Clustering minimal inhibitory concentration data through Bayesian mixture models: An application to detect...

Area of Science:

• Genomics
• Microbiology
• Public Health

Background:

• Antimicrobial resistance (AMR) poses a significant global health threat.
• Whole-genome sequencing (WGS) is increasingly affordable and rapid, enabling genomic variant analysis.
• Genomic mutations can be statistically linked to observed antimicrobial resistance patterns.

Purpose of the Study:

• To develop a robust method for identifying genomic mutations associated with antimicrobial resistance.
• To create and update catalogues of genomic variants linked to antibiotic resistance.
• To improve the prediction of antimicrobial resistance and inform patient-specific treatment strategies.

Main Methods:

• Analysis of minimal inhibitory concentration (MIC) distributions.
• Clustering of bacterial strains based on resistance levels.
• Regression analysis to identify statistically significant mutations associated with resistance.
• Application to a 96-well microtiter plate for Mycobacterium tuberculosis testing.

Main Results:

• A general method was proposed to identify clusters of strains with varying antimicrobial resistance.
• High statistical power was achieved in identifying resistance-associated mutations using regression.
• The method was successfully applied to Mycobacterium tuberculosis.

Conclusions:

• The proposed method enables robust identification of genomic mutations driving antimicrobial resistance.
• This facilitates the development of predictive models for resistance and personalized treatment plans.
• Continuous updating of genomic variant catalogues is crucial for combating AMR.