NGS 454   ERR315664 51 11 2 Corby already published/closest to centroid uncommon but well characterised – virulent in animal model and protozoa GenBank (NC_009494.2)   [34] 454 11 2 H091960011 second of cluster unique environmental strain (from Roman baths in Bath) NGS mate paired Illumina 268 ERR315665 Trichostatin A chemical structure 578 11 2 Alcoy already published responsible for a very big outbreak in Spain GenBank (NC_014125.1)   [35] 59 12 25 H070840415 closest to centroid quite common environmental strain – a few cases of LD. NGS mate paired Illumina 246 ERR315666 188 12 25 H075160080 second of cluster no particular data NGS paired end Illumina 298 ERR315667

36 13 9 Philadelphia already published/closest to this website centroid the type strain – well characterised caused the Philadelphia outbreak. GenBank (NC_002942.5)   [36] 37 13 9 H034680035 internationally significant in top six strains that cause disease NGS 454   ERR315668 186 13 9 H044500045 second of cluster unique clinical isolate NGS paired end Illumina 375 ERR315669 34 14 34 RR08000134 closest available

to centroid no particular significance NGS paired end Illumina 301 ERR315670 68 14 34 H074360710 second of cluster no particular significance NGS mate paired Illumina 179 ERR315671 707* 15 71 H091960009 only one in cluster unique environmental strain (from Roman baths in Bath) NGS paired end Illumina and mate paired Illumina Paired end 136 Mate paired 50 ERR315672 Asterisks designate strains that are likely to have plasmids based on the analysis described in the methods section of this manuscript. GBA3 The sequence data described can be obtained using the European Nucleotide Archive accession number listed in the table. De novo assembly The reads were assembled de novo into scaffolds. The genomic content of these scaffolds was assessed using BLAST Ring Image Generator [37] where the scaffolds were the query sequences and the reference sequence was the genome from the Corby strain (Figure  5). Corby was chosen since it is known to be virulent in both humans and

animal models and has extra mobile genetic elements not seen to date in the other sequenced legionella genomes [34, 38]. Regions showing a high level of variability compared to the Corby genome were investigated further by looking at the gene content of those regions (Additional file 1: Table S1). Figure 5 BRIG blast analysis of the Legionella genomes using the genome of Corby as a reference. The strains and figure colours used were from centre to outside ST152 (CST1 mauve), ST5 (CST1 light blue), ST611 (CST124 dark blue), ST454(CST2 medium blue), ST47(CST16 leaf green), ST376 (CST17 dark green), ST46(CST45 light green), ST59 (CST25 pink), ST42(CST14 red), ST84 (CST15 purple), ST337 (CST130 mauve), ST23 (CST19 light blue), ST37 (CST9 dark blue), ST68 (CST34 medium blue), ST154 (CST12 leaf green) and ST707 (CST71 dark green).