Fig. 6: Drug tolerance is target specific.
a, Principal component analysis of 8 drugs at 2 concentrations in drug tolerance space, based on Spearman correlations from 350 clinical M. abscessus isolates. b, Genome-wide association results for 16 drug tolerance phenotypes, highlighting the top 5 associated genes (moderate- or high-effect genotypes) per phenotype. The heat map shows association of 43 genes (top genotype per gene) with tolerance phenotypes using a mixed-effects model corrected for population structure (two-sided Wald test). Associations with P > 0.0001 are in white. c, Manhattan plot of 272,351 M. abscessus genotypes and their association with amikacin (low concentration) kill kinetics (two-sided Wald test), with the Bonferroni correction threshold (black line, 1.7 × 10−7). Multiple variants in MAB_0233 were strongly linked to amikacin killing (insert). d, Time–kill kinetics of control, ΔMAB_0233 and complemented strains (mean ± s.e.m., nine replicates per condition). AUC values of ΔMAB_0233 were compared with control or complemented strains using a two-sided Mann–Whitney U-test. P values for amikacin, linezolid and tigecycline were 4 × 10−5, except for tigecycline time–kill kinetics between ΔMAB_0233 and the control strain (P = 8 × 10−5). ****P < 0.0001.
