Distinguishing new from persistent infections at the strain level using longitudinal genotyping data

Area of Science:

• Epidemiology
• Genetics
• Biostatistics

Background:

• Longitudinal pathogen genotyping is crucial for understanding infection dynamics and disease progression, particularly in malaria.
• Distinguishing new infections from persistent strains is vital but challenging due to multiple strains per sample, shared alleles, and genotyping errors.
• Current methods for strain differentiation rely on simple rules based on allele observation time, limiting accuracy.

Purpose of the Study:

• To develop and validate a novel statistical framework, DINEMITES (Distinguishing New Malaria Infections in Time Series), for accurate analysis of longitudinal pathogen genotyping data.
• To estimate the probability of new infections, molecular force of infection (molFOI), and new infection events from time-series sequencing data.
• To improve the detection of malaria infections by addressing challenges like missing data, multiple strains, and allele sharing.

Main Methods:

• Developed DINEMITES, a Bayesian statistical model for analyzing longitudinal genotyping data from individual hosts.
• Incorporated capabilities to handle missing sequencing data, treatment history, and relevant covariates.
• Evaluated model performance using synthetic data and applied it to three real-world longitudinal malaria genotyping datasets.

Main Results:

• The DINEMITES Bayesian model accurately estimated key clinical parameters like molFOI, outperforming a clustering-based alternative and simple rule-based methods.
• Compared to simple rules, DINEMITES detected a substantially higher average number of infections per participant across three real datasets (33% to 359% increase).
• The method demonstrated robustness in handling complex genotyping data, including multiple strains, shared alleles, and stochastic marker dropout.

Conclusions:

• DINEMITES provides a statistically rigorous and scalable approach for analyzing longitudinal pathogen genotyping data to understand infection dynamics.
• The method significantly enhances the ability to detect new infections and estimate the force of infection, offering valuable insights for malaria control and research.
• This advancement has broad implications for infectious disease epidemiology, enabling more precise characterization of transmission patterns and intervention effectiveness.