Extended Data Fig. 5: Somatic Mutations Modulate the m⁶A Landscape.

a-b) Association of (a) copy number aberrations or (b) RNA abundance with 1) RNA or protein abundance of the corresponding gene, or total sample peak number (Student’s t-test, two-sided), 2) clinical features: ISUP Grade Group, IDC status, pathological T category (Pearson’s Χ² test, two-sided), PSA, and age (Student’s t-test, two-sided). Black background indicates Q value < 0.1, whilst dot size represents (a) effect size or (b) Spearman’s correlation. c) Association of mutations in m⁶A regulatory genes and prostate cancer driver events. P values from a hypergeometric test (one-sided for over-representation) are represented by the color of the background, whilst the size of the dot represents the difference between observed and expected co-occurring mutations. Covariate bars indicate mutation type and, if applicable, GISTIC peak the mutation is located within. d) Consensus clustering of 3,432 differentially methylated m⁶A peaks (k = 5) associated with driver mutations or mutations to m⁶A regulators (k = 5). For each mutation, colors in the main heatmap indicate the log₂ fold change in m⁶A level, and right barplot the number of m⁶A peaks significantly differentially m⁶A methylated, between samples with and without the corresponding mutation. MC: mutation cluster; DMPC: differentially-methylated peak cluster. e) Pathway enrichment analysis performed on genes with differential m⁶A abundance associated with mutations in m⁶A enzymes or prostate cancer drivers. Each node represents an enriched pathway term (from GO:BP or REACTOME ontologies). Node colors represent mutations from (d), with some pathways only enriched through joint consideration of all genes associated with a whole mutation cluster (indicated through ‘Contribution’ key). f) Forest plot showing the hazard ratio and 95% confidence intervals for presence of an m⁶A peak associated with biochemical recurrence, and left barplot indicates corresponding P values from a Cox Proportional Hazards model. g-h) Presence of a m⁶A peak on g) INHBA or h) ZFHX4, increases risk of biochemical recurrence. INHBA peak samples n = 6, ZFHX4 peak samples n = 9, total samples = 148. P values from a Cox Proportional Hazards model.