Detection of m6A from direct RNA sequencing using a multiple instance learning framework

• DOI: 10.1038/s41592-022-01666-1
• Publication Date: 2022-11-10T17:04:18Z
• Authors: Christopher Hendra; Ploy N. Pratanwanich; Yuk Kei Wan; W. S. Sho Goh; Alexandre Thiery; Jonathan Göke
• Keywords: Identification; RNA methylation; Retraining

Abstract RNA modifications such as m6A methylation form an additional layer of complexity in the transcriptome. Nanopore direct sequencing can capture this information raw current signal for each molecule, enabling detection using supervised machine learning. However, experimental approaches provide only site-level training data, whereas modification status single molecule is missing. Here we present m6Anet, a neural-network-based method that leverages multiple instance learning framework to specifically handle missing read-level labels data. m6Anet outperforms existing computational methods, shows similar accuracy approaches, and generalizes with high different cell lines species without retraining model parameters. In addition, demonstrate captures underlying stoichiometry, which be used approximate differences rates. Overall, offers tool transcriptome-wide identification quantification from run sequencing.