Fig. 2

Interferon treatment elicits more Ψ modification in mRNA. a Log10 expression levels of genes in untreated sample and IFN β-treated (left) or IFN γ-treated (right) samples. Expression level is calculated as the peak height of the piled reads. Red, genes with an increase of > 2 fold in expression. Blue, genes with a decrease of > 2 fold in expression. b Venn diagram of the GO terms of the genes containing the 500 U sites with the highest Ψ probabilities in each sample. c Scatter plot showing the mean modification probability change versus log10 expression fold change of each gene between untreated and IFN β-treated (left) or IFN γ-treated (right) sample. Red, genes with an increase of > 2 fold in expression. Blue, genes with a decrease of > 2 fold in expression. d Mean Ψ modification probability of genes assigned to groups based on expression fold change between untreated and IFN β-treated (left) or IFN γ-treated (right) samples. ***p<10^-3, and ****p<10^-4. e GO analysis of the 50 genes with highest mean Ψ probability change between untreated and IFN β-treated (top) or IFN γ-treated (bottom) samples. Blue vertical line indicates p=0.05. f Mean Ψ probability change of the highest 50 genes between untreated and IFN β-treated (left) or IFN γ-treated (right) samples. Genes with a significant increase in expression levels are marked in red (>10 fold) or orange (5–10 fold). g Relative Ψ level of mRNA transcript of ACTB (left panel, data from set 1 and set 2 primers) and ISG15 (right panel, data from set 1, set 2, and set 3 primers) in the untreated and interferon-treated samples measured by RT-qPCR. *p < 0.05; **p < 0.01. h Single read prediction results for the partially modified Ψ sites in human rRNA. The stoichiometry predicted by our method is compared with the stoichiometry reported previously by quantitative LC/MS. The correlation coefficient is 0.6566 (Pearson’s r). i Clustering heatmap showing the Ψ probability of two pairs of sites in single reads of the B2M transcript in the IFN γ-treated sample. Each row represents a read. Site numbers are defined as the chromosomal locations in the hg38 nomenclature. These two pairs show either negative linkage (left) or no linkage (right). j Reads in panel i are assigned to “Ψ” and “U” groups based on the posterior probabilities of site 1 in Gaussian mixture model (k=2). The cumulative distribution curves of Ψ probabilities of site 2 are drawn for reads in “Ψ” or “U” groups or for all reads. The curves for “Ψ” and “U” groups undergo two sample Kolmogorov-Smirnov test; p values are <2.2x10^-16 (left) and 0.7684 (right). k P value in the two sample Kolmogorov-Smirnov test for selected pairs of sites in the B2M transcript in the untreated and IFN treated samples