XPore software detects over 100 RNA modifications with AI

Researchers at Singapore Agency for Science, Technology and Research’s Genome Institute have developed xPore, a software that uses artificial intelligence to identify more than 100 RNA modifications in genomic data. These RNA modifications represent a “hidden layer of information” beyond standard RNA sequences and can influence cellular function, leading to disease risk and mRNA vaccine development. Unlike previous methods that require extensive laboratory work, xPore utilizes Nanopore direct RNA sequencing to analyze native RNA while preserving modifications. The team reused existing AI tools to detect the nuances created by these changes. “When we speak, the same word can have completely different meanings depending on its pronunciation and context,” said Dr. Jonathan Gauke, transcriptomics group leader at GIS, drawing parallels to how chemical changes change the function of RNA.

xPore software extracts RNA modifications from genomics data

More than 100 different RNA modifications are now known to influence cellular function, adding complexity to traditional understanding of RNA sequencing and prompting a re-evaluation of genomic data analysis. Researchers have long recognized that these modifications influence processes ranging from disease susceptibility to mRNA vaccine efficacy, with m6A methylation of adenosine being a common example. Previously, identifying these modifications required extensive laboratory work, limiting the scale and speed of research. However, xPore is designed to address these limitations. The software employs a machine learning approach, reusing tools from artificial intelligence research to accurately identify subtle differences that indicate RNA modifications. Recognizing that modifications disrupt this consistency, the research team focused on identifying consistent data from unmodified RNA sites, allowing the algorithm to accurately detect changes. This approach has already demonstrated potential in clinical practice.

The researchers, in collaboration with Professor Chng Wee Joo, Director of the Cancer Institute at the National University of Singapore, successfully used xPore to detect m6A RNA modifications in samples from multiple myeloma patients. “We were interested in studying m6A modifications in myeloma, as this could have important clinical and therapeutic implications for patients with poor outcomes, and now with xPore we have an important tool to accelerate our research,” added Professor Chng. Assistant Professor Sho Goh of the Shenzhen Bay Research Institute highlighted the flexibility of xPore, noting that its ability to map new RNA modifications without the need for specific reagents could facilitate the discovery of new RNA modification functions.

The ability to map new RNA modifications is essential to determining their function. xPore does not require specific reagents dedicated to identifying only a single RNA modification type, and therefore has the potential to detect other RNA modifications beyond m6A.

Dr. Shaw Goh, Assistant Professor, Shenzhen Bay Research Institute

Beyond the well-known sequences of RNA bases, there are over 100 different RNA modifications that influence cellular function and represent a complex regulatory landscape that scientists need to explore. This technique determines the sequence of native RNA and conclusively preserves its modifications. This is a significant advance over previous methods that required destructive chemical treatments. At the core of xPore is a machine learning approach that reuses established artificial intelligence tools to identify subtle differences in RNA molecules. The research team employed statistical models commonly used in data science to pinpoint these modification sites, allowing them to detect multiple RNA modifications simultaneously.

In this study, we introduce a computational approach that enables the profiling of differential RNA modifications across the transcriptome and provides a systematic resource of direct RNA-Seq data.

Professor Patrick Tan, GIS Executive Director