Statistical​ ​Methods​ ​for​ Challenging​ Problems​ ​in​ Public​ Health​ Microbiology

Collaborative Research Team Project: 2018-2021

Pathogenic microbial organisms cause a significant burden of disease not only in low-resource countries, but also in high-income countries, especially in hospital settings. One problem that is particularly relevant today is the problem of drug resistance, whereby a pathogen no longer responds to a drug treatment. Additionally, the appearance of pathogen outbreaks requires the development of surveillance tools to rapidly track, prevent, and ultimately disrupt the chain of transmissions.

The availability of fast, reliable and affordable whole-genome sequencing (WGS) methods has the potential to be a major boon for public health authorities attempting to control the development of drug resistance and the spread of epidemic outbreaks. However, in order to fully harness the power of these methods there is an urgent need for novel statistical and algorithmic techniques for microbial WGS data, but these methods are still in their infancy. This project consists of tackling three challenging and currently unsolved statistical problems that arise in public health microbiology, and deploying the developed methods in a publicly available computational platform.

The three aims are:

  • Developing a likelihood-based method for calling genomic variants (e.g. SNPs, insertions, deletions or alleles of specific genes) informed by a microbial evolutionary model
  • Training a statistical or machine learning algorithm to combine multiple signals into a single call to predict drug resistance, phylogenetic relatedness or epidemiological relatedness directly from whole-genome sequencing data
  • Designing methods for estimating the power of studies for detecting regular genotype-phenotype associations in bacteria, as well as epistatic interactions (which are known to require prohibitively high sample sizes in human genetics)

The team leaders are Alexandre Bouchard-Côté, Department of Statistics, University of British Columbia and Leonid Chindelevitch, School of Computing Science, Simon Fraser University. Collaborators include Luis Barreiro, Centre Hospitalier Universitaire Sainte-Justine, Montréal; Art Poon, Department of Pathology and Laboratory Medicine, University of Western Ontario; Jesse Shapiro, Department of Biological Sciences, Université de Montréal;
and Liangliang Wang, Department of Statistics and Actuarial Science, Simon Fraser University.

Comments are closed.