Title: Signatures of selection at drug resistance loci in Mycobacterium tuberculosis

Tuberculosis (TB) is the leading cause of death by an infectious disease, and global TB control efforts are increasingly threatened by drug resistance in Mycobacterium tuberculosis (M. tb). Unlike most bacteria, where lateral gene transfer is an important mechanism of resistance acquisition, resistant M. tb arises solely by de novo chromosomal mutation. Using whole genome sequencing data from two natural populations of M. tb, we characterized the population genetics of known drug resistance loci using measures of diversity, population differentiation, and convergent evolution. We found resistant sub-populations to be less diverse than susceptible sub-populations, consistent with ongoing transmission of resistant M. tb. A subset of resistance genes (“sloppy targets”) were characterized by high diversity and multiple rare variants; we posit that a large genetic target for resistance and relaxation of purifying selection contribute to high diversity at these loci. For “tight targets” of selection, the path to resistance appeared narrower, evidenced by single favored mutations that arose numerous times on the phylogeny and segregated at markedly different frequencies in resistant and susceptible subpopulations. These results suggest that diverse genetic architectures underlie drug resistance in M. tb, and combined approaches are needed to identify causal mutations. Extrapolating from patterns observed in well-characterized genes, we identified novel candidate variants involved in resistance. The approach outlined here can be extended to identify resistance variants for new drugs, to investigate the genetic architecture of resistance, and, when phenotypic data are available, to find candidate genetic loci underlying other positively selected traits in clonal bacteria.