Phylogenetic analysis of sasO and sasN across bacterial phyla

Phylogenetic trees were derived from the Genome Taxonomy Database (GTDB, release 220)⁸⁵. The fully annotated precomputed tree and the extracted actinomycete phylogeny were plotted in R with the ggtree package (Extended Data Fig. 4a,b) ⁸⁶. Hidden Markov Models (HMMs) for SasO and RsmC were generated from orthologs of WAC40 SasO and the methyltransferase domain of Thermus thermophilus RsmC (UniProtKB Q5SKW0, amino acids 185-375), respectively. The corresponding NCBI reference sequences are: SasO (Bifidobacterium longum WP_012577213.1, Bacillus cereus WP_410259422.1, Clostridium sp. NLP14976.1, Staphylococcus sp. WP_113608439.1, Chlamydia trachomatis CRH61495.1, Bacillus sp. WP_000763262.1) and RsmC (E. coli CQR83742.1, Klebsiella pneumoniae EJK92454.1, Vibrio cholerae WP_142735227.1, Caulobacter sp. HRD46923.1, Pseudomonas aeruginosa MCR3844603.1, Thermus thermophilus WP_244344254.1, Haemophilus influenzae SPX43210.1). HMMs were queried against all of the genomes from the GTBD reference and the distributions of error values were used to define thresholds (Supplementary Fig. 15) to assign presence of rsmC or sasO genes in the genome. When a protein gene was predicted by either model, it was classified by the lowest error value. Analysis was performed using custom scripts.