PyBDEI

PyBDEI estimates epidemiological parameters from phylogenetic trees using the Birth-Death Exposed-Infectious (BDEI) model to quantify transmission dynamics such as R_e and infectious period while accounting for incubation delays in pathogens like Ebola and SARS-CoV-2.

Key Features:

• BDEI model: Implements the Birth-Death Exposed-Infectious model that accounts for the delay between infection and becoming infectious (incubation period).
• Phylogenetic inference: Uses phylogenetic trees (genealogies of pathogen sequences) to infer epidemiological parameters including R_e and infectious time.
• Multi-type birth-death framework: Operates within a multi-type birth-death (MTBD) framework to represent structured transmission dynamics.
• Parallelizable ODE formulation: Employs a novel formulation of ordinary differential equations (ODEs) that enables parallel computation and preserves numerical stability.
• Scalability and speed: Processes large phylogenetic trees (reported up to 10,000 samples) with execution times on the order of minutes (approximately two minutes for 10,000 samples in reported benchmarks).
• Accuracy and uncertainty: Produces parameter estimates together with confidence intervals to quantify estimation uncertainty.
• Forests of trees: Extends to forests of phylogenetic trees to model multiple introductions or sub-epidemics.
• Comparative performance: Empirical comparisons on simulated data report improved speed and accuracy relative to existing implementations and include application to the 2014 Ebola epidemic in Sierra Leone.
• Extensibility: Methodological approaches are stated to be extendable to ClaSSE-like models for macroevolutionary applications.

Scientific Applications:

• Estimation of transmission parameters: Quantifies R_e and infectious period for pathogens with incubation periods, such as Ebola and SARS-CoV-2.
• Phylodynamic analysis: Infers epidemic dynamics from pathogen sequence genealogies under the BDEI and MTBD frameworks.
• Modeling complex epidemics: Represents multiple introductions and sub-epidemics via forests of phylogenetic trees.
• Real-world outbreak analysis: Applied to real outbreak data, including analysis of the 2014 Ebola epidemic in Sierra Leone.
• Macro-evolutionary extension: Supports methodological extension to ClaSSE-like models beyond epidemiology.

Methodology:

Inference is performed from phylogenetic trees using the BDEI model and a novel ODE formulation that enables parallel computation and numerical stability; the implementation estimates epidemiological parameters and their confidence intervals and can operate on forests of phylogenetic trees.