Nanopore direct RNA sequencing is becoming an important tool for studying the epitranscriptome, the collection of chemical modifications that decorate RNA molecules and influence how they function. These modifications can affect RNA stability, translation, and gene regulation, making accurate detection essential for understanding biology and disease.

Researchers at the Center for Genomic Regulation (CRG) in Barcelona, Spain, evaluated how well recently released Oxford Nanopore Technologies basecalling models identify several common RNA modifications. These models are designed to recognize N6-methyladenosine (m6A), inosine, pseudouridine, and 5-methylcytosine directly during sequencing.

The researchers tested four modification-aware basecalling models using both synthetic RNA molecules and biological samples from multiple species. In highly controlled synthetic samples containing abundant modifications, the models performed well, showing strong accuracy in identifying modified bases.

However, the picture changed when the researchers examined biological samples. Real RNA molecules contain modifications at much lower frequencies and in more complex contexts. Under these conditions, prediction accuracy dropped substantially, and false positive rates became much higher.

Experiments using ribosomal RNA samples confirmed these limitations. Even after applying controls designed to remove background signals, false discovery rates ranged from 50% to 100%. This means many predicted modifications were actually incorrect.

The team identified two major causes of these false positives. First, different RNA modifications can produce similar electrical signals, causing cross-reactivity between modification types. Second, modifications can alter the electrical current patterns of neighboring nucleotides, leading to errors at nearby positions.

The researchers also demonstrated that alternative approaches based on analyzing electrical current signals and sequencing errors can accurately identify modifications. These methods may be particularly valuable for RNA modifications that currently lack dedicated basecalling models.

The findings highlight both the promise and current limitations of modification-aware nanopore sequencing. As RNA sequencing technologies continue to improve, careful use of control samples and complementary analytical approaches will remain important for generating reliable epitranscriptomic data.