Skip to contents

MalReBay

Overview

MalReBay is an R package for Bayesian molecular correction of malaria therapeutic efficacy studies (TES). Given paired genotyping data from a patient’s initial infection (Day 0) and a later recurrence, it estimates the posterior probability that the recurrence is a recrudescence (treatment failure, same parasite persisting) versus a reinfection (new parasite acquired after treatment).

Unlike rule-based allele-matching methods, MalReBay:

  • models sequencing error, allelic dropout, and allele loss explicitly
  • handles polyclonal infections (MOI > 1) at every locus
  • incorporates local allele frequency information to weigh evidence
  • uses a full Bayesian MCMC engine (Stan HMC-NUTS) for principled uncertainty quantification
  • supports both length-polymorphic markers (microsatellites, MSP, GLURP) and amplicon sequencing (haplotype) data

Key Features

Feature Detail
Data types Length-polymorphic markers and amplicon sequencing
MCMC engine Stan HMC-NUTS via rstan — fast, well-mixing, gradient-based
Polyclonal infections Analytically marginalises over within-host clone multiplicity
Error model Estimates q_mismatch, q_loss, q_dropout from the data
Prior Recrudescence probability fixed at 0.5 (no directional bias)
Convergence Rank-normalised R̂, bulk/tail ESS, Geweke, Gelman-Rubin
Output Per-patient posterior probability + match-counting comparison table

Installation 

# Install the development version from GitHub
remotes::install_github("SwissTPH/MalReBay")

Note: MalReBay links against rstan and StanHeaders. On first use the Stan model is compiled once and cached. This may take a few minutes but only happens once per R installation.

Quick Start 

library(MalReBay)

# Run the full pipeline on the bundled Angola 2021 TES example data
results <- MalReBay(output_folder = "my_results")

# View posterior probabilities
head(results$posterior_probabilities)

Workflow

MalReBay wraps a four-step pipeline that can also be called step by step:

import_data()
    ↓
classify_infections()
    ↓
summarise_results()
    ↓
save_results()