Fig. 3: Comparison of m⁶Am methylome detected by m⁶Am-seq and existing m⁶Am detection methods.

a Boxplot showing that the normalized peak coverage by m⁶Am-seq (red, n = 1652), m⁶Am-Exo-seq (blue, n = 3169), and miCLIP (yellow, n = 2217). Statistical significance of the difference was determined by unpaired two-sided Mann–Whitney U-test. ****P < 2.2e⁻¹⁶ for m⁶Am-seq vs m⁶Am-Exo-seq, m⁶Am-seq vs miCLIP. Boxes represent 25th–75th percentile (line at the median) with whiskers at 1.5*IQR. b Metagene profiles of m⁶Am distribution identified by the three sequencing methods. Each segment was normalized according to its average length in RefSeq annotation. c Venn diagram showing the overlap of m⁶Am-marked genes by m⁶Am-seq and by miCLIP. d Boxplot showing the normalized peak coverage of shared m⁶Am peaks (yellow, n = 456, identified in both m⁶Am-seq and miCLIP) and m⁶Am-seq-unique m⁶Am peaks (pink, n = 1179) in the m⁶Am-seq and miCLIP data sets, respectively. Boxes represent 25th–75th percentile (line at the median) with whiskers at 1.5*IQR. e An example of m⁶Am (in the RBM48 gene), which was missed by miCLIP due to its limited sensitivity but is clearly identified by m⁶Am-seq. The m⁶Am peak is lost in PCIF1-KO data sets, further proving the detection confidence of m⁶Am-seq. Pink background color denotes m⁶Am signal. f An example of mis-annotated m⁶Am (in the ABCD1 gene) by miCLIP. This peak is not affected by FTO treatment nor PCIF1-KO, showing that it is a 5′-UTR m⁶A modification. Blue background color denotes m⁶A signal. Source data are provided as a Source Data file.