Fig. 3:

Rho avoidance drives purine bias in Bacilli genes. a, Nucleotide skew of 300-nt fragments of the B. subtilis (left) and E. coli (right) genomes derived from sense (blue) or antisense (green) strands of genes (Methods). Each point represents a single fragment. Contour lines mark 25^th, 50^th, and 75^th percentiles of Gaussian kernel density estimate. b, Phylogenetic tree of selected Bacilli species (155 species) colored by presence (purple) or absence (black) of a rho homolog in the genome⁴⁰ (Methods). c, Average excess purine content in leading strand genes of 309 Bacilli species, separated by the presence or absence of rho in the genome (Methods). Purine excess quantifies the difference between purine and pyrimidine counts, normalized to length. Median, 25^th, 75^th percentiles marked by horizontal lines. d, Beta-galactosidase (LacZ) assays for termination of brnQ homologs in wild-type and Δrho B. subtilis, as in Fig. 2c. Colored bars summarize result: grey, no termination; purple, Rho termination; pink, Rho-independent termination. Insert sequences will be available in Supplementary Table 2. Uncropped plate image in Extended Data Fig. 4b.