Mix infection analysis through machine learning

Mixed infections in genotypic drug-resistant Mycobacterium tuberculosis

Publication

Developed a custom Python pipeline using Gaussian Mixture Models to detect mixed M. tuberculosis infections from WGS data.

Analyzed over 5,000 public isolates to model expected allele frequency distributions and benchmark mixed infection detection thresholds.

Demonstrated that mixed infections are significantly associated with genotypic drug resistance, suggesting transmission complexity and diagnostic challenges.

Proposed integration of this method into existing TBProfiler workflows for improved strain-level resistance resolution.


TOAST Image 1

Mix infection over the world

TOAST Image 2

Mixed infection dissection pipeline