Fig. 5

Increased length of eIF4A-dependent 5′UTRs drives increased localized structure potential, which confers increased dependency on eIF4A activity when positioned close to the coding sequence. a Diagrammatic representation of the sliding window approach used in this figure. First, the Δ reactivity is calculated for every possible window, after setting the width of the window and the step. Then, these windows can be filtered by certain criteria. b, c Boxplots depicting the Δ reactivity of the windows with the b biggest decrease or c biggest increase Δ reactivity per 5′UTR for eIF4A-dependent (4A-dep) and eIF4A-independent (4A-indep) mRNAs with different sized windows. P values were calculated by an un-paired, two-sided Wilcoxon test, without correcting for multiple comparisons. d Density scatter plot depicting the correlation between the Δ reactivity of each 20-nt window which gained in structure the most per transcript against its 5′UTR length. P value and the r correlation coefficient were calculated by a Pearson correlation test. The number of transcripts within each hexagon is denoted in the legend. e Δ reactivity of the 20-nt windows with the biggest decrease in Δ reactivity per 5′UTR for 4A-dep and a 4A-indep group that is matched by 5′UTR length. P value is calculated by an unpaired two-sided Wilcoxon test. f Density scatter plot depicting the correlation between the Δ reactivity of each 20-nt window which gained in structure the most per transcript against its GC content. P value and the r correlation coefficient were calculated by a Pearson correlation test. The number of transcripts within each hexagon is denoted in the legend. g, h Violin plots showing the binned positions within the 5′UTRs of the 20-nt windows which g increase in structure the most per 5′UTR, and h decrease in structure the most per 5′UTR for 4A-dep and 4A-indep mRNAs. P value is calculated by an un-paired, two-sided Wilcoxon test