Extended Data Fig. 2

(a) Rarefaction analysis shows sequencing depth was sufficient to comprehensively catalogue the unique microbial strains present. (b) Composition of each Australian patient sample classified at the ‘family’ taxonomic level (n = 71). (c-e) Inverse Simpson’s index of alpha diversity for individual patients grouped by (c) maximum irAE grade experienced by each patient split by severe irAEs, (d) response/low irAE (R/G0-G2) (pink) and non-response/severe irAE (NR) (NR/G3- G5)) (orange) or (e) maximum irAE grade experienced by each patient split by severe gastrointestinal irAEs and non-gastrointestinal irAEs (NR with severe irAE ‘Adverse’ indicated in orange). (AUS, n = 7) (f & g) Tumour mutational burden (TMB) and tumour IFN-gamma signature split by response. Subset of patients Rozeman etal (2021)¹⁷ (n = 25). (h) Inverse Simpson’s index of alpha diversity for individual patients with high TMB grouped by response and non-response. (i-j) Absolute bacterial/archaeal faecal loads assessed using qPCR, with patients grouped by irAEs and ‘Adverse’ outcome groups (n = 71). (k) Correlation of diversity with 16 S rRNA gene number/mg faeces for each patient (n = 71). Each symbol represents an individual patient, bars indicate the median. Mann-Whitney U rank sum test (c-j). For linear regressions, p value was calculated on Spearman’s rank correlation (k). All statistical tests are two-sided where appropriate.