## 7. References and Further Reading

1. **P<sub>osmY</sub> Promoter Characterization:**  
   Yim, H. H., & Villarejo, M. (1992). osmY, a new hyperosmotically inducible gene, encodes a periplasmic protein in *Escherichia coli*. *Journal of Bacteriology*, 174(11), 3637-3644.

2. **mCherry Fluorescent Protein:**  
   Shaner, N. C., et al. (2004). Improved monomeric red, orange and yellow fluorescent proteins derived from *Discosoma* sp. red fluorescent protein. *Nature Biotechnology*, 22(12), 1567-1572.

3. **Microplate Reader Assays:**  
   Myers, J. A., et al. (2013). Improving accuracy of cell and chromophore concentration measurements using optical density. *BMC Biophysics*, 6(1), 4.

4. **Synthetic Biology Education:**  
   Smanski, M. J., et al. (2014). Functional optimization of gene clusters by combinatorial design and assembly. *Nature Biotechnology*, 32(12), 1241-1249.

---

## Appendix A: Preparation of 10× Concentrated LB

**Recipe (for 100 mL):**
- Tryptone: 100 g
- Yeast Extract: 50 g
- NaCl: 100 g
- Distilled H₂O: to 100 mL final volume

**Procedure:**
1. Weigh out components and dissolve in 80 mL distilled water with stirring
2. Adjust to 100 mL final volume
3. Autoclave at 121°C for 20 minutes
4. After cooling, filter-sterilize through 0.22 µm filter to remove particulates
5. Store at 4°C for up to 1 month; check for precipitates before use

**Quality Control:**
- Dilute 1:10 and measure pH (should be 7.0 ± 0.2)
- Perform growth test: diluted 10× LB should support E. coli growth equivalent to standard LB

---

## Appendix B: Chloramphenicol Stock Preparation

**1000× Stock (34 mg/mL in 100% ethanol):**
1. Weigh 340 mg chloramphenicol in chemical fume hood
2. Dissolve in 10 mL absolute ethanol (molecular biology grade)
3. Mix by vortexing until fully dissolved
4. Aliquot into 1 mL microcentrifuge tubes
5. Store at -20°C (stable for 1 year)
6. Thaw aliquot before use; do not refreeze

**Working Concentration:** Add 1 µL of 1000× stock per 1 mL of media for final concentration of 34 µg/mL

## LAYER 3: ELECTROCHEMICAL SENSING SUBSYSTEM

This is the new core of v2. Two unengineered biological signals are read directly from the well using miniaturized electrodes embedded in the spacer frame.

### Signal 1: pH (Metabolic Acid Production)

**What it measures:** As E. coli actively metabolize LB (consuming amino acids, sugars, and organic nitrogen), they excrete metabolic byproducts including acetate, succinate, formate, and lactate. This naturally acidifies the media. A healthy, growing culture drops pH from ~7.2 (fresh LB) toward ~5.8–6.2 over 4–12 hours.

**Game meaning:** pH dropping = bacteria are metabolically active = "alive and growing." pH flat at baseline = dormant or freshly fed. pH flat at low value for extended time = nutrients exhausted, culture stressed.

**Electrode design:**

The pH electrode cannot be a standard glass probe — too fragile, too large, and reference junction too bulky for a 7mm well. Instead, use a two-electrode system threaded through holes drilled or printed into the spacer frame sidewall:

- Working electrode: IrOx-coated platinum wire, 0.5mm diameter, inserted through spacer frame wall at a 30° downward angle into the agar layer. IrOx (iridium oxide) has a Nernstian pH response of ~59mV/pH unit, is biocompatible, and can be electrodeposited onto Pt wire in the lab (protocol: cycle in 0.4mM IrCl₃ + 0.4mM oxalic acid solution, 25 cycles, −0.5V to +1.4V at 50mV/s).
- Reference electrode: Ag/AgCl wire (0.5mm diameter silver wire, chloridized by immersion in FeCl₃ solution for 10 min or by brief anodization in NaCl). Inserted through the opposite side of the spacer frame into the liquid LB layer. This provides a stable reference potential independent of solution composition.

**Signal conditioning:** IrOx pH electrodes have high source impedance (~MΩ). The signal must be buffered before reaching the Arduino analog pin. Use a single-supply rail-to-rail op-amp in voltage-follower configuration (unity gain buffer) between the working electrode output and Arduino A0:

- Recommended op-amp: MCP6001 (1.8–6V supply, rail-to-rail I/O, single-supply, SOT-23 package, ~$0.40). Wire: VDD to 5V, VSS to GND, V+ from IrOx wire, V− connected to Vout (follower config), Vout to A0.
- Expected output range: ~0.2V (pH 4) to ~1.0V (pH 8) relative to Ag/AgCl reference. Calibrate at startup using two-point calibration buffers (pH 4.0 and pH 7.0 standard buffer solutions).

**Arduino read:** Analog A0, 10-bit ADC. Calibrate slope (mV/pH) and intercept at device startup using known buffers. Store calibration constants in EEPROM.

---

### Signal 2: Electrochemical Impedance Spectroscopy (EIS) via AD5933

**What it measures:** The AD5933 applies a small sinusoidal voltage across two electrodes in the well and measures the resulting current to extract the complex impedance (Z = R + jX) of the sample. In a bacterial culture:

- **Cell density:** As bacteria proliferate, they displace conductive medium and add membrane capacitance. Bulk resistance (R) increases with cell crowding at low frequencies; double-layer capacitance (X) increases with total cell membrane area. A growing healthy culture shows characteristic impedance changes detectable within 1–2 hours.
- **Membrane integrity:** Healthy cells maintain high intracellular ion concentration behind an intact lipid bilayer — this contributes a capacitive signature at 10–100kHz. Dead or lysed cells lose membrane integrity, causing characteristic phase angle shifts and capacitance drop. This distinguishes "cells present but dead" from "cells present and healthy."

**Game meaning:** Impedance stable in expected range = population healthy and intact. Impedance falling (phase angle shifting toward purely resistive) = cells dying, membranes compromised. Impedance rising abnormally = possible biofilm formation or media precipitation.

**IC: AD5933 (Analog Devices)**
- I2C interface, address 0x0D (no conflict with SSD1306 at 0x3C)
- Frequency range: 1kHz – 100kHz (programmable sweep)
- Output excitation: 200mV peak-to-peak (suitable for biological samples — non-damaging)
- Supply: 3.3V or 5V (tie to 3.3V to protect OLED on shared I2C bus)
- Returns: real (R) and imaginary (X) impedance components as 16-bit integers

**Gain-setting resistor (RFB):** Critical calibration component. RFB sets the transimpedance gain of the AD5933's internal current-to-voltage converter. It must be matched to the expected well impedance range. For LB agar + liquid layer at 7mm well diameter:

- Expected impedance: ~500Ω–5kΩ at 10–50kHz
- Recommended RFB: 1kΩ (1% tolerance) — this keeps the AD5933 output in its linear range for the expected well impedance. If impedance is consistently saturating or clipping (check by verifying |Z| = √(R²+X²) makes physical sense), swap RFB to 2.2kΩ or 510Ω accordingly.

**EIS electrode design:**

Two-electrode configuration (adequate for relative monitoring; 4-electrode Kelvin measurement is more accurate but unnecessary for game-state detection):

- Material: 0.5mm platinum wire (biocompatible, stable, doesn't corrode in LB, can be autoclaved)
- Placement: two Pt wires inserted through the spacer frame on opposing sides, penetrating 1–2mm into the agar layer. Space electrodes ~5mm apart (across the 7mm well diameter with ~1mm offset from center to avoid blocking the optical axis).
- Connection: short Teflon-insulated lead wires from spacer frame to PCB pads. The AD5933's VIN (excitation output) connects to electrode 1; the VOUT (current measurement input) connects to electrode 2 via RFB.

> **Important geometry note:** The EIS electrodes and the OPT101 optical axis must not conflict. The OPT101 sits directly below the coverslip. The EIS Pt wires enter from the sides of the spacer frame, parallel to the coverslip plane. At 30° downward angle, their tips sit in the agar without blocking the optical path. Verify clearance in the spacer frame design before printing.

**Calibration procedure at startup:**
1. Measure RFB resistance directly (known calibration resistor in place of well) — store as `Z_cal`
2. With well loaded, measure at 5 frequencies (e.g., 5, 10, 25, 50, 100kHz)
3. Store initial baseline |Z| values per frequency in EEPROM
4. Game logic uses *delta from baseline*, not absolute impedance values

**Software:** Use the ZSweeper or Electrochemistry Arduino library for AD5933. For v2 prototype, single-frequency measurement at 10kHz is sufficient. Full sweep (5 frequencies) gives richer data for debugging.

---

## LAYER 4: HARDWARE SYSTEM

### Component list (v2, updated)

**Microcontroller:**
- Arduino Nano (ATmega328P) — unchanged. Plugs into female headers on PCB.

**Sensors:**
- IrOx/Pt pH working electrode + Ag/AgCl reference (in well, through spacer frame) → analog A0 via MCP6001 buffer
- FSR force sensitive resistor (feed button, on case surface) → analog A1 via 10kΩ voltage divider
- OPT101 photodiode + transimpedance amplifier → analog A2 (reads mCherry fluorescence from below)
- AD5933 impedance analyzer (I2C, 0x0D) → reads EIS from Pt well electrodes

**Display:**
- SSD1306 128×64 OLED (I2C, 0x3C) → SDA/SCL (A4/A5)

**Excitation LED:**
- Amber LED 590nm on D6 (PWM) + 220Ω — excites mCherry only. No multiplexing required.

> **v1 → v2 hardware removals:** Blue 470nm LED (D3), Green 530nm LED (D5), and Velostat petting pad (A0 in v1) are all eliminated. This removes 3 components and simplifies the PCB significantly.

**Optical filter:**
- 600nm longpass filter (Rosco #19 "Fire" gel, or equivalent) — upgraded from 500nm longpass in v1. The tighter cutoff at 600nm more aggressively blocks the 590nm amber excitation scatter while still passing mCherry emission at 610nm. This improves SNR for mCherry specifically (critical since mCherry's Stokes shift is only 20nm).

**Translucent PDMS membrane:**
- Use standard Sylgard 184 cured at 10:1 base:crosslinker ratio — this gives optically clear, slightly translucent PDMS. The red mCherry emission at 610nm transmits through 300–500μm PDMS with minimal attenuation. Users looking at the top of the device can see the biological glow. This is intentional and is the primary user-facing aesthetic of v2.

**Optics stack (bottom to top):**
1. OPT101 sensor (face up, on PCB)
2. 600nm longpass filter (resting on sensor)
3. Glass coverslip #1.5 (170μm)
4. Spacer frame (7mm well, 2mm thick, black PLA or acrylic) — with IrOx/Pt pH electrode and two Pt EIS electrodes threaded through sidewalls
5. Agar layer (~0.5–0.8mm, bacteria embedded)
6. Liquid LB layer (~0.7–1.2mm)
7. Translucent PDMS membrane (300–500μm) — visible red glow surface
8. User's hand / ambient light above

**Power:** USB-powered via Nano's USB port. No battery in current design.

### Pin mapping (v2)

> **I2C bus note:** AD5933 (0x0D) and SSD1306 (0x3C) addresses do not conflict. Both operate at 3.3V VDD. Pull-up resistors: 4.7kΩ on SDA and SCL to 3.3V (not 5V — both devices are 3.3V I2C). The Nano's A4/A5 are 5V-tolerant with 3.3V pull-ups; this is fine.

### Physical layout (v2 updates)

Egg-shaped enclosure, black 3D-printed PLA, same form factor as v1 with the following changes:

- **Top surface:** Translucent PDMS window replaces the opaque membrane — this is now the primary visual feature. At high mCherry expression, a diffuse red glow is visible to the naked eye from several feet away in a dimmed room.
- **FSR feed button:** Remains on lower front face as a tactile soft pad.
- **Side port:** Feeding port (syringe injection for LB + kanamycin) unchanged.
- **Electrode ports:** Two additional small sealed ports in spacer frame for pH and EIS electrode wires. These are internal (not user-accessible); wires exit through the frame and terminate at PCB solder pads.
- **OLED position:** Unchanged — sits beside the well on the PCB, visible through cutout.
- **USB port:** Unchanged — bottom of device.

---

## LAYER 5: SIGNAL PROCESSING AND GAME LOGIC

### Sensing timeline and polling rates

| Signal | Hardware | Poll interval | Biological timescale |
|---|---|---|---|
| mCherry fluorescence | OPT101 + amber LED | Every 10 seconds | Hours (protein accumulation) |
| pH | IrOx electrode → A0 | Every 30 seconds | Hours (acid accumulation) |
| EIS impedance | AD5933 at 10kHz | Every 60 seconds | Hours (cell growth/death) |
| FSR (feed input) | A1 | Interrupt / 50ms | Instant |

### Startup calibration routine (mandatory)

Before entering game loop, the device runs a 3-step calibration:

1. **pH calibration:** Prompt user (OLED message) to place pH 7.0 buffer on well, read A0 for 10 seconds averaged → store as `pH7_ADC`. Then pH 4.0 buffer → store as `pH4_ADC`. Compute slope = (7.0 − 4.0) / (pH7_ADC − pH4_ADC) and intercept. Store in EEPROM. This is valid for the life of the IrOx electrode (weeks to months).

2. **EIS baseline:** With bacteria freshly loaded at OD600 ~0.25–0.5, run AD5933 sweep at 10kHz for 30 seconds (5 readings averaged) → store as `Z_baseline`. This baseline shifts when bacteria are reloaded — recalibrate at each reload.

3. **mCherry baseline:** Run 3 amber LED fluorescence readings averaged → store as `mCherry_baseline`. This accounts for any background signal from the agar matrix or optical path.

All subsequent readings use delta from baseline: `ΔpH`, `ΔZ`, `ΔmCherry`.

### Signal delta interpretation

Use exponential moving average (α = 0.1) on all three deltas to smooth noise before game-state evaluation.

### Game states (v2 — 5 states)

> **Disambiguation of THRIVING vs RECOVERING:** Both show pH dropping. The key difference: RECOVERING is entered explicitly after a feeding event (FSR press + confirmed LB injection timestamp). Use a `lastFedTime` variable — if a feeding occurred within the past 2 hours AND pH is dropping AND mCherry is falling, state = RECOVERING. THRIVING is the steady-state equivalent in the absence of a recent feeding event.

### State transition logic (pseudocode)

```cpp
// All deltas are EMA-smoothed before this evaluation
void evaluateGameState() {
  bool acidifying   = (deltaPH < -0.3);
  bool acidCrash    = (deltaPH < -0.8);
  bool impedOK      = (abs(deltaZ_pct) < 15);
  bool impedDrop    = (deltaZ_pct < -20);
  bool impedHigh    = (deltaZ_pct > +20);
  bool mCherryHigh  = (deltaMCherry_pct > 50);
  bool mCherryLow   = (deltaMCherry_pct < 20);
  bool recentFeed   = ((millis() - lastFedTime) < 7200000UL);  // 2hr window

  if (acidCrash && impedHigh && mCherryLow)  { state = OVERFED;     return; }
  if (impedDrop && mCherryHigh)              { state = SICK;        return; }
  if (acidifying && recentFeed && !mCherryHigh) { state = RECOVERING; return; }
  if (mCherryHigh && !impedDrop)             { state = HUNGRY;      return; }
  if (acidifying && impedOK && mCherryLow)   { state = THRIVING;    return; }
  // Default / ambiguous:                    { state = DORMANT;     }
}

1. All LEDs off → OPT101 reads ambient baseline → store as ambientRead
2. Amber LED (D6) ON at PWM 180/255 → wait 50ms → OPT101 reads for 100ms averaged
3. Amber LED OFF
4. mCherry_raw = reading − ambientRead
5. Apply EMA: mCherry_EMA = α * mCherry_raw + (1-α) * mCherry_EMA
6. Repeat every 10 seconds

</details>