Localized, Non-parametric Detection of RNA Structural Modification using Nanopore Basecalling.

Recently, much work has been done on chemical probing strategies with nanopore sequencing to identify RNA modifications at the single nucleotide level. Here, we examine the use of Oxford Nanopore's Guppy basecalling to identify structural modifications using localized, non-parametric peak detection. In a novel experiment, we evaluate whether detection of structural modifications is possible using the Guppy's basecalling error and determine the accuracy of our approach for selected RNA control sequences. Next, we use statistical analysis to determine the dominant structural bindings in a set of averaged read errors. Finally, we compare our approach to average reactivity determined by orthogonal experiments from SHAPE-CE and alternative approaches. We show that localized, non-parametric peak detection demonstrates improved accuracy and coverage of structural modifications in selected control RNA and that our method is agnostic to underlying changes in the distribution. Our approach allows for a more generalizeable methodology for detecting structural modification with nanopore sequencing and the subsequent generated probabilities can be used to refine further downstream analysis.