SoloTE for improved analysis of transposable elements in single-cell RNA-Seq data using locus-specific expression

Abstract

Transposable Elements (TEs) contribute to the repetitive fraction in almost every eukaryotic genome known to date, and their transcriptional activation can influence the expression of neighboring genes in healthy and disease states. Single cell RNA-Seq (scRNA-Seq) is a technical advance that allows the study of gene expression on a cell-by-cell basis. Although a current computational approach is available for the single cell analysis of TE expression, it omits their genomic location. Here we show SoloTE, a pipeline that outperforms the previous approach in terms of computational resources and by allowing the inclusion of locus-specific TE activity in scRNA-Seq expression matrixes. We then apply SoloTE to several datasets to reveal the repertoire of TEs that become transcriptionally active in different cell groups, and based on their genomic location, we predict their potential impact on gene expression. As our tool takes as input the resulting files from standard scRNA-Seq processing pipelines, we expect it to be widely adopted in single cell studies to help researchers discover patterns of cellular diversity associated with TE expression.

Transposable Elements are differentially activated in cell lineages during the developing murine submandibular gland

“…Transposable Element expression was measured using SoloTE (12). Briefly, SoloTE adopts the following strategy: first, reads not having the GN tag (i.e., not assigned to a gene) are selected, and the overlap between these reads and TEs is assessed using BEDtools (19).…”

“…We have previously taken advantage of the usage of this metric in Spatial Transcriptomics data, which follows a similar computational process during alignment and demultiplexing, to that of scRNA-Seq (20). Although not all TE locus are retrieved, it still shows an improvement to the default strategy of directly summarizing expression at the subfamily level (12). SoloTE generates as final output an expression matrix, that contains both gene and TE expression per cell.…”

“…Based on the number of mapping locations, SoloTE reports the locus of TEs if they have uniquely mapped reads, whereas for non-unique reads, TE expression is summarized at the subfamily level. Although not ideal, this strategy still allows the retrieval of locus-specific TE expression, which in turn, might help to understand their potential regulatory impact on genes (12).…”

“…Despite all the insights gained with scRNA-Seq and its computational applications, Transposable Elements are still not routinely considered to further aid in the understanding of cell processes and cell heterogeneity, as it was in the case of the work by Hauser et al It is thought that the study of these elements in scRNA-Seq works an further aid understand cell biology (10). Indeed, preliminary works have indicated that these elements can indeed be associated with specific cell types: a subset of mouse embryonic cells known as "2C-like" is marked by expression of the MERVL transposable element (11,12).…”

IRescue: uncertainty-aware quantification of transposable elements expression at single cell level

“…This tag-based layout decreases read mappability and makes it difficult to determine the exact genomic origin of TE-containing RNA fragments. Current tools for the quantification of TE subfamilies only use one alignment per UMI, disregarding the signal from ambiguous alignments on different TE subfamilies and do not take into account sequencing errors in UMI sequences (17, 18), that are common and introduce an overestimation of the UMI count (19). Approaches to rescue ambiguously mapping UMIs and obtain better expression estimates only exist for gene expression quantification, and leverage on finding UMI-transcript equivalence classes (20, 21).…”