Extended Data Fig. 6: Characterization of mutational events in antibiotic resistant evolved lines.

(A) Distribution of SNPs and indels in resistant lines across four bacterial species. Red and green colours denote short indels and SNPs, respectively. As expected, lines derived from FoR have accumulated fewer mutations (one-sided Wilcoxon rank sum test, P < 2.2e-16), when compared to lines derived from ALE. (B) Fraction of non-synonymous mutations. To assess the signatures of adaptive evolution in our genomic samples from ALE and FoR assay, we tested whether the fraction of non-synonymous mutational events within all SNPs in the coding region was higher than expected based on a purely neutral model of evolution using an established method⁹⁸. For each bacterial strain background, we identified all SNPs in the coding regions, counted the number of non-synonymous (nonsyn) and synonymous (syn) ones and calculated the observed fraction of non-synonymous mutations (red dashed line) as follows: nonsyn / nonsyn + syn, where nonsyn and syn are the number of non-synonymous and synonymous mutations, respectively. Next, we randomly generated the same number of SNPs at random coding positions along the genome as observed in the mutation dataset. We repeated this step 5000 times, then plotted the fraction of non-synonymous mutations as histograms for each species (columns) and strain type (rows). Next, we calculated the probability (P-value) that the fraction of nonsynonymous mutations was equal to or higher in the real data than that of in the randomly generated one. (C) Distribution of different mutational events. Top and bottom row correspond to the adapted lines originating for laboratory evolution (ALE) and frequency of resistance (FoR) assays, respectively.

Source data