From benchmarking HITS-CLIP peak detection programs to a new method for identification of miRNA-binding sites from Ago2-CLIP data

Abstract

Experimental evidence indicates that about 60% of miRNA-binding activity does not follow the canonical rule about the seed matching between miRNA and target mRNAs, but rather a non-canonical miRNA targeting activity outside the seed or with a seed-like motifs. Here, we propose a new unbiased method to identify canonical and non-canonical miRNA-binding sites from peaks identified by Ago2 Cross-Linked ImmunoPrecipitation associated to high-throughput sequencing (CLIP-seq). Since the quality of peaks is of pivotal importance for the final output of the proposed method, we provide a comprehensive benchmarking of four peak detection programs, namely CIMS, PIPE-CLIP, Piranha and Pyicoclip, on four publicly available Ago2-HITS-CLIP datasets and one unpublished in-house Ago2-dataset in stem cells. We measured the sensitivity, the specificity and the position accuracy toward miRNA binding sites identification, and the agreement with TargetScan. Secondly, we developed a new pipeline, called miRBShunter, to identify canonical and non-canonical miRNA-binding sites based on de novo motif identification from Ago2 peaks and prediction of miRNA::RNA heteroduplexes. miRBShunter was tested and experimentally validated on the in-house Ago2-dataset and on an Ago2-PAR-CLIP dataset in human stem cells. Overall, we provide guidelines to choose a suitable peak detection program and a new method for miRNA-target identification.

Post-transcriptional gene silencing mediated by microRNAs is controlled by nucleoplasmic Sfpq

“…Similarly, we found that the direct reduction of the binding sites for the endogenously expressed miRNAs (endogenous Ago2-miRNA peaks; Supplementary Fig. 6e ) 34 in P19 cells upon Sfpq knockdown statistically only occurred in the 3′UTR (Fig. 4e and Supplementary Fig.…”

“…To assess the Sfpq dependency of the endogenous miRNA-binding sites, we selected Ago2 peaks that overlap (Jaccard index > 0.10) between the full set of 2065 Ago2-miRNA peaks identified by HITS-CLIP analysis from P19 cells 34 and the Ago2-miRNA peaks enriched in siCtr-transfected P19 cells over the siSfpq-transfected P19 cells condition identified by dCLIP analysis. We found that 245 Ago2-miRNA peaks overlap with the peaks coming from such a filter, indicating their dependency on Sfpq.…”

“…Ago2-let-7a peaks recruited in absence of Sfpq were identified by comparing siSfpq-let-7a-transfected P19 cells to let-7a-transfected P19 cells with dCLIP analysis and selecting peaks with a score greater than 7. For bindings sites of endogenously expressed miRNAs, we analyzed Ago2-miRNA peaks in siSfpq-transfected P19 cells using pyicoclip 57 with the same protocol as described in Bottini et al 34 . In Supplementary Data 7 is shown the recruited Ago2-let-7a or Ago2-miRNA peaks in siSfpq-transfected P19 cells.…”

Data Science Issues in Understanding Protein-RNA Interactions

“…This is further complicated by the different strategies in identifying the crosslink sites by the various experimental protocols ( Table 1, Supplementary Table 1). Benchmarking of tools is challenging, owing to the differences in experimental protocols, and our limited understanding of the ground-truth regarding RNA binding sites in vivo (33,34) . Nevertheless, we attempt to demonstrate the impact of the different CLIP protocols and computational tools through use of the RNA maps, which combine CLIP with orthogonal functional data to derive an estimate of ground-truth from the perspective that RNA landmarks regulated by an RBP should contain its nearby RNA binding sites (Box 1).…”

Re-evaluating Strategies to Define the Immunoregulatory Roles of miRNAs

“…Mapping of RBP and miRNA transcript specificity using high-throughput approaches like crosslinking and immunoprecipitation (CLIP) technologies [45, 46] will allow us to resolve the cell type and stimulus-specific RBP-miRNA interactions that shape host immunity.…”