Fig. 1: Workflow of DiffScan.

a Taking raw reactivities as input, DiffScan first normalizes them relative to one another in the Normalization module (b) to correct for systematic bias, and then identifies SVRs in the Scan module (c). b The Normalization module transforms raw reactivities into normalized reactivities to remove systematic bias. The raw reactivities are from the icSHAPE SRP vivo dataset which has no SVRs. The normalized reactivities are comparable as far as possible across different cellular conditions. c Taking normalized reactivities as input, the Scan module first calculates the significance of any differential signals for each nucleotide position with two-sided Wilcoxon test, and then concatenates positional p values into a regional signal via scan statistic. The significance of the scan statistic for each enumerated region is evaluated by Monto Carlo sampling, and those regions crossing a specified significance threshold are reported as SVRs.