tailfindr: Alignment-free poly(A) length measurement for Oxford Nanopore RNA and DNA sequencing

Maximilian Krause; Adnan M. Niazi; Kornel Labun; Yamila N. Torres Cleuren; Florian S. Müller; Eivind Valen

We extended our algorithm to accommodate ONT DNA sequencing data output (Fig. 2B ). Most significantly, we account for the double-stranded nature of DNA and define the read type (poly(A)-or poly(T)-containing) by making use of known sequence motifs in Nanopore Adapters (details in Material and Methods). As for RNA sequencing, tailfindr provides the user with a tabular output of tail length measurements (including the read type) as well as optional raw data plots ( Figure S3 ). We tested the performance of the DNA-specific tailfindr algorithm on PCR products of eGFP coding sequence with known poly(A)/(T) length in two replicates, similar to the spike-ins generated for native RNA sequencing. As shown in Figure 2C Next we compared poly(A)-length estimates from PCR-DNA and native RNA sequencing. We observed that DNA sequencing results in significantly more precise estimation of poly(A) tail length, mainly due to fewer outliers towards longer poly(A) tail lengths (Fig. 2D ). Especially the shortest poly(A) tail length (10 nt) could be estimated more correctly in DNA sequencing (mode of poly(A) length estimation 10 in DNA vs 22 in RNA sequencing). On other poly(A) lengths the mode of poly(A) estimation does not differ dramatically, but the precision of measurement is significantly higher in DNA sequencing approaches (coefficient of variation 50% in native RNA sequencing vs 30% in DNA sequencing). In summary, tailfindr is able to estimate poly(A) and poly(T) tail size from ONT DNA sequencing with significantly higher precision compared to ONT RNA sequencing estimation.