Agency For Science Technology And Research
Press Release 21 Jul 2021

Scientists Use Artificial Intelligence (AI) To Reveal Hidden Layer Of Information Above RNA

SINGAPORE – A team of researchers from Agency for Science, Technology and Research’s (A*STAR) Genome Institute of Singapore (GIS) has developed xPore, a software that extracts RNA modifications (an additional layer of information above the genetic molecule RNA) from genomics data. Their research was published in Nature Biotechnology on 20 July 2021.

"When we speak, the same word can have very different meanings depending on the pronunciation and context. For RNAs, we have something similar as chemical molecules may change the function of the same RNA. These RNA modifications are widespread, but because they do not change the letters of the RNA. They are very difficult to identify," said Dr Jonathan Göke, Group Leader of Laboratory of Computational Transcriptomics at GIS. More than 100 RNA modifications are known to play different roles in cells. Some of these RNA modifications are associated with disease risk, while others are used in mRNA vaccines. One of the most common modifications is the m6A methylation of the adenosine in RNAs. In the past, identifying RNA modifications required labour- and time-intensive bench-experiment assays that only very few laboratories can perform.

To overcome these limitations, the team utilised Nanopore direct RNA-sequencing, a new transcriptomic technology that sequences native RNA molecule with its modifications retained. To extract the hidden layer of RNA modifications, they developed xPore, a machine learning-based method that re-purposes tools from AI research to precisely detect differences in RNA modifications. A property employed in the method is the consistent data of the unmodified sites, and the existence of modifications disrupts this consistency.

"Similar problems occur in other data-rich areas such as finance or speech recognition that tap on machine learning. Here, we adopted an existing statistical model that is used frequently in data science, so that it can precisely identify these modified sites," explained Dr Ploy Pratanwanich, formerly a Postdoctoral Fellow, Epigenetics and Epitranscriptomics at GIS, current Lecturer at Chulalongkorn University (Thailand), and first author of the study. In their study, the authors demonstrated that xPore is highly accurate and is able to overcome many of the previous limitations in studying RNA modifications.

Collaborating with Prof Chng Wee Joo, Director of the National University Cancer Institute, Singapore (NCIS), the team successfully detected the m6A RNA modification using xPore in multiple myeloma cancer patient samples, showing xPore’s potential for large-scale clinical analyses. "We have been interested in studying m6A modification in myeloma as this may have important clinical and therapeutic implications for patients with poor outcome. Now with xPore, we have an important tool to facilitate our studies," Prof Chng added.

Dr Sho Goh, Assistant Prof from Shenzhen Bay Laboratory who co-led the study, said, “The ability to map new RNA modifications is vital for determining their functions. Since xPore does not require specific reagents that specialise in identifying only a single RNA modification type, it can potentially detect other RNA modifications beyond m6A. Therefore, xPore’s flexibility can expedite our efforts to discover novel RNA modification functions.”

Prof Patrick Tan, Executive Director of GIS, said, “This study introduces a computational method that enables the profiling of differential RNA modifications transcriptome wide, and provides a systematic resource of direct RNA-Seq data. It will be valuable as a benchmark data set for modification detection, with the potential to lead to better patient outcomes.”

– END –



ANNEX A – Notes to Editor



Photo Gallery