DeepRNA-Reg: A Deep-Learning Based Approach for Comparative Analysis of CLIP Experiments

DeepRNA-Reg employs advances in deep learning to enable high-fidelity comparative analysis of paired datasets of high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP). In a HITS-CLIP experimental paradigm where Ago2 activity is precisely perturbed via gene knock-out of a microRNA cluster, DeepRNA-Reg offers a superior prediction set than the current best prescription for differential HITS-CLIP; furthermore, DeepRNA-Reg predictions adhere better to the ground-truth of the RNA primary and secondary structural motifs that enable miRNA-mediated targeting of RNA. DeepRNA-Reg uncovered novel mediators in the mechanism of microRNA-mediated restraint of type-2 immunity in T-Helper 2 cells. In a comparative analysis, DeepRNA-Reg predictions show greater translatability across distinct biological milieux, offering prediction sets with wide applicability for investigators. To investigate the ability of DeepRNA-Reg to identify differentially protein-bound loci between a pair of HITS-CLIP experiments, we performed Ago2 HITS CLIP on mouse Th2 cells with the Mirc11 and Mirc22 clusters (miRNAs 23, 24, and 27) knocked-out and Th2 cells with a WT genotype.