DM-tRNA-Seq

We quantified mature and premature tRNA DM-tRNA-Seq datasets similar to the analysis described in¹⁹. Reads were trimmed using Cutadapt to remove adapters, artificially appended G’s generated as an artifact of the NextSeq platform, and sequences below the quality cutoff of 30⁴⁸. Reads were filtered to only retain species with a minimum length of 50 and maximum length of 200, reducing fragmented or partial reads which may introduce error during quantification. All mapping steps were performed using Bowtie2 with the following settings: -I: 50, -X: 300, -phred33, -end-to-end, -N: 0, -L: 20, -i: S,1,0.5, -n-ceil: L,0,0.15, -k: 2, -D: 20, -re-seed: 3. To discriminate between pre- and mature-tRNA species we first mapped data to an assembly of mature-tRNA sequences wherein 5’ and 3’ leader sequences are absent, introns are spliced, and 3’ CCA tails are present. Mature and premature-tRNA assemblies for hg38 were generated using the tRAX pipeline⁴⁹. Mitochondrial and nuclear tRNA loci were based on GtRNAdb^50,51 and tRNAscan-SE⁵² for hg38. Unaligned reads were then mapped to a modified host (hg38) genome in which tRNA genes were masked and instead appended as an additional chromosome containing pre-tRNA sequences with introns, and 5’ and 3’ leader sequences. We assessed whether the length of reads mapping to mature and premature-tRNA assemblies matched expectation using CollectInsertSizeMetrics. The median size for mature-tRNA mapped reads was 70-75 nt. Reads mapped to premature-tRNA has a multi-modal distribution peaking around 76, 87, and 100 nt. These sizes gave increased confidence regarding accurate discrimination of pre- and mature-tRNA. Transcript quantification was performed individually on mature- and premature-tRNA BAMs using salmon quant in alignment mode⁵³. Raw counts were normalized relative per million in vitro transcribed tRNA spike-in controls and per kilobase pair. Assessment of differentially expressed tRNA was performed using normalized counts matrices in EdgeR with a p-value adjusted (Benjamini-Hochberg, TMM normalization) threshold of 0.05 to achieve significance⁵⁴. As our pipeline maps sequentially to mature- and then premature-tRNA, the amount of pre-tRNA reads quantified will be an underestimate. Additionally due to high sequence overlap between tRNA genes for the same isodecoder, unless the premature reads contain locus specific flanking regions or introns these reads will be split between genes encoding the same isodecoder.