Fig. 3

Examples of structure score aggregation using the data from Structure Surfer across RBP motif match sites for three RBPs, PABPC5 (a-b), SNRPA (c-d), and SRSF7 (e-f). Human structure scores are aggregated at match sites in the human exome (a, c, and e), and mouse scores are aggregated at match sites in the mouse exome (b, d, and f). In all examples, the RBP interacting motif sequence is a heptamer occupying nucleotide positions 21–27. The score at each position is calculated as the average score across all nucleotides at that position relative to the RBP motif. PABPC5 shows a consistent dip in secondary indicating that sites matching its motif have, on average, less secondary structure than surrounding nucleotides. The SNRPA motif shows the opposite trend. Specifically, the average structure scores at sites containing this motif are higher than the surrounding nucleotides indicating that these sites tend to be double stranded. Sites for SRSF7 show a more complex pattern in which the different experiments do not form a consensus. PARS demonstrates evidence for a peak in average secondary structure at SRSF7 motifs, while ds/ssRNA-seq and DMS display evidence for a dip in average secondary structure. The icSHAPE experiments both show a region where some positions appear to be involved in base pairing and others appear unpaired