Skip to main content
. 2021 Feb 17;12:1076. doi: 10.1038/s41467-021-21394-y

Fig. 3. Selective constraints on the upstream open reading frame (uORF) start codons.

Fig. 3

a Age distribution of start codons (uoAUGs) of human uORFs. The number of origination events assigned to each branch was inferred with the maximum parsimony method. b The scheme showing how the branch length scores (BLSs) for the start codons of two uORFs are calculated based on their presence or absence across species. In this hypothetical example, the length of each branch is denoted with ai (i = 1–8). c Empirical cumulative distribution function (ECDF) of the BLSs for uoAUGs, noncanonical start codons, and the other triplets in human 5′ untranslated regions (UTRs). The BLS of uoAUGs (total or translated) or noncanonical start codons was significantly larger than that of the other triplets (P < 8 × 10−58, two-sided Wilcoxon rank-sum tests). d Signal-to-noise ratios of the BLSs of uoAUGs and noncanonical start codons relative to other triplets in humans based on different thresholds of minimum BLS. The dashed line delineates a signal-to-noise ratio of 1 expected under neutral evolution. e ECDF of BLS for uoAUGs and the other triplets in fly 5′ UTRs. The BLS of uoAUGs (total or translated) was significantly larger than that of the other triplets (P < 2 × 10−88, two-sided Wilcoxon rank-sum tests). f The signal-to-noise ratio of BLSs for uoAUGs relative to other triplets in flies based on different minimum BLS thresholds. Source data are provided as a Source Data file.