Extended Data Fig. 1: Ribosome profiling captures active translation in the human adult and prenatal brain, Related to Fig. 1.

(a) Pie chart displaying the fraction of raw sequence reads derived from tRNA, ribosomal RNA (rRNA), mitochondrial RNA (mtRNA), and remaining aligned reads (clean) from human adult and prenatal brain RNA-seq and Ribo-seq. (b) Beeswarm plot of sequenced ribosome footprint lengths across all 73 adult and prenatal brain samples. Red line indicates the average percentage of Ribo-seq reads assigned to a given read length across all samples. (c) Bar plot of the percentage of reads mapping to the coding sequence (CDS) and untranslated regions (5′ and 3′ UTR) of annotated protein-coding genes (Refseq hg38). Each bar represents an individual sample. (d) Bar plot of the number of ORFs identified by RibORF in each sample after filtering (see Methods). Notably, we identify many more ORFs in the prenatal brain compared to the adult brain, which is likely at least in part a result of the longer post-mortem interval in adult compared to prenatal samples. (e) Pie chart of ORF types detected in our study as well as four previous studies^4,11,12,13. Because each study defines ORF types differently, ORFs are in hues denoting similar ORF types. Blue = CDS ORFs including non-canonical, out-of-frame ORFs; pink = uORFs and overlapping uORFs; orange = dORFs; yellow = ORFs translated from previously annotated non-coding RNAs. (f-g) PCA analysis of all genes in the human brain (f) RNA-seq and (g) Ribo-seq, colored by sample type (adult vs prenatal), post-mortem interval for adult samples, adult age, prenatal age (pcw), sex, and read depth (based on DESeq2 scale factors of estimated library size). The validity of combining samples into two groups in subsequent analyses was confirmed by the finding that these two groups were well separated by PCA analysis for both the transcriptome and translatome. (h) PCA analysis of batches of samples processed for RNA-seq (left) and Ribo-seq (right) to test for batch effect. We note a mild batch effect in some cases. To address the remaining batch effects in our differential expression analysis, we employed DTEG, an algorithm that uses DESeq2 to normalize across samples and includes batch correction. (i) Linear regression between postmortem tissue RNA integrity number (RIN) and read depth for RNA-seq and Ribo-seq. There is no significant correlation between the two variables (two-sided significance test for linear regression, p > 0.05 for RNA-seq and Ribo-seq). Gray shading = 95% CI. (j) Stacked bar plot of ORF types distributed by translation probability value, as calculated by RibORF. While RibORF uses a translation probability cutoff of 0.7 to determine significantly translated ORFs, we detect novel ORFs with a large range of translation probabilities.