The ribosome profiling landscape of yeast reveals a high diversity in pervasive translation

Figures

• 

  Fig. 1. Construction of the Ribo-Seq landscape. Left: example of calculation of the fractions of F0, F1, and F2 reads of one CDS (a in orange) and two aORFs (b and c in blue). Each Ribo-Seq read (vertical bars) can be assigned to a unique nucleotide according to its genomic coordinates and associated with a specific frame. For each considered ORF, we calculate its metagene, reflecting the distribution of reads over the ORF’s F0, F1, and F2 frames, as well as the read fractions in the three frames (see Methods). The ORF is then represented on a 2D graph according to its fractions of reads in its F0 and F1 frames (the fraction of reads in its + 2 frame (F2) can be calculated as F2 = 1 − (F0 + F1)) (middle—top part). As such, the CDS a is associated with the coordinates (85,9). In contrast, the aORF b whose metagene exhibits a + 2 shifted read periodicity will be associated with the coordinates (10, 8). Right: The same is done for inter-CDS regions. Theoretically, scanning ribosomes are expected to lead to reads evenly distributed across the three frames, leading approximately to F0 = F1 = F2 = 33%. Above 33% of in-frame reads, ORFs are considered enriched in in-frame reads. Despite this…

• 

  Fig. 2. Ribo-Seq landscape of coding regions. A Ribo-Seq landscape of the coding regions’ ORFs, including CDSs and aORFs, colored orange and blue, respectively. Each ORF with at least 50 reads and 30% of codon coverage is represented as a dot on the Ribo-Seq landscape according to its fractions of F0 reads (x-axis) and F1 reads (y-axis). B Metagenes of the three extremities of the Ribo-Seq landscape. Each metagene is calculated for the first 20 codons from all the ORFs associated with a given Ribo-Seq area (see the “Methods” section) and represents the average number of reads per nucleotide over all the ORFs of the metagene. The corresponding Ribo-Seq landscape areas are colored on the schematic Ribo-Seq landscapes of each plot. Nucleotide positions that are covered by in-frame reads relative to the ORFs of the metagene are colored in orange or blue, while positions covered by out-of-frame reads (F1 and F2) are colored in gray. C Ribo-Seq landscape of the CDSs colored with respect to their evolutionary age (in Mya) as estimated by phylostratigraphy (see the “Methods” section). D Top: Metagene of the CDS YNL046W representing the number of reads of each nucleotide summed over the 54…

• 

  Fig. 3. A Ribo-Seq landscape of the aORFs and CDSs observed in our MS data. Same representation as Fig. 2A, but aORFs and CDSs for which we detected a peptide product are colored in dark blue or dark orange, respectively. B Ribo-Seq landscape of the CDSs colored according to their protein abundance in ppm as provided by PaxDb [67]. C Venn diagram of the number of aORFs detected with MS by the two experimental conditions, standard (left) and inhibited proteasome (right). D Number of codons of aORFs detected with MS experiments (blue) and aORFs that were not detected (light blue). E Number of Ribo-Seq datasets in which an aORF has been observed for detected (blue) and not detected aORFs (light blue). F Reads per kilobase of exon model per Million mapped reads (RPKM) computed over all datasets for detected (blue) and not detected aORFs (light blue). G Distribution of the reads across the three overlapping frames for the aORFs detected in MS (top) and aORFs not detected in MS (bottom). H RPKM computed over all datasets of CDSs that overlap aORFs detected with MS (orange) and CDSs whose overlapping aORFs were not detected (light orange)

• 

  Fig. 4. Ribo-Seq landscape of noncoding regions. A Ribo-Seq landscape of iORFs. Each read-associated iORF is represented as a purple dot according to its fractions of F0 reads (x-axis) and F1 reads (y-axis). Cyan dots correspond to iORFs detected with MS. B–F Metagenes of the five regions of the Ribo-Seq landscape representing the average number of Ribo-Seq reads of each nucleotide for the first 20 codons. Nucleotide positions covered by in-frame reads (F0) for the considered iORFs are represented in purple, while those covering reads in the + 1 and + 2 frames are colored in gray

• 

  Fig. 5. Robustness of Ribo-Seq signals across individual datasets. A Distributions of the fractions of datasets in which the ORFs present in the global Ribo-Seq landscape were detected. Distributions for iORFs, aORFs, and CDSs are colored purple, blue, and orange, respectively. B Cumulative fractions of the ORFs of the global Ribo-Seq landscape that were detected when cumulating the reads of the different datasets. Fractions for iORFs, aORFs, and CDSs are colored purple, blue, and orange, respectively. Datasets are ranked according to their number of reads. C Distributions of the Euclidean distances calculated between the 2D coordinates of an ORF in the individual datasets

• 

  Fig. 6. Impact of the initiation codon on the Ribo-Seq signals. A Ribo-Seq landscape of the iORFs for which we detected the translation initiation codon. iORFs are colored according to the type of the predicted start codon with dark blue, light blue, and gray dots corresponding to initiation with ATG, near cognate codons, or other codons, respectively. Densities of the three different types of codons with respect to the fraction of in-frame reads are represented on the top of the plot. B Barplots representing the frequencies of the 61 codons at the predicted start codon for intermediate-FIR (salmon) and high-FIR iORFs (dark red). The small circles indicate the expected frequency of each codon according to its frequency in the intermediate- or high FIR iORFs. Codons that are significantly enriched at the predicted start codon are indicated with a star (one proportion z-test, P < 0.05). Near-cognate codons are represented in bold

• 

  Fig. 7. Topology of the yeast Ribo-Seq landscape. Schematic representation of the entire diversity of Ribo-Seq signals that allows (i) their comprehensive and comprehensible classification and (ii) the identification of areas that are either specific to pervasive translation or present in both pervasive and regular translation