Experimental and computational workflow for the analysis of tRNA pools from eukaryotic cells by mim-tRNAseq

Abstract

Quantifying tRNAs is crucial for understanding how they regulate mRNA translation but is hampered by their extensive sequence similarity and premature termination of reverse transcription at multiple modified nucleotides. Here, we describe the use of modification-induced misincorporation tRNA sequencing (mim-tRNAseq), which overcomes these limitations with optimized library construction and a comprehensive toolkit for data analysis and visualization. We outline algorithm improvements that enhance the efficiency and accuracy of read alignment and provide details on data analysis outputs using example datasets. For complete details on the use and execution of this protocol, please refer to Behrens et al. (2021).

Keywords: Bioinformatics; Gene Expression; Molecular Biology; RNAseq; Sequencing; Systems biology.

Graphical abstract

Figure 1

Testing mimseq environment and installation (A) Running mimseq --version displays the mimseq logo and version number, which should be higher than v1.1. (B) mimseq --help displays the help documentation on mimseq parameters.

Figure 2

Typical gel images of key steps during mim-tRNAseq library construction Gel regions to be excised are indicated by a red bracket. (A) Total RNA after step 66. (B) Adapter-ligated tRNA after step 92. (C) cDNA after successful reverse transcription in steps 105–113. (D) Library DNA after step 135. (E) cDNA after suboptimal reverse transcription in steps 105–113.

Figure 3

Schematic of the mimseq computational pipeline Highlighted are the main customizable parameters and the analysis steps they affect. See text in running mimseq and mimseq --help for detailed parameter descriptions.

Figure 4

Fundamental principles of the new cluster deconvolution algorithm (A) Schematic representation of deconvolution methodology. For each cluster, the set of mismatches distinguishing each transcript is found. For each set, the minimal unique subsets are found, from which the most 3′ subset is chosen. Reads are assessed individually for these mismatches in order to assign them to a member transcript within a cluster. (B) Schematic representation of conditions when clusters or transcripts cannot be deconvoluted. Either coverage at a required mismatch is too low (left), the mismatch is a potentially modified site (middle), or >10% parent assigned reads contain mismatches to the parent.

Figure 5

Alignment statistics for sample human dataset Shown are the uniquely mapping, multi-mapped, and unmapped read proportions per library after realignment (if enabled with --remap as in step 5).

Figure 6

Quality control for tRNA coverage and full-length transcripts (A) Metagene plots of coverage per nuclear-encoded tRNA isotype in each library specified in the sample data input file. Coverage is normalized to total mapped reads and scaled to the second last bin. See cov/coverage_byaa_norm_scaled.pdf. (B) Global heatmaps of average proportions of stops to RT per canonical tRNA position for each unique tRNA transcript with coverage above 0.005 (as per --min-cov) in tRNA sequencing data from human K562 cells (n = 2).

Figure 7

Assessing replicate similarity with variance stabilizing transformed (vst) tRNA count data from DESeq2 (A) Distance matrix representing pairwise Euclidean distance for each pair of samples. (B) Principal component analysis (PCA) plot using the first two principal components from tRNA isodecoder analysis. Percent variance explained by each principal component is given in axis titles.