Abstract
Mycobacterium tuberculosis can adapt to changing environments by non-heritable mechanisms. Frame-shifting insertions and deletions (indels) may also participate in adaptation through gene disruption, which could be reversed by secondary introduction of a frame-restoring indel. We present ScarTrek, a program that scans genomic data for indels, including those that together disrupt and restore a gene's reading frame, producing "frame-shift scars" suggestive of reversible gene inactivation. We use ScarTrek to analyze 5977 clinical M. tuberculosis isolates. We show that indel frequency inversely correlates with genomic linguistic complexity and varies with gene-position and gene-essentiality. Using ScarTrek, we detect 74 unique frame-shift scars in 48 genes, with a 3.74% population-level incidence of unique scar events. We find multiple scars in the ESX-1 gene cluster. Six scars show evidence of convergent evolution while the rest shared a common ancestor. Our results suggest that sequential indels are a mechanism for reversible gene silencing and adaptation in M. tuberculosis.