EpiFusion

REF	Judge et al. (2024)
Docs
Github	github.com/ciarajudge/EpiFusion
Last commit	Nov, 2024
Installation

Brief description

Brief summary of EpiFusion method from the paper

EpiFusion is a Bayesian framework designed to estimate the effective reproduction number by jointly analyzing epidemiological (case incidence) and phylodynamic (genomic) data using particle filtering within a particle Markov Chain Monte Carlo (pMCMC) framework. It addresses the limitations of using only epidemiological or genomic data, particularly in under-sampled outbreaks. EpiFusion combines a stochastic infection dynamics model with dual observation models: one for case incidence data and another for phylodynamic tree data. The approach involves sequential particle filtering to simulate infection trajectories, with particles weighted and resampled based on their fit to both data sources. Parameter inference is achieved through Metropolis-Hastings MCMC. EpiFusion has been validated through simulations, benchmarking against existing tools, and application to real-world outbreaks, including the 2014 Ebola outbreak in Sierra Leone.

Methods

This package contains methods that estimate \(R(t)\) from both phylodynamic (time-scaled trees estimated from genetic sequences) and epidemiological (case incidence) data. Therefore, a discussion of these methods is somewhat outside the scope of this document.

Assessment

Features
Ability to nowcast/forecast	No, Designed for retrospective analysis
Incorporates delay distributions	Yes, Handles delays between infection and reporting implicitly
Estimates expected cases	Yes
Communicates uncertainty	Yes, Highest Posterior Density (HPD) intervals
Validation
Documentation of package methods	Yes
Documentation of package implementation	No

Sample code

Tutorials for how to use EpiFusion are given in this Github repository.

Judge C, Vaughan T, Russell T, Abbott S, Plessis L, Stadler T, et al. 2024. EpiFusion: Joint inference of the effective reproduction number by integrating phylodynamic and epidemiological modelling with particle filtering. PLOS Computational Biology 20:e1012528; doi:10.1371/journal.pcbi.1012528.