Bayesian inference of antibody evolutionary dynamics

A collection of analyses for the paper, Bayesian inference of antibody evolutionary dynamics using multitype branching processes.

There are also accompanying software packages which implement branching process simulators and likelihoods:

• thanasibakis/gcdyn.jl
• matsengrp/gcdyn

Getting started

0. Please ensure you have a working Python and Julia installation. (This codebase was tested with Python 3.12 and Julia 1.10.)

1. Set up a Python environment for the gcdyn dependency.

python -m venv .venv/
echo "*" > .venv/.gitignore
source .venv/bin/activate

pip install -e lib/gcdyn

2. Set up the Julia environment.

julia  --project -e "import Pkg; Pkg.instantiate()"

Repository structure

Visit the following directories for specific instructions on how to run each analysis:

data-analysis/

• Analysis of our real experimental dataset
• See Section Analysis of experimental data in the paper

simulation-studies/

• A collection of simulation studies to test our method under various model misspecifications
• See Section Simulation studies in the paper

type-spaces/

• Code to compute our type space and type change rate matrix
• See Section Adapting our methodology to the experimental setup of DeWitt, Vora, Taraki et al. (2025) in the paper