Nanopore-based native RNA sequencing provides insights into prokaryotic transcription, operon structures, rRNA maturation and modifications

Felix Grünberger; Robert Knüppel; Michael Jüttner; Martin Fenk; Andreas Borst; Robert Reichelt; Winfried Hausner; Jörg Soppa; Sebastien Ferreira-Cerca; Dina Grohmann

It is noteworthy that the predicted large number of modifications lead to a loss in read quality. This 613 is for example seen when analysing the Pyrococcus data set, as the quality score is also based on 614 the comparison of detected to theoretical read signal (compare Supplementary Fig. 3c) . To look at 615 this observation in more detail, we compared the results from the raw squiggle Tombo approach 616 that is not dependent on mapping of the reads to the mapping properties of the reads. This 617 methodology is based on an approach that was recently used to identify m 6 A RNA modifications 618 from native RNA reads with an accuracy of ~90% 14 . After mapping, the frequency for the 619 respective nucleotide at every position in the 16S rRNA to be correct, wrong or deleted was 620 calculated. Homopolymer stretches in the RNA increase the likelihood that the correct number of 621 nucleotides is not correctly identified and this, in turn, leads to an increased number of deleted 622 nucleotides, which is also visible in the Haloferax data sets at the G-stretch adjacent to the m 6 2A 623 modifications ( Supplementary Fig, 16a,b) . The overall frequency of wrongly assigned and deleted 624 nucleotides was much higher in P. furiosus (16.4%) than in E. coli (10.7%) and H. volcanii (10.4%) 625 ( Supplementary Fig. 16c ). Interrogating the full-length 16S rRNA, we found that miscalled 626 positions are equally distributed among the whole transcript ( Supplementary Fig. 16d ,e,f). 627