SLAM-seq analysis of mouse transcriptome

Gene and 3′ untranslated region (UTR) annotations were obtained from the UCSC table browser (https://genome.ucsc.edu/cgi-bin/hgTables, mm10 vM14 3′ UTR). Adapters were trimmed from raw reads using cutadapt through the trim_galore wrapper tool with adapter overlaps set to 3 bp for trimming. For Quant-seq, concatenated fastq files were trimmed for adapter sequences, and masked for low-complexity or low-quality sequences using trim_galore, then mapped to the mm10 whole genome using HISAT v.2.2.1 with the default parameters. The number of reads mapped to the 3′ UTR of genes was determined using featureCounts. Raw reads were normalized to CPM. SLAM-seq analysis was performed as previously described^{58 (link)} using the SlamDunk package^{59 (link)}. Trimmed reads were further processed with SlamDunk (v.0.3.4 16). The ‘Slamdunk all’ command was executed with the default parameters except ‘-rl 74 -t 8 fastq.gz -n 100 -m -mv 0.2 -o Slamdunk2’, running the full analysis procedure (slamdunk all) and aligning against the mouse genome (GRCm38), filtering for variants with a variant fraction of 0.2. Unless indicated otherwise, reads were filtered for having ≥2 T>C conversions. The remaining parameters were left as defaults.
Analysis of differential gene expression was restricted to genes with ≥10 reads in at least one condition. Differential gene expression calling was performed on raw read counts with ≥2 T>C conversions using DESeq2 with the default settings, and with size factors estimated on corresponding total mRNA reads for global normalization. Downstream analysis was restricted to genes that passed all internal filters for FDR estimation by DESeq2. Plots of differential gene expression were visualized using the ggplot2 package in R with significant genes (P value < 0.05, |log2FC| ≥ 1). Reproducibility of replicates is shown in Supplementary Fig. 3.