BORIS

BORIS reconstructs transmission networks for infectious disease outbreaks by integrating genomic and epidemiological data in a Bayesian framework to infer who infected whom and estimate farm-level parameters such as transmissibility, susceptibility, incubation and latent periods for pathogens including foot-and-mouth disease.

Key Features:

• Bayesian Inference Framework: Extends Lau's systematic Bayesian inference framework by incorporating additional parameters for farm-level covariates such as predominant species and animal numbers.
• Markov Chain Monte Carlo (MCMC) Algorithm: Uses a reformulated Bayesian MCMC algorithm to perform statistical inference and was verified and pseudo-validated on 100 simulated outbreaks with demographic data from Japan and Australia.
• Improved Model Accuracy: Demonstrated significant improvements in overall accuracy compared to the SCOTTI model implemented in BEAST2 on simulated outbreak scenarios.
• Farm-Level Covariate Integration: Integrates farm-level covariates (e.g., species predominance, animal numbers) to estimate farm-specific transmissibility and susceptibility, including estimates that pig farms had five times the transmissibility and 49% lower susceptibility than cattle farms in the 2010 foot-and-mouth disease outbreak in Japan.
• Incubation and Latent Period Estimation: Estimates farm-level incubation periods and reports incubation periods shorter than latent periods.
• Outbreak Seeding and Cluster Linkages: Infers timing of outbreak seeding and identifies linkages between clusters and farms involved in dissemination.

Scientific Applications:

• Epidemiological Reconstruction: Reconstructs transmission trees for infectious disease outbreaks, applicable to pathogens such as foot-and-mouth disease.
• Parameter Estimation: Estimates farm-level transmissibility, susceptibility, incubation and latent periods using integrated genomic and epidemiological data.
• Outbreak Investigation and Source Tracing: Infers timing of seeding events and cluster-to-farm linkages to support tracing of origin and spread.
• Model Comparison and Validation: Enables comparative evaluation against approaches such as SCOTTI in BEAST2 and pseudo-validation using simulated outbreaks with real demographic data.

Methodology:

Integrates genomic data with epidemiological information within an extended Lau Bayesian framework and applies a reformulated Bayesian MCMC algorithm; validation included pseudo-validation on 100 simulated outbreaks using demographic data from Japan and Australia.