process constraint matrix through two-stage Linear Programming loop
adaptive_results = []

for bound in custom_bounds:
    with model_phb:
        # Enforce step-specific boundary constraint
        model_phb.reactions.get_by_id(N_RXN).lower_bound = bound

        # Stage 1: Compute true capacity ceiling for condition biomass
        model_phb.objective = BIO_H
        s1 = model_phb.optimize()
        cond_max = s1.objective_value if s1.status == 'optimal' else 0

        if cond_max < 1e-9:
            adaptive_results.append({
                'Applied_N_Bound': round(bound, 5),
                'Biomass_Flux': 0,
                'PHB_Flux_Proxy': 0,
                'Status': 'No Growth'
            })
            continue

        # Stage 2: Impose 1% operational floor constraint, maximize precursor sink
        model_phb.reactions.get_by_id(BIO_H).lower_bound = 0.01 * cond_max
        model_phb.objective = 'PHB_demand'
        s2 = model_phb.optimize()

        adaptive_results.append({
            'Applied_N_Bound': round(bound, 5),
            'Biomass_Flux': round(cond_max, 5),
            'PHB_Flux_Proxy': round(s2.objective_value, 5) if s2.status == 'optimal' else 0,
            'Status': s2.status
        })

df_adaptive = pd.DataFrame(adaptive_results)
print("Final Dataset - Nutrient Stress Adaptive Mapping Matrix:")
print(df_adaptive.to_string(index=False))

Step 5: Combinatorial adaptive FBA grid (nitrogen x phosphorus x carbon source)

• Method: Run the full combinatorial grid using empirically-calibrated nutrient baselines per carbon source. For each carbon source (glucose and acetate), the actual N and P uptake fluxes the cell uses during optimal growth are measured first; stress fractions then scale relative to those real biological values. This is more accurate than scaling from model lower bound constants, which may not reflect actual in silico uptake. The two-stage optimization applies throughout.
• Automation: COBRApy nested context managers; itertools.product; pandas.
• Expected result: A ranked table showing that carbon source type is the primary FBA determinant of acetyl-CoA flux in C. vulgaris, with glucose significantly outperforming acetate. This finding motivates the NIT1 promoter design in Aim 2, since nutrient limitation alone is insufficient to drive PHB accumulation without active transcriptional control.
• Timeline: Day 3 to 4

# ── CELL 7: Combinatorial adaptive FBA grid (iCZ843 C. vulgaris) ──────────────
# Measures empirical N and P baseline uptake per carbon source, then applies
# stress fractions relative to those real biological values.
 
with model_phb:
    model_phb.reactions.get_by_id(GLC_RXN).lower_bound = default_glc
    model_phb.reactions.get_by_id(AC_RXN).lower_bound  = 0
    model_phb.objective = BIO_H
    sol_glc    = model_phb.optimize()
    real_n_glc = sol_glc.fluxes[N_RXN] if sol_glc.status == 'optimal' else default_n_lb
    real_p_glc = sol_glc.fluxes[P_RXN] if sol_glc.status == 'optimal' else default_p_lb
 
with model_phb:
    model_phb.reactions.get_by_id(AC_RXN).lower_bound  = default_ac
    model_phb.reactions.get_by_id(GLC_RXN).lower_bound = 0
    model_phb.objective = BIO_H
    sol_ac    = model_phb.optimize()
    real_n_ac = sol_ac.fluxes[N_RXN] if sol_ac.status == 'optimal' else default_n_lb
    real_p_ac = sol_ac.fluxes[P_RXN] if sol_ac.status == 'optimal' else default_p_lb
 
print(f"Glucose baseline: N = {real_n_glc:.5f} | P = {real_p_glc:.5f}")
print(f"Acetate baseline: N = {real_n_ac:.5f}  | P = {real_p_ac:.5f}")
 
n_fracs = [0.1, 0.25, 0.5, 0.75, 1.0]
p_fracs = [0.1, 0.25, 0.5, 0.75, 1.0]
 
carbon_modes = {
    'glucose': (GLC_RXN, default_glc, AC_RXN,  0, real_n_glc, real_p_glc),
    'acetate': (AC_RXN,  default_ac,  GLC_RXN,  0, real_n_ac,  real_p_ac),
}
 
results_3d = []
for n_f, p_f, (c_name, (open_rxn, open_val, close_rxn, close_val, target_n, target_p)) in itertools.product(
        n_fracs, p_fracs, carbon_modes.items()):
    with model_phb:
        if target_n == 0 or target_p == 0:
            continue
        model_phb.reactions.get_by_id(N_RXN).lower_bound    = target_n * n_f
        model_phb.reactions.get_by_id(P_RXN).lower_bound    = target_p * p_f
        model_phb.reactions.get_by_id(open_rxn).lower_bound  = open_val
        model_phb.reactions.get_by_id(close_rxn).lower_bound = close_val
        model_phb.objective = BIO_H
        s1 = model_phb.optimize()
        cond_max = s1.objective_value if s1.status == 'optimal' else 0
        if cond_max < 1e-9:
            results_3d.append({'N': n_f, 'P': p_f, 'Carbon': c_name,
                                'PHB_flux': 0, 'status': 'no_growth'})
            continue
        model_phb.reactions.get_by_id(BIO_H).lower_bound = 0.01 * cond_max
        model_phb.objective = 'PHB_demand'
        s2 = model_phb.optimize()
        results_3d.append({
            'N': n_f, 'P': p_f, 'Carbon': c_name,
            'PHB_flux': round(s2.objective_value, 5) if s2.status == 'optimal' else 0,
            'status': s2.status
        })
 
df_3d = pd.DataFrame(results_3d).sort_values('PHB_flux', ascending=False)
print("\nTop 15 conditions by predicted PHB precursor flux:")
print(df_3d.head(15).to_string(index=False))
print(f"\nGlucose max PHB = {df_3d[df_3d.Carbon=='glucose']['PHB_flux'].max():.5f}")
print(f"Acetate max PHB = {df_3d[df_3d.Carbon=='acetate']['PHB_flux'].max():.5f}")

Step 6: Export top conditions and generate FBA landscape visualizations

• Method: Export the top conditions as a CSV for OT-2 protocol parameterization, then generate heatmaps and line plots showing the PHB precursor flux landscape across the nitrogen-phosphorus-carbon combinatorial space.
• Automation: pandas to_csv; matplotlib and seaborn for visualization; files.download for Colab export.
• Expected result: A CSV file ready for OT-2 parameterization and two publication-quality figures showing the FBA optimization landscape.
• Timeline: Day 4

# ── CELL 8: Export top conditions ─────────────────────────────────────────────
from google.colab import files
 
top_conditions = df_3d.head(12).reset_index(drop=True)
top_conditions.to_csv('fba_top_conditions.csv', index=False)
print(top_conditions[['N', 'P', 'Carbon', 'PHB_flux']].to_string())
files.download('fba_top_conditions.csv')
 
# ── CELL 9: Heatmap visualization ─────────────────────────────────────────────
import matplotlib.pyplot as plt
import seaborn as sns
 
matrix_glc = df_3d[df_3d.Carbon=='glucose'].pivot(index='N', columns='P', values='PHB_flux')
matrix_ac  = df_3d[df_3d.Carbon=='acetate'].pivot(index='N', columns='P', values='PHB_flux')
 
fig, axes = plt.subplots(1, 2, figsize=(16, 6), sharey=True)
sns.heatmap(matrix_glc, annot=True, fmt=".5f", cmap="YlGnBu", ax=axes[0],
            cbar_kws={'label': 'PHB flux proxy (mmol/gDW/hr)'})
axes[0].set_title('PHB Flux Landscape: Glucose', fontweight='bold')
axes[0].set_xlabel('Phosphate fraction (P)')
axes[0].set_ylabel('Nitrate fraction (N)')
axes[0].invert_yaxis()
sns.heatmap(matrix_ac, annot=True, fmt=".5f", cmap="YlOrRd", ax=axes[1],
            cbar_kws={'label': 'PHB flux proxy (mmol/gDW/hr)'})
axes[1].set_title('PHB Flux Landscape: Acetate', fontweight='bold')
axes[1].set_xlabel('Phosphate fraction (P)')
plt.suptitle('Combinatorial FBA Landscape — iCZ843 C. vulgaris (Zuniga et al. 2016)',
             fontsize=13, fontweight='bold', y=1.02)
plt.tight_layout()
plt.savefig('fba_heatmap.png', dpi=300, bbox_inches='tight')
plt.show()
files.download('fba_heatmap.png')

Step 7: Design the PHB biosynthetic construct in Benchling

• Method: Design a synthetic operon containing phaA, phaB, and phaC genes from Cupriavidus necator H16, codon-optimized for E. coli BL21 (DE3) expression; add an sfGFP transcriptional reporter under the BBa_J23119 constitutive promoter as a proxy for construct activity; use the T7 promoter for operon expression and the pET-28a(+) backbone with kanamycin resistance and ColE1 high-copy origin for selection and propagation. The insert (4,725 bp) was designed in Benchling and submitted to Twist Bioscience as a clonal gene synthesis order for insertion into pET-28a(+), producing a complete 10,017 bp expression plasmid.
• Automation: Benchling sequence editor for annotation; IDT Codon Optimization Tool or Benchling codon optimizer for sequence design.
• Expected result: A fully annotated plasmid sequence in GenBank format confirmed by Twist Bioscience tools, ready for synthesis and delivery to Ginkgo Bioworks.
• Timeline: Day 4 to 5 The E. coli expression insert (phaABC_Ecoli_Construct_2, 4,725 bp) GenBank file is shown below. This insert was submitted to Twist Bioscience for clonal synthesis into the pET-28a(+) backbone, producing the complete 10,017 bp expression plasmid (phaABC_Ecoli_Final_Con):

LOCUS       phaABC_Ecoli_Construct_ 4725 bp ds-DNA     circular     12-MAY-2026
DEFINITION  PHB biosynthetic operon insert for pET-28a(+) cloning. Contains
            T7-driven phaA-phaB-phaC operon (E. coli codon-optimized from
            C. necator H16) with individual RBS elements; sfGFP reporter under
            BBa_J23119 constitutive promoter; BBa_B0015 double terminator.
            Submitted to Twist Bioscience for clonal gene synthesis.
FEATURES             Location/Qualifiers
     promoter        1..19
                     /label="T7 promoter"
                     /note="promoter for bacteriophage T7 RNA polymerase"
     RBS             20..33
                     /label="RBS"
                     /note="Shine-Dalgarno ribosome binding site for phaA"
     CDS             34..1215
                     /label="phaA_Ecoli_optimized2"
                     /gene="phaA"
                     /product="beta-ketothiolase"
                     /note="Converts 2x acetyl-CoA to acetoacetyl-CoA;
                             first committed step in PHB biosynthesis;
                             E. coli codon-optimized from C. necator H16;
                             1182 bp coding sequence"
                     /codon_start=1
     RBS             1224..1237
                     /label="RBS2"
                     /note="Ribosome binding site for phaB"
     CDS             1238..1978
                     /label="phaB_Ecoli_optimized2"
                     /gene="phaB"
                     /product="acetoacetyl-CoA reductase (NADPH-dependent)"
                     /note="Reduces acetoacetyl-CoA to (R)-3-hydroxybutyryl-CoA;
                             E. coli codon-optimized from C. necator H16;
                             741 bp coding sequence"
                     /codon_start=1
     RBS             1987..1999
                     /label="RBS3"
                     /note="Ribosome binding site for phaC"
     CDS             2000..3766
                     /label="phaC_Ecoli_optimized2"
                     /gene="phaC"
                     /product="polyhydroxyalkanoate synthase class I"
                     /note="Polymerizes (R)-3-hydroxybutyryl-CoA into PHB
                             granules; rate-limiting enzyme; E. coli
                             codon-optimized from C. necator H16;
                             1767 bp coding sequence"
                     /codon_start=1
     terminator      3767..3814
                     /label="T7 terminator"
                     /note="transcription terminator for bacteriophage T7
                             RNA polymerase; terminates phaABC operon"
     promoter        3815..3849
                     /label="BBa_J23119"
                     /note="Strong constitutive Anderson promoter;
                             drives sfGFP reporter expression independently
                             of T7 polymerase; confirms construct presence"
     CDS             3850..4596
                     /label="sfGFP2"
                     /gene="sfGFP"
                     /product="superfolder green fluorescent protein"
                     /note="Transcriptional activity reporter; Ex 485 nm /
                             Em 510 nm; 747 bp coding sequence"
                     /codon_start=1
     terminator      4597..4725
                     /label="BBa_B0015"
                     /note="Bidirectional double terminator; insulates sfGFP
                             from downstream vector sequences"
ORIGIN
        1 taatacgact cactataggA AGGAGATATA CATATGACTG ATGTTGTCAT AGTTTCAGCT
       61 GCCCGCACGG CGGTCGGCAA ATTTGGGGGG TCACTGGCTA AAATTCCGGC ACCTGAACTG
      121 GGTGCCGTTG TAATCAAAGC GGCGCTGGAG CGCGCGGGCG TCAAACCGGA ACAGGTGAGC
      181 GAAGTTATTA TGGGCCAGGT GTTGACCGCA GGTAGCGGCC AGAACCCCGC ACGTCAGGCC
      241 GCGATTAAAG CTGGTTTGCC CGCGATGGTG CCGGCAATGA CCATTAACAA AGTGTGTGGT
      301 TCTGGCCTGA AGGCCGTGAT GCTGGCGGCC AACGCGATCA TGGCGGGTGA TGCGGAAATC
      361 GTTGTGGCGG GAGGTCAAGA AAATATGAGC GCGGCGCCTC ATGTTCTGCC GGGAAGCCGG
      421 GATGGTTTCC GAATGGGCGA TGCGAAATTG GTCGATACCA TGATTGTCGA TGGTCTGTGG
      481 GACGTGTACA ATCAGTACCA CATGGGTATT ACAGCGGAGA ACGTGGCCAA AGAATATGGG
      541 ATTACACGCG AAGCGCAGGA TGAGTTTGCT GTAGGCTCTC AAAACAAAGC AGAAGCCGCG
      601 CAGAAGGCCG GCAAATTTGA CGAAGAAATC GTCCCGGTTC TGATCCCGCA GAGGAAAGGA
      661 GATCCGGTAG CCTTCAAAAC CGACGAGTTC GTGCGCCAGG GCGCTACGCT GGACAGCATG
      721 AGTGGCCTCA AACCAGCCTT TGACAAGGCA GGCACGGTAA CCGCGGCCAA CGCAAGTGGC
      781 TTAAATGACG GCGCCGCCGC GGTTGTGGTG ATGTCGGCGG CCAAAGCTAA GGAACTTGGG
      841 TTAACCCCGT TAGCCACGAT TAAGAGCTAT GCAAATGCTG GTGTTGATCC AAAAGTTATG
      901 GGTATGGGCC CGGTGCCAGC CTCCAAGCGC GCCCTGTCAC GCGCCGAATG GACCCCTCAA
      961 GATCTAGATC TCATGGAAAT CAATGAGGCA TTTGCCGCAC AGGCGCTTGC GGTGCATCAG
     1021 CAAATGGGCT GGGACACTTC CAAGGTCAAT GTGAACGGGG GCGCGATTGC AATTGGACAC
     1081 CCGATTGGTG CGTCGGGCTG CCGTATTCTT GTGACTCTGC TCCATGAAAT GAAACGTCGT
     1141 GATGCAAAAA AAGGCTTAGC CTCGCTGTGC ATCGGCGGCG GAATGGGAGT AGCGCTGGCA
     1201 GTAGAGCGTA AATAAGCTCT AGAGAAAGAG GGGACAAATG ACTCAGAGGA TTGCGTATGT
     1261 GACCGGAGGC ATGGGTGGTA TAGGAACGGC CATTTGTCAA CGGTTGGCAA AAGATGGCTT
     1321 TCGTGTGGTT GCGGGTTGCG GCCCAAACTC GCCGCGCCGC GAAAAGTGGC TGGAACAACA
     1381 GAAGGCGCTG GGTTTTGATT TCATTGCCAG TGAAGGCAAT GTGGCTGACT GGGACTCCAC
     1441 CAAAACTGCA TTCGATAAGG TCAAGAGCGA GGTCGGAGAG GTTGATGTTC TGATCAACAA
     1501 CGCCGGTATT ACACGTGATG TGGTGTTTCG CAAAATGACA CGCGCCGATT GGGATGCGGT
     1561 GATCGACACC AATCTGACCA GCTTATTTAA TGTAACGAAA CAGGTGATTG ACGGCATGGC
     1621 CGACCGTGGC TGGGGCCGCA TTGTAAACAT CAGCTCTGTT AACGGGCAGA AAGGCCAATT
     1681 TGGTCAGACG AATTACTCCA CCGCAAAAGC CGGCCTGCAT GGCTTTACCA TGGCATTAGC
     1741 GCAGGAAGTT GCGACTAAAG GTGTGACCGT TAACACGGTA TCTCCGGGGT ATATTGCCAC
     1801 CGACATGGTC AAAGCGATCC GTCAGGATGT GCTTGATAAA ATTGTAGCGA CGATCCCGGT
     1861 CAAACGACTT GGCCTACCTG AAGAAATCGC TAGCATTTGC GCATGGTTGT CATCAGAGGA
     1921 AAGTGGTTTC AGCACCGGCG CGGATTTCTC GCTCAATGGC GGGCTGCACA TGGGATAAAT
     1981 GCATGCAGGA GGTACAAAAA TGGCCACCGG AAAAGGAGCC GCAGCATCCA CGCAGGAGGG
     2041 TAAGAGCCAG CCGTTTAAGG TAACCCCGGG TCCATTCGAC CCGGCCACGT GGCTGGAATG
     2101 GTCCCGCCAG TGGCAAGGTA CCGAAGGTAA CGGTCATGCA GCGGCTTCTG GCATCCCGGG
     2161 CCTAGATGCA CTGGCAGGGG TGAAGATAGC GCCCGCCCAG CTGGGGGATA TACAGCAGCG
     2221 CTACATGAAA GATTTTTCAG CCTTGTGGCA GGCTATGGCG GAAGGCAAAG CGGAAGCCAC
     2281 GGGTCCGCTA CACGATAGGC GTTTTGCGGG CGATGCTTGG CGCACAAACC TCCCGTACCG
     2341 CTTTGCGGCG GCGTTTTACC TGCTGAACGC ACGTGCGCTG ACCGAACTGG CGGATGCCGT
     2401 TGAAGCTGAC GCCAAGACGA GACAGCGGAT TCGCTTCGCG ATTAGCCAAT GGGTCGATGC
     2461 GATGAGCCCG GCGAATTTCT TGGCGACAAA CCCTGAAGCA CAGCGTTTAC TTATCGAGTC
     2521 CGGTGGAGAA AGCCTTCGGG CAGGGGTTCG CAATATGATG GAAGACCTGA CCCGAGGGAA
     2581 GATTAGTCAA ACTGATGAAT CGGCTTTTGA GGTGGGTCGA AACGTAGCGG TAACCGAGGG
     2641 CGCCGTGGTT TTTGAAAATG AATATTTTCA GCTGCTTCAA TATAAACCTC TGACGGATAA
     2701 AGTGCACGCC CGTCCGTTGC TGATGGTCCC GCCATGCATT AATAAATATT ACATTTTAGA
     2761 TCTGCAACCG GAAAGCAGCT TGGTCCGCCA TGTCGTAGAA CAGGGCCACA CGGTTTTCCT
     2821 AGTTTCATGG CGTAACCCGG ATGCTAGCAT GGCTGGCTCG ACCTGGGATG ACTATATCGA
     2881 ACATGCCGCC ATTCGCGCCA TTGAAGTAGC ACGCGACATT TCGGGCCAAG ATAAAATCAA
     2941 CGTGCTGGGT TTCTGCGTCG GTGGGACCAT CGTCTCGACT GCGTTAGCCG TTCTCGCGGC
     3001 GCGTGGCGAG CACCCTGCCG CGTCAGTGAC CCTGCTGACT ACCCTCCTGG ATTTTGCAGA
     3061 TACTGGAATC CTGGACGTGT TCGTAGACGA GGGCCATGTT CAGTTGCGTG AGGCGACCCT
     3121 GGGCGGCGGC GCAGGCGCCC CTTGCGCACT TCTCCGCGGC TTAGAACTGG CGAACACCTT
     3181 CAGCTTTCTG CGCCCAAATG ACCTGGTATG GAATTATGTG GTGGATAACT ACTTAAAAGG
     3241 CAACACGCCC GTGCCCTTCG ACCTTCTGTT CTGGAATGGA GATGCGACAA ACTTACCGGG
     3301 CCCCTGGTAC TGTTGGTATC TCCGTCACAC CTACTTGCAG AATGAACTGA AAGTGCCTGG
     3361 CAAACTTACA GTTTGTGGGG TGCCGGTCGA CCTCGCAAGT ATTGATGTGC CGACTTATAT
     3421 CTATGGTTCA CGTGAGGATC ATATTGTTCC GTGGACCGCA GCGTACGCCA GCACCGCCTT
     3481 ACTTGCGAAT AAGTTACGCT TTGTGCTGGG TGCTTCTGGC CATATTGCGG GCGTGATTAA
     3541 TCCACCGGCG AAAAATAAAC GTAGTCATTG GACGAACGAT GCCTTGCCGG AAAGTCCTCA
     3601 GCAGTGGCTG GCAGGCGCAA TCGAACATCA CGGCTCTTGG TGGCCAGACT GGACCGCCTG
     3661 GCTGGCGGGA CAAGCCGGCG CGAAACGTGC CGCTCCGGCA AATTATGGTA ACGCGCGTTA
     3721 CCGCGCCATT GAGCCCGCCC CGGGTCGCTA TGTTAAAGCC AAAGCGctag cataacccct
     3781 tggggcctct aaacgggtct tgaggggttt tttgTTGACA GCTAGCTCAG TCCTAGGTAT
     3841 AATGCTAGCA TGTCGAAAGG TGAGGAACTT TTTACCGGCG TAGTCCCGAT CTTGGTGGAG
     3901 CTGGACGGGG ATGTAAACGG CCACAAATTT AGCGTACGCG GCGAGGGGGA AGGCGACGCG
     3961 ACCAACGGTA AACTGACCTT AAAGTTTATT TGTACCACCG GTAAACTCCC GGTGCCCTGG
     4021 CCAACCCTGG TTACTACCCT GACGTATGGC GTTCAGTGCT TCAGCCGTTA TCCGGATCAC
     4081 ATGAAACGCC ATGATTTCTT CAAATCCGCC ATGCCGGAAG GCTACGTCCA GGAACGTACG
     4141 ATTTCTTTTA AGGATGATGG AACATACAAA ACTCGGGCAG AAGTGAAATT CGAGGGCGAT
     4201 ACGTTGGTTA ATCGAATTGA ACTGAAGGGT ATCGATTTTA AAGAGGATGG CAACATTCTG
     4261 GGTCATAAAC TGGAATACAA CTTTAATAGC CATAACGTCT ATATTACCGC CGACAAACAG
     4321 AAAAACGGAA TTAAGGCGAA CTTCAAAATC CGCCATAATG TTGAAGATGG AAGTGTGCAG
     4381 CTGGCCGATC ATTACCAGCA AAATACGCCT ATCGGCGACG GGCCGGTTTT ACTGCCTGAC
     4441 AACCACTATC TGTCGACACA ATCAGTGCTA TCAAAGGACC CGAATGAAAA ACGTGATCAC
     4501 ATGGTGCTTC TCGAATTTGT GACCGCGGCG GGTATAACTC ATGGCATGGA CGAATTATAC
     4561 AAAGCGGCAA ATGATGAAAA TTATGCCTTG GCAGCTccag gcatcaaata aaacgaaagg
     4621 ctcagtcgaa agactgggcc tttcgtttta tctgttgttt gtcggtgaac gctctctact
     4681 agagtcacac tggctcacct tcgggtgggc ctttctgcgt ttata
//

The complete expression plasmid (phaABC_Ecoli_Final_Con, 10,017 bp circular) with the pET-28a(+) backbone is the Twist Bioscience synthesis target. Key backbone features are summarized below:

LOCUS       phaABC_Ecoli_Final_Con 10017 bp ds-DNA     circular     12-MAY-2026
DEFINITION  phaABC_Ecoli_Construct_2 in pET-28a(+). Complete expression plasmid
            for PHB biosynthesis in E. coli BL21(DE3). Insert (1-4725) contains
            T7-driven phaABC operon and sfGFP reporter. Backbone (4726-10017)
            provides kanamycin resistance, ColE1 high-copy origin, lacI repressor,
            and T7/lac regulatory elements. Created using Twist Bioscience tools.
FEATURES             Location/Qualifiers
     misc_feature    1..4725
                     /label="insert"
                     /note="phaABC operon + sfGFP reporter; see
                             phaABC_Ecoli_Construct_2.gb for full annotation"
     misc_feature    4732..4749
                     /label="6x_His_tag"
                     /note="C-terminal His-tag in backbone for optional
                             protein purification"
     misc_feature    4760..4888
                     /label="T7"
                     /note="T7 promoter in backbone (upstream read-through)"
     misc_feature    5447..6262
                     /label="kan"
                     /note="Kanamycin resistance; select at 50 ug/mL;
                             NOTE: pET-28a uses kanamycin, NOT ampicillin"
     misc_feature    6384..6972
                     /label="colE1_high_copy"
                     /note="ColE1 high-copy origin; ~15-20 copies per cell"
     misc_feature    7399..7590
                     /label="ROP"
                     /note="ROP protein; regulates plasmid copy number"
     misc_feature    8365..8386
                     /label="CAP_binding"
     misc_feature    8399..9490
                     /label="lacI"
                     /note="Lac repressor; represses T7lac promoter
                             until IPTG induction"
     misc_feature    9482..9559
                     /label="lacIQ_promoter"
     misc_feature    9868..9886
                     /label="T7"
                     /note="T7lac promoter driving insert expression"
     misc_feature    9886..9913
                     /label="lacO"
                     /note="Lac operator; IPTG-inducible expression"
     misc_feature    9926..9948
                     /label="RBS_3"
                     /note="Ribosome binding site upstream of insert"
     misc_feature    9967..9983
                     /label="6x_His_tag"
                     /note="N-terminal His-tag"
//

The Chlorella vulgaris expression construct (pPHB_Chlorella, 5,614 bp linear) GenBank annotation for Aim 2 is provided below. This construct was designed in Benchling with phaA, phaB, and phaC codon-optimized for C. vulgaris nuclear expression under the nitrate-inducible NIT1 promoter:

LOCUS       pPHB_Chlorella          5614 bp ds-DNA     linear       27-MAY-2026
DEFINITION  Synthetic PHB biosynthetic operon for Chlorella vulgaris nuclear
            expression. phaA-phaB-phaC from C. necator H16, codon-optimized
            for C. vulgaris, under NIT1 nitrate-inducible promoter. RbcS2
            terminator (BBa_K2984021) for transcript stability. AphVIII
            paromomycin resistance selectable marker for algal transformation.
FEATURES             Location/Qualifiers
     Promoter        1..1000
                     /label="pNIT1"
                     /note="Nitrate reductase NIT1 promoter from C. vulgaris;
                             1 kb upstream regulatory region; activated under
                             nitrogen limitation conditions; directly couples
                             the FBA-predicted metabolic stress signal to
                             PHB operon transcription in the target organism"
     CDS             1031..2212
                     /label="phaA_Chlorella_opt"
                     /gene="phaA"
                     /product="beta-ketothiolase"
                     /note="C. vulgaris nuclear codon-optimized; 1182 bp;
                             source: C. necator H16; converts 2x acetyl-CoA
                             to acetoacetyl-CoA"
                     /codon_start=1
     CDS             2225..2965
                     /label="phaB_Chlorella_opt"
                     /gene="phaB"
                     /product="acetoacetyl-CoA reductase (NADPH-dependent)"
                     /note="C. vulgaris nuclear codon-optimized; 741 bp;
                             reduces acetoacetyl-CoA to (R)-3-hydroxybutyryl-CoA"
                     /codon_start=1
     CDS             2978..4744
                     /label="phaC_Chlorella_opt"
                     /gene="phaC"
                     /product="polyhydroxyalkanoate synthase class I"
                     /note="C. vulgaris nuclear codon-optimized; 1767 bp;
                             rate-limiting enzyme; polymerizes
                             (R)-3-hydroxybutyryl-CoA into PHB granules"
                     /codon_start=1
     Terminator      4745..5014
                     /label="BBa_K2984021"
                     /note="RbcS2 terminator B6-C1; Rubisco small subunit 2
                             3-prime UTR; provides transcript stability and
                             correct 3-prime processing in C. vulgaris and
                             Chlamydomonas reinhardtii"
     CDS             5015..5614
                     /label="AphVIII"
                     /gene="aphVIII"
                     /product="aminoglycoside phosphotransferase VIII"
                     /note="Paromomycin resistance selectable marker; 600 bp;
                             standard for nuclear transformation in C. vulgaris
                             and C. reinhardtii; select on TAP agar with
                             10-50 ug/mL paromomycin"
                     /codon_start=1
ORIGIN
        1 CCCGGGGCAC ATCCGGGCAT CCCCTACCTC GGCATATATA CTGCGCCCCT CTCCCAACCG
       61 GGGCGCAGCA GCAGGGAGCA GGTGCTGGTG TAGGGTAAGG AAAGGATCTG ACGCACATCC
      121 ATCAACGGGT CAACAAGCCC AAGCATCTAG ATATGCACGC CAGGCTTGCG GTTGAAGGGG
      181 CATCAGGCTC GAGGCGAGAC GTCGAGGGCG TGGGCTCTGT ATGGCTGGGT AACGGTACGT
      241 ATAATTCCAG GTACAAGCTA GAGCAGACGG TGGTGAGAAG CATTAGAAGC ATTGTCCCGA
      301 GTGTGGTGGC TAGAATCCCG GCCCACGAAT CACAGTGAAT GGGTACATGT ACAGGTGCCC
      361 CGCCAGCCCC CGCTCCTCTG CTGCCTCTGA TGCCTCATGC CAAAAGTCCT GACGCGGCGC
      421 CCTCACATCC CCGTCCGGGT AATCTATGAG TTTCCCTTAT CGAGCATGTA CGCGATAGTG
      481 GACGGGGCTC AGGGTGGGGG GTGGGTGGGT GGGAGGGGCG TTCCTTCAGA CACCCTGGAG
      541 GGGTGGCTAG AAAAGCGGCC GCGCGCCAGA AATGTCTCGC TGCCCTGTGC AATAAGCACC
      601 GGCTATATTG CTCAGCGCTG TTCGGCGCAA CGGGGGGTCA GCCCTTGGGA AGCGTTGGAC
      661 TATATGGTAG GGTGCGAGTG ACCCCGCGCG ACTTGGAGCT CGATGGCCCC GGGTTGTTTG
      721 GGGCGTCCGC CTCTCGCGCT ATTCTGAGCT GGAGACCGAG GCGCATGAAA ATGCATTCGC
      781 TTCCATAGGA CGCTGCATTG TGGCTTGAAG GTTCAAGGGA AGGGTTCAAA CGACCCCGCC
      841 GTACGAACTT TTGTCGGGGG GCGCTCCCGG CCCCGGGCTC TTGTGCGCGC ATTAGGGCTT
      901 CGGGTCGCAA GCAAGACGAT ACAGGAACCG ACCAATCGAT AGTCTTGTGC GACCGTGCAC
      961 GTGTGCAGCA ATAGTTAGGT CGATAACCAC GTTGAACTTG aaaaacaaag caacagcaca
     1021 ccaacgacac ATGACTGACG TGGTGATCGT GTCCGCCGCC CGCACCGCGG TGGGCAAGTT
     1081 CGGCGGGTCG CTCGCCAAGA TCCCGGCACC CGAACTGGGT GCTGTGGTCA TCAAGGCCGC
     1141 GCTGGAGCGG GCCGGCGTAA AGCCGGAGCA GGTTAGCGAG GTCATTATGG GCCAAGTGCT
     1201 GACCGCCGGT TCGGGCCAGA ATCCCGCACG CCAGGCCGCG ATTAAGGCTG GCCTACCCGC
     1261 GATGGTGCCC GCCATGACCA TCAACAAGGT GTGCGGCTCG GGCCTTAAGG CCGTGATGCT
     1321 GGCGGCTAAC GCGATCATGG CGGGCGACGC TGAGATCGTG GTGGCTGGCG GGCAGGAGAA
     1381 CATGAGCGCC GCCCCCCATG TTCTGCCTGG CTCGCGGGAT GGTTTCCGGA TGGGCGACGC
     1441 CAAGCTGGTC GACACCATGA TCGTCGACGG CTTGTGGGAC GTTTACAACC AGTACCACAT
     1501 GGGCATTACT GCAGAGAACG TGGCGAAGGA ATACGGCATC ACACGCGAGG CGCAGGATGA
     1561 GTTCGCGGTG GGCAGCCAGA ACAAGGCAGA GGCCGCGCAG AAGGCGGGCA AGTTTGACGA
     1621 GGAAATCGTC CCAGTGCTGA TTCCCCAGCG CAAGGGCGAC CCGGTGGCGT TCAAGACCGA
     1681 CGAGTTCGTG CGCCAGGGTG CCACGCTGGA CAGCATGTCC GGCCTCAAGC CCGCCTTCGA
     1741 CAAGGCCGGC ACGGTGACCG CGGCCAACGC GAGCGGCCTG AACGACGGGG CCGCCGCTGT
     1801 GGTGGTGATG TCAGCGGCCA AGGCCAAGGA GCTGGGCCTG ACGCCTCTGG CCACGATCAA
     1861 GAGTTACGCC AACGCCGGTG TCGATCCCAA GGTGATGGGC ATGGGCCCGG TGCCGGCCTC
     1921 CAAGCGCGCT CTCAGCCGCG CCGAGTGGAC CCCGCAAGAC CTGGACCTGA TGGAAATCAA
     1981 CGAGGCCTTC GCGGCGCAGG CGCTCGCGGT GCACCAGCAG ATGGGATGGG ACACCTCCAA
     2041 GGTCAACGTG AACGGAGGCG CTATCGCCAT CGGGCACCCC ATCGGCGCGT CTGGCTGCAG
     2101 GATACTGGTG ACGCTGCTGC ACGAGATGAA GCGCCGTGAC GCTAAGAAGG GCCTGGCGTC
     2161 GCTGTGCATC GGCGGCGGCA TGGGGGTGGC GCTGGCAGTC GAGCGCAAAT AAgtttaaaa
     2221 cttaATGACT CAACGCATTG CGTACGTGAC CGGCGGGATG GGTGGCATCG GAACGGCCAT
     2281 TTGCCAGCGG CTGGCCAAAG ACGGCTTTCG AGTGGTGGCA GGTTGCGGCC CCAACTCCCC
     2341 TCGCCGCGAA AAGTGGCTGG AGCAGCAGAA GGCCCTGGGC TTCGACTTCA TTGCGTCTGA
     2401 GGGCAATGTG GCTGACTGGG ACAGCACCAA GACGGCTTTC GACAAGGTCA AGTCCGAGGT
     2461 CGGCGAGGTT GATGTGCTGA TCAACAACGC GGGAATCACC CGCGACGTGG TGTTTAGGAA
     2521 GATGACGCGC GCCGACTGGG ATGCGGTGAT CGACACCAAC CTGACTAGCC TGTTCAACGT
     2581 AACCAAGCAG GTGATCGACG GCATGGCTGA CCGTGGCTGG GGCCGCATCG TTAACATCAG
     2641 CTCGGTGAAC GGGCAGAAGG GCCAGTTCGG CCAGACCAAC TACTCCACGG CCAAGGCCGG
     2701 CCTGCACGGC TTCACCATGG CACTTGCGCA GGAGGTGGCG ACAAAGGGCG TGACCGTCAA
     2761 CACGGTGTCA CCGGGCTACA TCGCCACCGA CATGGTCAAG GCGATCCGCC AGGACGTGCT
     2821 CGACAAGATC GTGGCGACGA TCCCCGTCAA GCGGCTGGGC CTGCCGGAGG AGATCGCCTC
     2881 GATTTGCGCC TGGTTGAGCT CGGAGGAGTC CGGTTTCAGC ACCGGCGCCG ACTTCTCGCT
     2941 CAACGGCGGG CTGCACATGG GCTAAgttta aaacttaATG GCCACCGGAA AGGGTGCTGC
     3001 GGCTTCCACG CAGGAGGGCA AATCCCAGCC CTTCAAGGTG ACCCCGGGAC CGTTTGACCC
     3061 CGCTACCTGG CTGGAGTGGA GCCGACAGTG GCAGGGCACC GAGGGCAACG GTCACGCGGC
     3121 CGCTTCCGGC ATCCCCGGCC TCGACGCCCT GGCCGGCGTA AAGATCGCCC CCGCGCAGTT
     3181 AGGCGACATC CAGCAGCGGT ACATGAAGGA TTTCAGCGCT CTGTGGCAAG CCATGGCCGA
     3241 GGGCAAGGCC GAGGCGACGG GCCCCCTGCA CGACCGCCGC TTTGCGGGCG ACGCGTGGCG
     3301 CACGAACCTC CCTTACCGCT TCGCCGCGGC GTTCTACCTG CTGAACGCCC GCGCGCTGAC
     3361 CGAGCTCGCA GACGCCGTGG AGGCGGACGC CAAGACGCGT CAGCGCATCC GCTTCGCTAT
     3421 CAGCCAGTGG GTGGACGCCA TGAGCCCCGC GAACTTCCTG GCTACCAATC CGGAGGCCCA
     3481 GCGGCTGCTA ATCGAGAGCG GCGGGGAGAG CCTGAGGGCC GGCGTGCGCA ACATGATGGA
     3541 GGACCTGACA CGCGGCAAGA TTTCACAAAC CGACGAGTCG GCTTTCGAGG TGGGTCGCAA
     3601 CGTCGCGGTG ACCGAGGGGG CCGTGGTATT CGAGAACGAG TACTTCCAGT TGCTGCAGTA
     3661 CAAGCCCCTG ACCGACAAGG TGCACGCGAG GCCGCTCCTC ATGGTGCCGC CGTGCATTAA
     3721 CAAGTACTAT ATTCTGGACC TTCAGCCGGA GTCTAGCCTG GTGCGGCACG TTGTGGAGCA
     3781 GGGCCACACC GTGTTCCTAG TCTCGTGGCG CAACCCAGAC GCCTCTATGG CAGGCAGCAC
     3841 ATGGGACGAC TATATCGAAC ACGCGGCCAT CCGCGCGATT GAGGTGGCCC GTGACATCTC
     3901 GGGCCAGGAC AAGATCAACG TCCTCGGGTT CTGCGTGGGT GGTACCATTG TGAGTACTGC
     3961 CCTGGCCGTC CTGGCCGCCC GCGGCGAGCA CCCCGCGGCT TCGGTCACCC TGCTCACCAC
     4021 CCTGCTGGAC TTCGCGGACA CGGGCATCCT GGACGTGTTC GTGGACGAGG GGCATGTCCA
     4081 GCTGCGCGAG GCAACGCTGG GCGGGGGCGC CGGCGCGCCC TGCGCGCTGC TCCGCGGCCT
     4141 GGAGCTGGCG AACACGTTCT CCTTCTTACG CCCCAACGAC CTGGTGTGGA ACTACGTGGT
     4201 TGATAACTAC CTGAAGGGCA ACACTCCCGT GCCGTTTGAC CTGCTGTTCT GGAACGGCGA
     4261 CGCGACCAAC CTGCCTGGCC CGTGGTACTG CTGGTACCTG CGGCACACCT ACCTGCAGAA
     4321 CGAGCTGAAG GTCCCAGGAA AGCTGACGGT GTGTGGCGTG CCTGTGGACC TGGCCTCAAT
     4381 CGACGTGCCC ACGTACATAT ACGGCTCGCG CGAGGATCAC ATCGTGCCAT GGACCGCAGC
     4441 CTACGCCTCG ACGGCCCTTC TGGCGAACAA GCTGCGGTTC GTCTTGGGCG CTTCCGGCCA
     4501 CATCGCCGGC GTGATCAACC CCCCTGCGAA GAACAAGCGC AGCCACTGGA CTAACGATGC
     4561 CCTTCCGGAG TCCCCCCAGC AGTGGCTGGC GGGCGCAATC GAGCATCACG GCAGTTGGTG
     4621 GCCTGACTGG ACAGCCTGGC TGGCAGGCCA GGCCGGTGCG AAGCGCGCGG CACCCGCGAA
     4681 TTACGGCAAC GCCCGTTACC GCGCCATCGA GCCCGCTCCG GGCCGGTACG TTAAGGCCAA
     4741 GGCGCTGAAG ACTACTCAGC TTCCGCTCCG TGTAAATGGA GGCGCTCGTT GATCTGAGCC
     4801 TTGCCCCCTG ACGAACGGCG GTGGATGGAA GATACTGCTC TCAAGTGCTG AAGCGGTAGC
     4861 TTAGCTCCCC GTTTCGTGCT GATCAGTCTT TTTCAACACG TAAAAAGCGG AGGAGTTTTG
     4921 CAATTTTGTT GGTTGTAACG ATCCTCCGTT GATTTTGGCC TCTTTCTCCA TGGGCGGGCT
     4981 GGGCGTATTT GAAGCGCGCT CGAGGAGTCT TCAGatggcc gccatgaccg acctcgaccg
     5041 cctcggcgtg ctcgagcccg ccgaccccga gcgcatccgc gccaagtaca tcgagctcgc
     5101 ccgcgagcgc gccaccgacg ccgactacta cctcgccacc gccttcgacg tcgaggccgc
     5161 cgacgaggac cgccgcgtca tcgccgtcct cgccgaggcc ctcgacacca ccggcgaggt
     5221 cgccgccgcc gtcgtctacg gcaacgcccc gctcgccaag gccctcgacc gcgtcctcgc
     5281 cgagtacccc gccgtcgccc gcctcaaggg cgtctccatg ctctacggcc gccacgaggt
     5341 cgccctcgac ctcgacgcca tcaccaccct caaggcccag ggcggccgcg agatcatcct
     5401 cgccgacgcc gtcatggacc agcgcgagat catcgagaag gccgccgacc gcatccccat
     5461 cgacgccacc ctcaccacca tcgagaccat ccgctacctc atggcccacg agccctccac
     5521 cctcgagcac ctcggctggc gcaagtcctc caagccctcc gccgaggcca agccctccaa
     5581 ggcccccaag gcccccaagg cccccaagac cacc
//

Step 8: Place the Twist Bioscience DNA synthesis order

• Method: Submit the finalized pPHB-Ecoli GenBank sequence to Twist Bioscience via their online ordering portal using the Clonal Gene Synthesis with selected backbone service; specify shipping to Ginkgo Bioworks for subsequent experimental steps.
• Automation: Twist Bioscience online ordering portal (twistbioscience.com)
• Expected result: Order confirmation email; estimated synthesis and delivery turnaround of 2 to 3 weeks.
• Timeline: Day 5 to 6

Step 9: Program and run the OT-2 media preparation module (Checkpoint A)

• Method: Write a Python script using the Opentrons OT-2 API to prepare 96-well deep-well plates with the top 8 FBA-predicted nutrient stress media conditions in triplicate; the protocol ends cleanly at Checkpoint A for users who want only prepared media before proceeding to cell culture in a separate facility.
• Automation: Opentrons OT-2 (Python API v2.15); Plateloc plate sealer with A4s breathable seal after preparation
• Plate: 96-v-eppendorf-951033502-deep
• Expected result: Eight distinct media conditions prepared in triplicate (24 wells), sealed with breathable membrane and ready for inoculation at Ginkgo or an external laboratory.
• Timeline: Day 6 to 8

from opentrons import protocol_api
 
metadata = {
    'protocolName': 'PHB_NutrientStress_MediaPrep_CheckpointA',
    'author': 'Fabrizio Flores Huaman',
    'description': (
        'Automated preparation of FBA-predicted nutrient stress media for PHB '
        'screening. CHECKPOINT A: protocol terminates here if only media prep '
        'is needed. Seal plate with A4s breathable seal and store at 4C (Tundrastore) '
        'or proceed to inoculation. Continue to Screening Module if cells are ready.'
    )
}
requirements = {'robotType': 'OT-2', 'apiLevel': '2.15'}
 
def run(protocol: protocol_api.ProtocolContext):
    # Labware
    tiprack_300 = protocol.load_labware('opentrons_96_tiprack_300ul', 1)
    tiprack_20  = protocol.load_labware('opentrons_96_tiprack_20ul', 4)
    reservoir   = protocol.load_labware('nest_12_reservoir_15ml', 2)
    deep_plate  = protocol.load_labware('nest_96_wellplate_2ml_deep', 3)
 
    p300 = protocol.load_instrument('p300_multi_gen2', 'right', tip_racks=[tiprack_300])
    p20  = protocol.load_instrument('p20_multi_gen2',  'left',  tip_racks=[tiprack_20])
 
    # Reservoir layout
    base_media = reservoir['A1']   # LB or BG-11 minimal salts base
    n_source   = reservoir['A2']   # NaNO3 stock (nitrogen source)
    p_source   = reservoir['A3']   # KH2PO4 stock (phosphorus source)
    c_glucose  = reservoir['A4']   # 20% glucose stock (carbon source A)
    c_acetate  = reservoir['A5']   # Sodium acetate stock (carbon source B)
    diluent    = reservoir['A6']   # Sterile water (diluent for gradient)
 
    # Dispense 1200 uL base media to all 24 test wells (columns 1-3, rows A-H)
    test_cols = deep_plate.columns()[:3]
    for col in test_cols:
        p300.transfer(1200, base_media, col,
                      mix_after=(3, 200), new_tip='always')
 
    # Apply FBA-predicted nutrient stress recipes
    # Row layout (conditions in triplicate across columns 1-3):
    # Row A: N0%-P100%-Glucose   | Row B: N25%-P100%-Glucose
    # Row C: N50%-P100%-Glucose  | Row D: N0%-P50%-Glucose
    # Row E: N0%-P0%-Glucose     | Row F: N50%-P50%-Acetate (top FBA candidate)
    # Row G: N100%-P100%-Glucose | Row H: N100%-P100%-No exogenous carbon
 
    nitrogen_vols = [0, 12.5, 25, 0, 0, 25, 50, 50]  # uL of NaNO3 stock per row
    phosphorus_vols = [50, 50, 50, 25, 0, 25, 50, 50]
    carbon_sources = [c_glucose, c_glucose, c_glucose, c_glucose,
                      c_glucose, c_acetate, c_glucose, diluent]
 
    for row_idx, (n_vol, p_vol, c_src) in enumerate(
            zip(nitrogen_vols, phosphorus_vols, carbon_sources)):
        row_letter = 'ABCDEFGH'[row_idx]
        wells = [deep_plate[f'{row_letter}{col+1}'] for col in range(3)]
        if n_vol > 0:
            p20.transfer(n_vol, n_source, wells, new_tip='always')
        if p_vol > 0:
            p20.transfer(p_vol, p_source, wells, new_tip='always')
        p20.transfer(50, c_src, wells, mix_after=(2, 15), new_tip='always')
 
    protocol.comment("=" * 60)
    protocol.comment("CHECKPOINT A: Media preparation complete.")
    protocol.comment("Seal plate with A4s breathable seal using Plateloc.")
    protocol.comment("Store at 4C in Tundrastore if not inoculating immediately.")
    protocol.comment("To continue: proceed to Screening Module (Step 10 onward).")
    protocol.comment("=" * 60)

Step 10: Inoculate cultures and incubate

• Method: Inoculate each prepared deep-well with E. coli BL21 (DE3) harboring pPHB-Ecoli at a starting optical density at 600 nm (OD600) of 0.05; seal plate with A4s breathable seal.
• Automation: Plateloc (sealing with A4s breathable membrane); Cytomat shaking incubator at 37 degrees Celsius, 250 rpm
• Plate: 96-v-eppendorf-951033502-deep
• Expected result: Growth visible in positive control wells within 6 to 8 hours; reduced growth rate in nutrient-limited wells confirming that stress conditions are active.
• Timeline: Day 8 to 9 (overnight incubation, approximately 16 hours)

Step 11: Measure growth dynamics by optical density

• Method: Transfer 100 µL from each deep-well to a flat-bottom 96-well plate using the Echo525; measure OD600 using the Spark Plate Reader to assess growth rates and confirm that nutrient limitation is reducing but not eliminating cell viability.
• Automation: Echo525 (precise volume transfer); Spark Plate Reader (OD600 measurement)
• Plate: 96-round-axygen-pdw11cs-halfdeep for transfer; flat 96-well for reading
• Expected result: OD600 values confirming a gradient of growth suppression correlated with increasing nutrient limitation; no complete cell death in any well.
• Timeline: Day 9

Step 12: Perform Nile Red staining for PHB quantification

• Method: Add Nile Red dissolved in dimethyl sulfoxide (DMSO) at 1 mg/mL to each well using the Echo525 at a 1:1000 final dilution (1 µL per mL culture); incubate 15 minutes at room temperature in the dark; transfer 50 µL to 384-well black clear-bottom plate for fluorescence reading.
• Automation: Echo525 (Nile Red addition, nanoliter precision); Multiflo (50 µL culture transfer to 384-well); PHERAstar FSX (Ex 530 nm / Em 580 nm)
• Plate: 384 Greiner black-well clear-bottom
• Expected result: Higher Nile Red fluorescence in nutrient-limited conditions compared to full-media controls, consistent with increased PHB granule accumulation.
• Timeline: Day 9 to 10

Step 13: Validate sfGFP reporter expression

• Method: Measure sfGFP fluorescence (Ex 485 nm / Em 510 nm) on the Spark Plate Reader using the same 384-well plate as the Nile Red assay, before Nile Red addition, to confirm construct expression and identify any wells where failed transformation rather than PHB pathway activity is responsible for low Nile Red signal.
• Automation: Spark Plate Reader (sequential dual-channel fluorescence measurement)
• Plate: 384 Greiner black-well clear-bottom
• Expected result: Measurable sfGFP signal in all transformed wells; absence of sfGFP in empty vector controls; allows normalization of Nile Red signal to expression level.
• Timeline: Day 9 to 10 (before Nile Red addition)

Step 14: Validate PHB gene expression by quantitative PCR

• Method: Harvest cells from the top 3 performing conditions and the negative control; extract total RNA using an RNeasy kit; perform quantitative polymerase chain reaction (qPCR) to measure phaA, phaB, and phaC transcript levels relative to the 16S ribosomal RNA housekeeping gene.
• Automation: CFX Opus qPCR machine
• Plate: 384-pcr-eppendorf-9510207XX
• Expected result: Upregulation of phaA, phaB, and phaC transcripts in the top nutrient stress conditions relative to full-media controls, confirming that the T7 promoter is active and that the operon is being transcribed in proportion to the metabolic state of the cell.
• Timeline: Day 10 to 11

Step 15: Close the DBTL loop by refining the FBA model

• Method: Feed the experimental Nile Red fluorescence values and qPCR expression data back into the COBRApy model as refined constraints; adjust exchange reaction bounds to reflect observed growth rates and acetyl-CoA sink capacity; re-run FBA with the updated model to identify any systematic discrepancies between prediction and observation.
• Automation: COBRApy model update; pandas for data integration; optional MEMOTE framework for model quality scoring
• Expected result: An updated model with improved predictive accuracy for a second round of experimental screening or for informing the C. vulgaris construct design in Aim 2.
• Timeline: Day 11 to 13

Step 16: Optional supplementary characterization via FTIR and GC-MS

• Method: Lyophilize the cell pellets from the top two performing conditions; extract PHB polymer using a standard chloroform solvent extraction protocol; analyze bulk polymer by Fourier-transform infrared spectroscopy (FTIR) for functional group confirmation and gas chromatography-mass spectrometry (GC-MS) via the Waters Corporation platform at Ginkgo Bioworks or a partner facility for monomer composition and yield quantification.
• Automation: Waters GC-MS system (Ginkgo Bioworks partner; connection to Waters Corporation industry partner)
• Expected result: FTIR spectra showing a characteristic PHB carbonyl peak at 1720 cm-1; GC-MS confirmation of 3-hydroxybutyrate monomer and quantitative PHB yield in mg per gram dry cell weight.
• Timeline: Day 12 to 14 (if resources permit)

Assay Plate Layout

The following 96-well plate layout illustrates the screening design for the Nile Red fluorescence assay. Conditions are derived from the top FBA-predicted nutrient stress recipes, with all experimental conditions run in duplicate.

        Col 1           Col 2           Col 3           Col 4           Col 5           Col 6
Row A  [STD  0µg/mL]  [STD  0µg/mL]  [POS_CTRL]      [POS_CTRL]      [N0-P100-Glc]   [N0-P100-Glc]
Row B  [STD  1µg/mL]  [STD  1µg/mL]  [POS_CTRL]      [POS_CTRL]      [N25-P100-Glc]  [N25-P100-Glc]
Row C  [STD  2µg/mL]  [STD  2µg/mL]  [NEG_CTRL]      [NEG_CTRL]      [N50-P100-Glc]  [N50-P100-Glc]
Row D  [STD  5µg/mL]  [STD  5µg/mL]  [NEG_CTRL]      [NEG_CTRL]      [N0-P50-Glc]    [N0-P50-Glc]
Row E  [STD 10µg/mL]  [STD 10µg/mL]  [BLANK]         [BLANK]         [N0-P0-Glc]     [N0-P0-Glc]
Row F  [STD 25µg/mL]  [STD 25µg/mL]  [BLANK]         [BLANK]         [N50-P50-Ace]   [N50-P50-Ace]
Row G  [STD 50µg/mL]  [STD 50µg/mL]  [BLANK]         [BLANK]         [N100-P100-Glc] [N100-P100-Glc]
Row H  [STD100µg/mL]  [STD100µg/mL]  [BLANK]         [BLANK]         [N100-P100-NoC] [N100-P100-NoC]

Plate Legend:

• STD: Polyhydroxybutyrate (PHB) polymer standard curve (0 to 100 µg/mL; Nile Red fluorescence calibration; Millipore Sigma catalog 363502)
• POS_CTRL: Positive control (E. coli BL21 DE3 with a known PHB-producing construct or C. necator ATCC 17699)
• NEG_CTRL: Negative control (E. coli BL21 DE3 with empty pUC19 vector; no PHB genes)
• BLANK: Sterile media only; no cells; used for background fluorescence subtraction
• N[x]-P[x]-[Carbon]: Experimental conditions where x represents percent of standard nitrogen or phosphorus concentration
• Glc: Glucose; Ace: Sodium acetate; NoC: No exogenous carbon source (autotrophic simulation)
• All experimental conditions run in duplicate (two adjacent wells per condition)
• After initial assay, the 384-well Greiner black clear-bottom plate is used for the actual Nile Red fluorescence measurement with 50 µL transfers from each 96-well condition

SECTION 4.2: TECHNIQUES, TOOLS, AND TECHNOLOGY

Course Technique Checklist

• Metabolic modeling and Flux Balance Analysis (FBA) using COBRApy
• DNA construct design and sequence annotation using Benchling
• Synthetic gene and whole plasmid synthesis ordering from Twist Bioscience
• Codon optimization for heterologous expression (E. coli and C. vulgaris)
• Lab automation and liquid handling (Opentrons OT-2, Echo525, Multiflo)
• Bacterial whole-cell expression in E. coli BL21 (DE3)
• Cell-free protein synthesis (CFPS) for rapid construct validation
• Fluorescence assay development and microplate reader detection (Spark, PHERAstar FSX)
• Quantitative PCR (qPCR) for gene expression validation (CFX Opus)
• Bioinformatics and Python scripting for data analysis and model interrogation
• Microplate screening and high-throughput assay design (96-well and 384-well formats)
• Biosafety and SecureDNA screening for synthetic DNA sequences
• CRISPR genome editing (relevant for Aim 2 in C. vulgaris; not executed in Aim 1)
• Flow cytometry (optional alternative to plate reader Nile Red assay)
• Next-generation sequencing (could be used for construct integration verification in Aim 2)

We discussed and practiced various techniques related to synthetic biology throughout the semester. Place a check next to the techniques relevant to your project.

Pipetting

• Pipetting
• Lab Safety
• Bioethical Considerations (must check this box)

DNA Gel Art

• DNA Sequencing
• DNA Editing
• DNA Construct Design
• Restriction Enzyme Digestion
• Gel Electrophoresis
• DNA Purification From Gel
• Databases (e.g., GenBank, NCBI, Ensembl, and UCSC Genome Browser)

Lab Automation

• Creating Code for Laboratory Automation
• Using Liquid Handling Robots (e.g., Opentrons)
• Designing a Twist Order
• Creating a plan to use the Autonomous lab at Ginkgo Bioworks

Protein Design

• Protein Design
• Use of Boltz or PepMLM
• Use of Asimov Kernel
• Use of Benchling
• Models and Notebooks
• Databases

Bioproduction

• Bioproduction
• Chassis Selection (e.g., DH5alpha)
• Registry of Standard Biological Parts
• Plasmid Preparation
• Bacterial Culturing
• Quality Control/Analysis
• Bacterial Processing (e.g., Centrifugation, Lysis, DNA Purification)

Cell-Free Systems

• Cell Free Reactions
• Freeze-Dried Cell Free Systems
• miniPCR Tools
• Protein Purification

Gibson Assembly

• Primer Design or Selection
• PCR Reactions
• Gibson Assembly
• Other Cloning Methods (e.g., Restriction Enzyme Digestion or Gateway Cloning)

CRISPR

• CRISPR/Cas9
• Designing Prime Editing gRNA

Technique Expansion

Technique 1: Flux Balance Analysis (FBA) with COBRApy

Flux Balance Analysis is a constraint-based computational method that represents cellular metabolism as a network of stoichiometric equations, where each equation describes how metabolites are converted into one another by enzymatic reactions. COBRApy is the Python implementation of the COBRA (COnstraint-Based Reconstruction and Analysis) toolbox, which allows researchers to load genome-scale metabolic models in SBML format and apply linear programming solvers such as GLPK or Gurobi to find the flux distribution that maximizes a defined biological objective, such as biomass growth or production of a target metabolite. In the context of this project, FBA is uniquely powerful because it enables the simultaneous exploration of hundreds of nutrient combinations without requiring physical experiments, making it a fast and cost-effective Design phase tool in the DBTL cycle. The key limitation of FBA is that it assumes steady-state metabolic conditions and does not capture dynamic gene regulatory responses, such as the transcriptional induction of PHB pathway genes under stress, which is why experimental validation through the OT-2 pipeline and qPCR is an essential complement to the computational predictions. By connecting FBA outputs directly to parameterized OT-2 protocols, this project creates a tightly coupled workflow where computational and physical experimentation reinforce rather than simply follow each other.

Technique 2: Nile Red Fluorescence Assay for Intracellular PHB Quantification

Nile Red is a lipophilic fluorescent dye that partitions selectively into hydrophobic intracellular environments, including lipid droplets and polymer granules such as PHB inclusions, producing a fluorescence signal with an excitation maximum near 530 nm and an emission maximum near 580 nm that scales linearly with polymer content across several orders of magnitude. The assay is performed directly in microplate wells without cell disruption, making it fully compatible with high-throughput robotic workflows and the PHERAstar FSX plate reader available at Ginkgo Bioworks, and allowing the same plate to be used for both growth monitoring and biopolymer quantification. One important technical consideration is that Nile Red is not strictly specific to PHB and will also stain other neutral lipids such as triacylglycerols, so the assay is best interpreted as a measure of total neutral lipid content unless confirmatory techniques such as FTIR or GC-MS are applied to establish polymer identity. In this project, Nile Red serves as the rapid, high-throughput screening layer in the pipeline, allowing the top-performing nutrient conditions to be ranked and selected before committing to the more resource-intensive and time-consuming confirmatory analyses, demonstrating how orthogonal methods at different resolution levels can be strategically combined in an automated screening workflow.

Identify any How To Grow (Almost) Anything Industry Council companies which are associated with your final project (optional)

• Addgene
• Asimov (Kernel)
• ATCC
• Basecamp Research
• BioFabricate
• Biome Consortia
• Bolt
• Boltz.bio
• Cultivarium
• DeepCure
• Epibone
• Ginkgo Bioworks
• Helix Nano
• Millipore Sigma
• Mycoworks
• New England Biolabs
• Nuclera
• Opentrons
• SecureDNA
• Takeda Pharmaceuticals
• Thermo Fisher Scientific
• Transfyr.ai
• Twist Biosciences
• Upside Foods
• Waters Corporation

SECTION 5: RESULTS AND QUANTITATIVE EXPECTATIONS

10a: Validation Choice

The validation experiment for this project is a cell-free expression assay of the phaA-phaB-phaC operon in an E. coli BL21 (DE3) lysate system, followed by Nile Red fluorescence measurement to confirm PHB granule formation from the synthetic construct in the absence of a complete living cell chassis. This approach is particularly well-suited for a primarily in silico project because it provides a rapid, contained, and robotically compatible way to confirm that the designed construct produces active enzymes and that the phaABC-encoded pathway generates detectable PHB signal within hours of receiving the DNA, establishing proof-of-function before committing to multi-day culture experiments.

10b: Validation Protocol

1. Prepare E. coli BL21 (DE3) cell-free protein synthesis (CFPS) master mix using a Ginkgo Bioworks CFPS kit or equivalent commercial system (e.g., PURExpress from New England Biolabs).
2. Thaw the pPHB-Ecoli plasmid received from Twist Bioscience on ice and dilute to 10 nM in nuclease-free water.
3. Set up the following controls in separate wells: (a) positive control with a 10 nM sfGFP-only expression plasmid, (b) negative control with no DNA added, (c) PHB polymer standard curve at 0, 1, 2, 5, 10, 25, 50, and 100 µg/mL using Millipore Sigma PHB standard (catalog 363502).
4. Prepare 10 µL CFPS reactions in 1.5 mL Eppendorf tubes by combining 7.5 µL CFPS master mix with 1 µL plasmid (final 10 nM) and 1.5 µL nuclease-free water; also supplement with 0.5 mM acetyl-CoA (Millipore Sigma A2056) and 0.1 mM coenzyme A to support PHB precursor availability.
5. Dispense 10 µL reactions into 384-well Greiner black clear-bottom plate using Echo525 acoustic liquid handler for nanoliter precision.
6. Seal plate using Plateloc with a standard film seal (not breathable) to prevent evaporation during incubation.
7. Incubate at 30 degrees Celsius for 4 hours in the Inheco Plate Incubator.
8. Peel seal using XPeel; measure sfGFP fluorescence (Ex 485 nm / Em 510 nm) on the Spark Plate Reader to confirm construct expression before Nile Red addition.
9. Add 0.5 µL Nile Red stock (1 mg/mL in DMSO) per well using the Echo525 (1:20 dilution for 10 µL wells); final Nile Red concentration approximately 50 µg/mL.
10. Incubate plate at room temperature for 15 minutes in the dark (cover with aluminum foil or store in lightproof box).
11. Measure Nile Red fluorescence (Ex 530 nm / Em 580 nm) on the PHERAstar FSX using the well-resolved fluorescence intensity module.
12. Subtract background fluorescence values from no-DNA blank wells from all experimental wells.
13. Normalize Nile Red signal to sfGFP signal in each well to correct for variation in expression level between replicate reactions.
14. Generate a standard curve from PHB polymer standards and convert normalized fluorescence values to estimated PHB content in µg/mL.
15. A validation result is positive if the pPHB-Ecoli construct produces a Nile Red signal greater than 3 standard deviations above the no-DNA control after normalization to sfGFP expression.

10c: Techniques Used

This validation experiment integrates four core techniques that span the full technical scope of the project. First, cell-free protein synthesis (CFPS) provides a rapid in vitro expression system that bypasses the multi-week timelines of stable cell transformation and selection, allowing construct function to be assessed within a single working day of receiving the synthesized plasmid from Twist Bioscience. Second, Echo525 acoustic liquid handling ensures nanoliter-precision dispensing of both the CFPS reagents and the Nile Red dye, eliminating the pipetting variability that would otherwise mask subtle differences in PHB signal between conditions and making the assay compatible with fully automated workflows at Ginkgo Bioworks. Third, dual fluorescence detection on both the Spark Plate Reader (for sfGFP) and the PHERAstar FSX (for Nile Red) provides an internal normalization strategy, ensuring that differences in Nile Red signal reflect genuine differences in PHB polymer accumulation rather than artifacts from varying expression efficiency between reaction wells. Fourth, the use of a PHB polymer standard curve converts raw relative fluorescence units into physically interpretable concentration units, grounding the validation results in quantitative terms that can be directly compared to the acetyl-CoA flux values predicted by the FBA model and used to evaluate whether the computational predictions are consistent with experimental reality.

10d: Hypothetical Data

The following table represents a hypothetical subset of Nile Red fluorescence results from the cell-free validation experiment, normalized to sfGFP expression and converted to estimated PHB content using the standard curve:

These hypothetical results suggest a positive and roughly linear correlation between the FBA-predicted acetyl-CoA flux and the measured Nile Red-normalized PHB content, consistent with the hypothesis that FBA can accurately rank nutrient stress conditions by their bioplastic potential. The FBA-predicted optimal condition (N50%-P50%-Acetate carbon source) produces the highest normalized PHB signal, validating the in silico approach and justifying the use of acetyl-CoA flux as a proxy for polymer accumulation.

Troubleshooting

One of the most likely challenges in this project is encountering errors when loading or running FBA on the iCZ843 model, since genome-scale models can contain blocked reactions, metabolite charge imbalances, or infeasible constraint combinations that prevent the optimizer from finding a solution. If the model returns an infeasible status or unrealistically extreme objective values, the recommended approach is to run the MEMOTE model quality assessment tool to identify and annotate problematic reactions, and to apply standard growth media constraints before adding the PHB demand reaction. A second potential issue is low or absent Nile Red fluorescence signal in CFPS reactions despite confirmed sfGFP expression, which most likely results from insufficient acetyl-CoA availability in the cell-free lysate rather than a non-functional phaABC operon. In that case, supplementing the CFPS reaction with exogenous acetyl-CoA at 0.5 to 2 mM and coenzyme A at 0.1 mM is a standard troubleshooting step that can significantly improve PHB signal. A third limitation is that the E. coli CFPS system provides only a coarse proxy for PHB production in a living cell, and in particular does not capture the growth-competition dynamics between biomass production and polymer accumulation that FBA is designed to model. Finally, the codon-optimized sequences designed for E. coli will not directly predict expression efficiency in C. vulgaris in Aim 2, and the significantly different GC content preferences and translation initiation contexts between the two organisms mean that the algal construct will require independent validation through a separate CFPS system or direct transformation, which is why Aim 2 is framed as a distinct experimental stage requiring specialized facilities rather than a simple extension of Aim 1.

SECTION 6: ADDITIONAL INFORMATION

References

• Orth JD, Thiele I, Palsson BO. (2010). What is flux balance analysis? Nature Biotechnology, 28(3), 245-248. https://doi.org/10.1038/nbt.1614

• Peoples OP, Sinskey AJ. (1989). Poly-beta-hydroxybutyrate (PHB) biosynthesis in Alcaligenes eutrophus H16. Identification and characterization of the PHB polymerase gene (phbC). Journal of Biological Chemistry, 264(26), 15298-15303.

• Ebrahim A, Lerman JA, Palsson BO, Hyduke DR. (2013). COBRApy: COnstraint-Based Reconstruction and Analysis for Python. BMC Systems Biology, 7, 74. https://doi.org/10.1186/1752-0509-7-74

• Zuñiga C, Li CT, Huelsman M, Andrews J, Kerkhoven EJ, Zengler K. (2016). Genome-scale metabolic model for the green alga Chlorella vulgaris UTEX 395 accurately predicts phenotypes under autotrophic, heterotrophic, and mixotrophic growth conditions. Plant Physiology, 172(1), 589-602. https://doi.org/10.1104/pp.16.00593

• Al-Hammadi M, Güngörmüşler M. (2024). New insights into Chlorella vulgaris applications. Biotechnology and Bioengineering, 121(5), 1486-1502. https://doi.org/10.1002/bit.28666

• Hempel F, Bozarth AS, Lindenkamp N, Klingl A, Zauner S, Linne U, Steinbüchel A, Maier UG. (2011). Microalgae as bioreactors for bioplastic production. Microbial Cell Factories, 10, 81. https://doi.org/10.1186/1475-2859-10-81

• Cooksey KE, Guckert JB, Williams SA, Callis PR. (1987). Fluorometric determination of the neutral lipid content of microalgal cells using Nile Red. Journal of Microbiological Methods, 6(6), 333-345.

• Oscanoa Huaynate AI, Cervantes Gallegos MA, Flores Ramos L, Ruiz Soto A. (2021). Evaluación del potencial de Desmodesmus asymmetricus y Chlorella vulgaris para la remoción de nitratos y fosfatos de aguas residuales. Revista Peruana de Biología, 28(1), e18082. https://doi.org/10.15381/rpb.v28i1.18082

• Mariano Astocondor M, Mayta Huatuco E, Montoya Terreros H, Tarazona Delgado R. (2017). Crecimiento poblacional y productividad de la microalga nativa Chlorella peruviana bajo diferentes salinidades. Revista de Investigaciones Veterinarias del Perú, 28(4), 976-986. https://doi.org/10.15381/rivep.v28i4.13004

• Condori MAM, Gutierrez MEV, Oviedo RDN, Choix FJ. (2024). Valorization of nutrients from fruit residues for the growth and lipid production of Chlorella sp.: A vision of the circular economy in Peru. Journal of Applied Phycology, 36(1), 101-111. https://doi.org/10.1007/s10811-023-03153-2

• Loera-Quezada MM, Rios-Castro E, Orta-Zavalza E, et al. (2016). CRISPR-Cas9-based genome editing approaches in the model green microalga Chlamydomonas reinhardtii. Plant Methods, 12, 30. [Referenced for algal CRISPR design rationale applicable to Aim 2.]

• Arora et al. (2024). Metabolic response to a heterologous poly-3-hydroxybutyrate (PHB) pathway in Phaeodactylum tricornutum. Applied Microbiology and Biotechnology. https://doi.org/10.1007/s00253-023-12823-7

• Yi YC, Ng IS. (2025). Sustainable polyhydroxybutyrate production via metabolic flux redirection using CRISPRi in Escherichia coli and carbon capture with microalgae. ACS Sustainable Chemistry and Engineering. https://doi.org/10.1021/acssuschemeng.4c09128

• Chaturvedi et al. (2023). Systematizing microbial bioplastic production for developing sustainable bioeconomy: metabolic nexus modeling, economic and environmental technologies assessment. Journal of Polymers and the Environment. https://doi.org/10.1007/s10924-023-02787-0

• Addgene. (2026). sfGFP plasmid resources and sequence data. https://www.addgene.org

• SecureDNA Foundation. (2026). Biosecurity screening for synthetic DNA. https://securedna.org

• Twist Bioscience. (2026). Clonal gene synthesis ordering portal. https://www.twistbioscience.com

• Opentrons. (2026). OT-2 Python API documentation v2.15. https://docs.opentrons.com

Supplies and Budget