Fig. 1: Polishing nanopore sequencing basecalling using an iterative approach.

A Workflow overview. Sequence⁺⁺ represents sequences corrected with the reference. Basecaller⁺⁺ represents the basecaller updated using Sequence⁺⁺. B Mappability and alignment accuracy of RNA control oligos, in which counterparting canonical nucleotides were substituted using N1-methyladenosine (m1A), N6-methyladenosine (m6A) or 5-methylcytosine (m5C) respectively. Distributions of alignment accuracy were shown with ecdf (empirical cumulative distribution function). Guppy, results produced by the Guppy basecaller; Round1-4, results produced by each iteration. C RNA control oligo alignments from Guppy and iterative basecalling Round4 were visualized with IGV. Without losing generality, reads aligned to the reference contig “5mers_1_curlcake_1” were shown. D Alignment position comparison between Guppy and iterative basecalling Round4 results. Train, results for train datasets used in (B, C); Test, results for test datasets that are independent of “Train”. Arrows represent alignment directions. E Mappability and alignment accuracy comparison between train and test datasets.