A survey of RNA modifications across the human transcriptome by direct RNA nanopore sequencing

Abstract
RNA modifications have been demonstrated to be increasingly important for the function of most transcripts. Direct RNA nanopore sequencing preserves RNA modifications during sequencing, which can be identified by characteristic changes to the ionic current signals or by systematic errors in the aligned sequences. There have been many community software tools that take advantage of these types of signals to detect the presence of RNA modifications. These tools typically compare the signal from a sample of interest with another sample where the RNA modification is absent, such as with a knockdown or knockout of a modification writing enzyme. Another approach to detect RNA modifications is to compare the sample of interest against an in vitro transcribed version of the original sample, which will erase the signal for all RNA modifications. Here, we generated an in vitro transcribed sample from K562 cells and used Nanocompore to detect all RNA modifications in the mappable reads. We found more than 26,000 RNAmodified sites among 2,500 transcript isoforms from 1,700 genes. We were able to molecularly phase the RNA modifications and found a consistent signal that m5C and m6A were co-deposited on the same molecules. At the site level, we found that 11.5% of m5C sites could be detected within 25 nucleotides of an m6A site. Furthermore, we found that most transcript isoforms of the same gene were differentially modified. This survey of RNA modifications across the transcriptome demonstrates the potential for direct RNA nanopore sequencing for detecting all modifications simultaneously without the need for additional biochemical assays, critical for generating new hypotheses and promoting the study of epitranscriptomics.

Biography
Logan Mulroney is an ETPOD fellow at EMBL-EBI and the Center for Genomic Science at the Italian Institute of Technology where he works to detect and understand RNA modifications using nanopore data.