Differential and Coherent Processing of Small RNA-Seq Data

Small RNA transcripts are not only differentially expressed across different tissues, but they can also be differentially processed, exemplified by microRNA arm switching. They can also be remarkably coherent in terms of their processing, and this is the case for many microRNAs and snoRNAs. However, the extent of this has not been analyzed systematically. Using a pipeline implemented around the deepBlockAlign [1] tool, we analyzed the processing pattern of small RNA transcripts, represented in the form of their read profile, across nine human tissues (with replicates) from the ENCODE project. We obtained in total 701 genomic loci with expression in both biological replicates (ρ>0.7) of all nine tissues of which 97 loci (14%) showed differential processing and 353 loci (50%) were coherently processed. The 97 instances of differentially processed loci more than double the 39 known cases (all microRNA arm switching) and cover novel microRNA, snoRNA, tRNA and unannotated cases. For the 353 coherently processed loci, we observed 195 instances of distinctly small (<22 nt) transcripts enriched (p<0.01) for overlap to transcription start sites and DNase I hypersensitive sites, interestingly with no overlapping annotation of non-coding RNA. The remaining 158 loci mostly overlap annotated miRNAs and snoRNAs. Our analysis suggests that differential and coherent processing is persistently appearing in the human transcriptome, and we anticipate that it is present in the transcriptome of other organisms as well.

 [1] deepBlockAlign: a tool for aligning RNA-Seq profiles of read block patterns. Langenberger, Pundhir, Ekstrom, Stadler, Hoffmann, Gorodkin. Bioinformatics, 28:17-24, 2012.