A Supervised Statistical Learning Approach for Accurate Legionella pneumophila Source Attribution during Outbreaks

Area of Science:

• Genomics and Bioinformatics
• Infectious Disease Epidemiology
• Statistical Learning in Public Health

Background:

• Public health relies on genomics for Legionnaires' disease investigations.
• Legionella pneumophila exhibits extreme genome conservation, complicating outbreak source identification.
• Outbreaks can involve multiple L. pneumophila genotypes from a single source, further challenging standard phylogenomic methods.

Purpose of the Study:

• To develop and validate a statistical learning approach for accurate Legionnaires' disease outbreak cluster identification and source prediction.
• To overcome limitations of phylogenomic methods in identifying L. pneumophila outbreak sources due to its unusual population structure.

Main Methods:

• Applied a statistical learning approach using core genome single nucleotide polymorphism (SNP) comparisons of L. pneumophila.
• Utilized discriminant analysis of principal components (DAPC) to build a multivariate model predicting exposure sources from cooling tower isolates.
• Validated the model using 234 L. pneumophila isolates from Melbourne, Australia (1994-2014), and a UK hospital outbreak.

Main Results:

• The statistical learning model achieved 93% congruence with epidemiological data in Australian outbreak investigations, including a large aquarium outbreak.
• The approach demonstrated 86% predictive ability in a UK hospital Legionnaires' disease investigation.
• The method effectively identified cooling tower-specific genomic signatures for accurate source attribution.

Conclusions:

• A statistical learning method based on L. pneumophila core genome SNPs provides objective source attribution for Legionnaires' disease outbreaks.
• This approach effectively addresses the challenges posed by L. pneumophila's genetic diversity, improving outbreak investigation accuracy.
• The developed method offers a promising advancement for microbial outbreak investigations, enhancing public health response.