Figure 2. Results of Cartesian Pooling-Coordinate Sequencing on a 96 × 96-well transposon insertion library of M. bovis BCG.

(a) Visualization of the M. bovis BCG Pasteur reference genome. GC% (black) and Tn insertions that were identified and coordinate-determined in our 96 × 96-well mutant library (green bars). Duplicated regions of the genome are marked in red (DU1 and DU2). (b) Distribution between disrupted ORFs for which the location in the library of at least one disruption mutant is known (purple), disrupted ORFs for which the locations of their disruption mutants are unknown (‘untraceable'; grey) and untargeted ORFs (brown). (c) Relation between the fraction and the number of TA's in the ORF. Genes for which the orthologues are known to be essential in M. tb are omitted from the analysis. (d) Gene distribution according to gene-essentiality of M. tb orthologues of both the mutated and unmutated/untraceable gene fraction in our M. bovis BCG Tn insertion library. Genes for which no reliable data are available with regard to their essentiality for in vitro growth are shown in grey. (e) Theoretical model (based on the ‘coupon collector's' problem) to estimate the library size needed to target a certain percentage of the non-essential genome with at least one disruption every 12 TA's (=median number of TA's per gene in the M. bovis BCG Pasteur genome). The model predicts that with 6,072 mutants, 71% of the non-essential genes should be hit at least once, which is slightly above what we observed in our data set (64%). Doubling the amount of mutants will allow disrupting almost all targetable genes in this genome.