Fig. 5

Relationship between non-conversion level at candidate m⁵C sites and mRNA translation state. bsRNA-seq libraries were grouped as biological triplicates per fraction (e.g. LibB1, LibC1 and LibE1 each report on sites detected in bsRNA-seq fraction 1), allowing the calculation of average non-conversion levels per individual site and per fraction. a Boxplots showing distribution of candidate site non-conversion levels across the polysome profile. Shown from left to right are all sites in protein-coding RNA (c.f. Figures 2c and 4d), as well as subsets of these sites in the 5′ untranslated region (UTR), coding sequence (CDS), 3′UTR and introns. Asterisks indicate Student’s t test p value comparing adjacent fractions. b Non-conversion levels per individual site across the polysome profile were partitioned into nine soft clusters using Mfuzz (see Fig. S9). A total of 254 candidate m⁵C sites in exonic mRNA regions (5′UTR, CDS, 3′UTR designation in panel a) were included based on having coverage in at least 9 out of 12 bsRNA-seq fraction samples and ≥ 10 average read coverage in each of the four bsRNA-seq fractions. Mfuzz clusters were grouped into three translation state trend categories by visual inspection, showing a negative (N = 154), neutral (N = 51) or positive trend (N = 49) with polysome association. Top panels: line graphs displaying individual site average non-conversion levels across fractions. Middle panels: boxplots showing distribution of site non-conversion levels in each fraction. Asterisks indicate Student’s t test p value comparing adjacent fractions. Bottom panels: boxplots showing distribution of site read coverage in each fraction. c Stacked bar charts showing distribution of sites in the different translation state trend categories across mRNA regions (top) and distribution of sites in different mRNA regions across translation state trend categories (bottom). Asterisks indicate p values following binomial test against the distribution of all sites. The legend given in panel a is applicable to all panels. See Fig. S10 for cluster analysis of additional sites with sufficient coverage only in bsRNA-seq fractions 2–4