UVC Power Analysis

SecureBio

Spring 2023

Introduction

This repository contains code to simulate viral outbreaks in isolated environments (oil rigs and cruise ships). It was developed for a power analysis of a proposed experiment to evaluate the efficacy of 222 nm far-UVC in reducing infection rates on oil platforms. It contains:

• Code to simulate viral spread among workers on an oil platform (src/rig.py)
• Code to simulate viral spread among passengers on a cruise ship (src/cruise.py)
• A snakemake pipeline (Snakefile) to demonstrate use of the code and reproduce the power analysis figures in the proposal.

Oil platform model

The oil platform simulations are individual-based have three component models:

1. Crew turnover: Individual workers cycle between shifts onboard and shifts on the mainland, keeping the number of workers onboard approximately constant. Shifts are staggered so that only a fraction of the crew changes at a time.
2. Viral dynamics: We track workers as they move between four viral states: Susceptible, Exposed, Infectious, and Recovered. Susceptible workers on the mainland become Exposed at specified, time-dependent rate. Susceptible workers on the platform become Exposed at a rate proportional to the number of Infectious workers on the platform. Exposed workers become Infectious a fixed time after their first exposure, and Infectious workers become Recovered after another fixed interval. Recovered workers do not become Susceptible again. We simulate multiple viruses, each characterized by its infection rate on the mainland, basic reproductive number on the rig, and times to infectiousness and recovery.
3. Infection detection: Workers onboard the platform are tested at fixed intervals. Workers test positive if and only if they are Infectious. Each simulation is summarized by the number of total new infections across all viruses detected on the platform over the course of the simulation.

Installation

The simulation code requires Python >= 3.9. We recommend using pyenv to manage Python versions.

First, clone the repository:

git clone https://github.com/securebio/uvc-power.git && cd uvc-power

Next, set up a virtual environment (optional, but recommended) and install the requirements:

python -m venv env &&
source env/bin/activate &&
python -m pip install -r requirements.txt

Now the simulations are ready to run.

Usage

To run all the simulations and make the figures:

snakemake --cores

If the simulation output already exists and you'd like to force snakemake to rerun the pipeline, use the -F flag.

Warning: The simulations are not optimized for speed and may take a long time to run (~1hr using a 10-core machine). To test the simulations without waiting as long, you can set the number of replicate simulations per condition (default=4000) to a smaller number (200 is recommended):

snakemake --cores --config n_sims=200