🦠 Bacterial Growth Simulation with Dosing Strategies 🦠

📋 Overview

This project sets up a compartmental ODE model. It aims to simulate the way bacterial populations shift and grow when facing antibiotic pressure. The model builds in suppression effects that depend on the dose given. It also looks into various dosing approaches by running four organized simulations.

Key Features:

• ODE-based bacterial growth model with antibiotic pharmacodynamics (PK/PD)
• Configurable dosing strategies (constant and pulsed)
• Parameter fitting via least-squares optimization
• Publication-quality visualizations
• Modular, documented R code architecture

🛠️ Project Structure

.
├── src/                      # R source code
│   ├── main.R               # Pipeline orchestration
│   ├── model.R              # ODE definition
│   ├── simulate.R           # Data generation
│   ├── fit.R                # Parameter fitting
│   ├── visualize.R          # Plotting
│   └── io.R                 # Config & I/O
├── config/
│   └── sim_params.yaml      # Parameters
├── figures/                 # Output plots (8 files)
├── results/                 # Output metrics (4 CSV files)
├── data/                    # User data (optional)
└── README.md

---

📋 The Model

ODE Equation:

dN/dt = r·N·(1 - N/K) - k_max · [A(t)^h / (EC50^h + A(t)^h)] · N

Parameters:

• N(t): Bacterial density (cells)
• r: Growth rate (hr⁻¹)
• K: Carrying capacity (cells)
• A(t): Antibiotic concentration (μg/mL, time-dependent)
• k_max: Maximum kill rate (hr⁻¹)
• EC50: Half-maximal effect concentration (μg/mL)
• h: Hill coefficient (sigmoidal slope)

Dosing Modes:

• Constant: A(t) = A₀ throughout
• Pulse: Square pulses every N hours for M-hour duration

---

Results: 4-Run Comparison

Run	Configuration	EC50 (μg/mL)	Hill (h)	Best Conc	AUC	Key Finding
1	Constant, fit EC50+h	2.43	1.17	20	1.06e+06	Baseline reference
2	Pulse (12h), fit EC50+h	2.43	1.17	20	1.06e+06	Clinical relevance
3	Constant, fit EC50 only	2.00	1.50*	20	2.31e+10	Parameter sensitivity
4	Extended [0-50], fit EC50+h	2.00	1.50	50	9.23e+05	Saturation effects

*Hill coefficient fixed at 1.50

---

🏃‍♂️ Run Descriptions

Run 1: Baseline Analysis

Configuration: Constant dosing [0–20 μg/mL], both parameters fitted

Finding: EC50 = 2.43 μg/mL; Hill = 1.17
Establishes standard dose-response curve. Lower Hill suggests weak cooperativity.

Interpretation: Antibiotic effect is relatively linear rather than sigmoidal.

---

Run 2: Pulse Dosing Strategy

Configuration: Pulsed dosing (12h interval, 2h duration), both parameters fitted

Finding: EC50 = 2.43 μg/mL; Hill = 1.17 (identical to Run 1)
AUC remains consistent despite intermittent dosing.

Interpretation: Clinical relevance—simulates realistic antibiotic therapy. Bacterial rebound between doses visible in time-courses, explaining resistance development risk.

---

Run 3: Parameter Sensitivity

Configuration: Constant dosing, EC50 fitted only (Hill fixed at 1.50)

Finding: EC50 = 2.00 μg/mL; AUC = 2.31e+10 (much higher)
EC50 shifts when Hill is constrained.

Interpretation: Demonstrates parameter interdependence. When prior knowledge constrains Hill, EC50 adjusts to compensate. Shows importance of data quality for reliable parameter estimation.

---

Run 4: Extended Dose Range & Saturation

Configuration: Extended range [0–50 μg/mL], both parameters fitted

Finding: EC50 = 2.00 μg/mL; Best concentration shifts to 50 μg/mL; AUC = 9.23e+05
Model saturates at high doses—diminishing returns beyond 20 μg/mL.

Interpretation: Identifies safe and effective dosing windows. High doses don't provide additional benefit, suggesting toxicity thresholds should be considered in clinical practice.

---

📊 Understanding Outputs

Time-Course Plots

• X-axis: Time (hours)
• Y-axis: Bacterial density (CFU/mL, log scale)
• Panels: One per antibiotic concentration
• Pattern: Higher concentrations show stronger suppression

Dose-Response Curves

• X-axis: Antibiotic concentration (μg/mL)
• Y-axis: AUC (area under curve — cumulative bacterial exposure )
• Red line: EC50 mark (half-maximal effect)
• Sigmoid curve: S-shaped dose-response

Metrics CSV

Column	Meaning
AUC	Total bacterial burden (lower = better)
max_density	Peak population size
time_to_50pct_reduction	Hours to 50% suppression
estimated_EC50	Fitted potency
estimated_h	Fitted cooperativity

👨‍💻 Quick Start

Installation

# Clone repository
git clone https://github.com/jobinjohn24/bacterial-growth-simulation.git
cd bacterial-growth-simulation

# Install dependencies in R
install.packages(c("yaml", "deSolve", "ggplot2"))

Run Simulation

Rscript src/main.R

Output: Plots in figures/ and metrics in results/

Usage

Change Antibiotic Concentrations

Edit config/sim_params.yaml:

A_levels: [0, 0.5, 1.5, 3, 5]  # Custom range

Switch Dosing Schedule

dose_schedule: "pulse"
pulse_interval: 8      # Every 8 hours
pulse_duration: 1      # For 1 hour

Use Your Own Data

Place CSV in data/growth.csv with columns:

time, concentration, density

👨‍🔬 Customization

Fit Different Parameters

fit_free: ["EC50"]              # Fit only EC50
fit_free: ["r", "K"]            # Fit growth parameters
fit_free: ["EC50", "h"]         # Fit potency & cooperativity

Extend Simulation Time

tmax: 72                          # 72 hours instead of 48

Adjust Noise Level

noise_sd: 1.0e7                   # Increase measurement noise