Fig. 1

Short-read RNA-seq data sources and the analysis workflow to identify expressed TE loci. a A schematic of the sampling regime for the generation of grapevine short-read RNA-seq data. The grapevine RNA-seq data was derived from the V. vinifera embryogenic callus subjected to time-series stress treatment (each with three biological replicates). Published RNA-seq data of Arabidopsis thaliana seedlings of wild-type [40] and ddm1 [17] were also used in the analysis pipeline. b Computational workflow to identify TE expression candidates. The first and second sub-pipelines apply HISAT2 [42] for the alignment of sequencing reads against the reference genome and then use htseq-count [43] and Bedtools toolset [44], respectively, to quantify reads overlapping with TEs. While htseq-count only adopts unique-mapping reads, Bedtools incorporates both unique- and multi-mapping reads. The third sub-pipeline uses BWA [45] to align reads against TE sequences, after which the mates of TE-mapped reads would be fed into TEFingerprint [46] for mapping against the reference genome to capture danglers. The three sets of TEs passing through the filtering steps are joined together as a pool of expression candidates (see Methods, Fig. S1 and additional file 1 for detailed descriptions of definitions and thresholds)