a, The RNA fragment sizes were quantified using a 2100 Bioanalyzer. Top, total RNA after rRNA depletion. Middle, RNA after fragmentation. Bottom, RNA after C1 beads pull-down. b, c, The RIC-seq libraries were purified from the gel (the bracketed region) and quantified using a 2100 Bioanalyzer. −pCp denotes samples without pCp–biotin labelling. d, The mapping pipeline for RIC-seq data. e, Read types and their relationship to reference transcripts. f, The numbers of intra- and intermolecular chimeric reads. g, The intra- and intermolecular chimeric reads show a 4.2- and 5.5-fold increase, respectively, in the intron-to-exon ratio, as compared with RNA-seq data. h, The intra- and intermolecular RIC-seq signals do not decrease after the branch point (red star). i, Transcripts enriched in different subcellular fractions revealed by RNA-seq45 in HeLa cells. Two replicates for each fraction (rep 1 and rep 2). j, The number of RNAs enriched (blue) in each subcellular fraction and the numbers of RNAs that could be captured by RIC-seq (red). k, Cartoon of RBP-mediated RNA proximal interactions and data presentation. The structural models of the RBPs were generated in PyMOL on the basis of PDB accessions 5B16 and 4V6X. Light blue lines between chimeric reads represent gaps. Pairwise junctions are visualized as arc lines or plotted as a heat map. Colour intensity indicates the number of junctions within each 5 × 5-nt pixels. l, Scatter plots showing the reproducibility of RIC-seq-detected pairwise interactions in two biological replicates. The reproducibility for chimeric reads per transcript (R = 0.932) is displayed in a red box. RNA abundance is normalized for both scatter plots. m, Heat map showing the cross-species RNA–RNA interactions observed by the cell mixing strategy. The boxed region in dashed lines represents random ligations. The experiments in b, c were independently repeated three times with similar results.