Identifying proximal RNA interactions from cDNA-encoded crosslinks with ShapeJumper

Author summaryIn principle, crosslinking represents a simple and elegant way to measure important features of RNA structure. Crosslinking-derived, close-in-space structural information can be highly useful for modeling complex higher-order RNA structure and for generating hypotheses regarding how an RNA functions. In practice, extracting the information from an RNA crosslinking experiment, rigorously and at nucleotide resolution, has been difficult and imprecise. This work outlines the development and optimization of an analysis pipeline, called ShapeJumper, that substantially facilitates analysis of RNA crosslinking experiments, based on easily implemented JuMP technology. Both the crosslinking experiment and the analysis software described here are readily implemented by non-expert users. SHAPE-JuMP is a concise strategy for identifying close-in-space interactions in RNA molecules. Nucleotides in close three-dimensional proximity are crosslinked with a bi-reactive reagent that covalently links the 2'-hydroxyl groups of the ribose moieties. The identities of crosslinked nucleotides are determined using an engineered reverse transcriptase that jumps across crosslinked sites, resulting in a deletion in the cDNA that is detected using massively parallel sequencing. Here we introduce ShapeJumper, a bioinformatics pipeline to process SHAPE-JuMP sequencing data and to accurately identify through-space interactions, as observed in complex JuMP datasets. ShapeJumper identifies proximal interactions with near-nucleotide resolution using an alignment strategy that is optimized to tolerate the unique non-templated reverse-transcription profile of the engineered crosslink-traversing reverse-transcriptase. JuMP-inspired strategies are now poised to replace adapter-ligation for detecting RNA-RNA interactions in most crosslinking experiments.