Accounting for contact tracing in epidemiological birth-death models

Area of Science:

• Epidemiology
• Computational Biology
• Genomics

Background:

• Phylodynamics estimates epidemiological parameters from pathogen phylogenetic trees.
• Standard models assume independent sampling, which is often violated in real-world epidemics, particularly for sexually transmitted infections like HIV-1 where contact tracing is common.

Purpose of the Study:

• To extend phylodynamic multi-type birth-death (MTBD) models to incorporate contact tracing (CT).
• To develop tools for simulating trees under these new models, testing for CT, and estimating model parameters.

Main Methods:

• Developed a simulator for MTBD and MTBD-CT models.
• Proposed a non-parametric test for detecting CT in phylogenetic trees.
• Derived a closed-form likelihood for the BD-CT(1) model and implemented a maximum-likelihood estimator.

Main Results:

• The CT detection test demonstrated high specificity and sensitivity on simulated data.
• Accurate parameter inference was achieved using the BD-CT(1) model on simulated and real HIV-1 data.
• Ignoring CT in phylodynamic models leads to biased parameter estimates, such as overestimating the becoming-non-infectious rate.

Conclusions:

• The developed MTBD-CT models and associated tools offer a more accurate approach to phylodynamic analysis when contact tracing is employed.
• Accounting for contact tracing is crucial for unbiased estimation of epidemiological parameters, particularly for diseases like HIV-1.
• Freely available software facilitates the application of these advanced phylodynamic methods.