A framework for detecting signatures of positive selection in microbial genomes
Shapiro BJ, David LA, Friedman J, Alm EJ. (2009) Trends in Microbiology.
Shapiro BJ, Alm EJ. (2009) The ISME Journal.
Shapiro BJ, Alm EJ. (2008) PLOS Genetics.
Most of the population genetics toolkit was developed with sexual eukaryotic populations in mind, and these papers provides a detailed discussion of which tools are theoretically justified for use with microbes, and proposes new tests for selection at the protein level. Only certain classes of population genetic tests for positive selection are amenable to both clonal and recombining populations. One of these tests, evolutionary convergence, relies on the independent fixation of the same mutation (or allele) in different lineages under similar selective pressures. With collaborators Megan Murray and Maha Farhat at Harvard School of Public Health, we are currently developing a convergence test to discover antibiotic resistance mutations in tuberculosis.
Mycobacterium tuberculosis is successfully evolving antibiotic resistance, threatening attempts at tuberculosis epidemic control. Mechanisms of resistance, including the genetic changes favored by selection in resistant strains, remain poorly understood. In this large collaborative effort, involving M. tuberculosis strains sampled worldwide and sequenced at the Broad Institute, we identified regions of the genome under positive selection, specifically in drug resistant genomes. We developed a convergence test to identify mutations that are repeatedly associated with antibiotic drug resistance in different clonal lineages of M. tuberculosis. Convergence is simply the repeated appearance of the same mutation (or mutations in the same gene) in multiple different independent lineages. Similar selective pressures (antibiotic treatment, for example) in different lineages of M. tuberculosis are expected to lead to convergent mutations conferring resistance, or boosting fitness in the presence of drugs. Of course, convergence will sometimes occur by chance, due to random fixation of non-selected mutations. Our convergence test controls for such chance fixations, and identifies statistically significant convergence events, likely due to selection on drug resistant strains.