Abstract
Gene expression studies and transcriptome analyses are essential tools of molecular biology although, until a few years ago, the respective protocols suffered from two limitations, namely Polymerase Chain Reaction (PCR) amplification and read length. All the gene expression study-protocols included at least one step of PCR preventing, de facto, the possibility of measuring the “real amount” of a transcript. Transcript characterization, instead, is limited by the maximal read length achievable by the sequencing technology, especially in species with high heterozygosity or with a high degree of genomic duplications. Oxford Nanopore Technology (ONT) direct RNA sequencing, is a peculiar solution to the complementary DNA conversion needed for other platforms. Besides, direct RNA can potentially sequence entire transcripts and also resolve the complexity due to isoforms and to the alternative splicing processes. Since the error rate in ONT (flowcells R9.4.1) is quite high (5-15%), the investigation of the errors, especially in DNA sequencing, has been widely done to understand the main sources of insertion, deletion, and substitution other than natural biological variation. However, very few studies are present in the literature investigating the errors occurring in direct RNA sequencing. In this work we present a comparison between the overall error occurrence together with a more detailed investigation on each source of error (insertion, deletion, substitution), in human transcriptome reads obtained with direct RNA sequencing and the classical rates known for the DNA sequencing. Error-rate assessments are performed by parsing transcripts alignment files in search of single errors category that are then summarized in the complete error-rate. Insertion, deletion, and substitution-rates showed a similar pattern for both direct RNA and DNA sequencing indicating that all the software and pipelines developed for ONT cDNA protocols can be also used for the analysis of data obtained from direct-RNA experiments.