New metric speeds RNA alignment across arbitrary pseudoknots

A sequence-based representation and alignment metric called SERNA can compare RNA secondary structures containing arbitrary pseudoknots with competitive clustering performance and lower computational cost than structure-based approaches, according to research published in BMC Bioinformatics. The authors also introduce SERNAlign, an open-source tool that calculates the SERNA distance through dynamic programming with quadratic time complexity.

Comparison of RNA secondary structures is fundamental to RNA classification, motif analysis and evolutionary studies, but methods designed for pseudoknot-free structures are not equally suited to structures with arbitrary pseudoknots. The study tackles that computational challenge by representing a secondary structure as an integer sequence, termed a structural sequence. On this representation, the researchers define the SERNA distance as an extension of classical edit distance that includes constraints intended to maintain structural correctness.

The researchers prove that SERNA is a metric, establishing that it meets the mathematical properties required for measuring distances. They then evaluated it within a clustering framework through two complementary experiments. The first involved classification of experimentally validated pseudoknot motifs and directly targeted the design goals of SERNA. The second examined phylogenetic clustering of ribosomal RNAs as a robustness check against existing structural distances.

Across both tasks, SERNA produced clustering performance competitive with state-of-the-art comparison methods. In complex motif settings, the method also showed improved discrimination, while requiring significantly less computational cost than structure-based approaches. The supplied study abstract does not provide detailed effect sizes, test statistics or uncertainty estimates for these performance differences, so the scale of the improvements cannot be quantified from the reported summary.

SERNA’s central technical advance is its use of integer sequences to encode RNA structures for comparison, combined with dynamic programming to compute alignments. SERNAlign implements the approach in quadratic time, meaning that its computational cost grows with the square of the input size. The two evaluation settings were intended to examine complementary aspects of the method: motif classification addressed the pseudoknot structures that motivated its design, while ribosomal-RNA clustering tested whether it could remain competitive on a broader evolutionary comparison involving established molecular data.

The work is a computational bioinformatics study rather than an investigation of biological mechanisms or a clinical intervention. Its reported conclusions are limited to the mathematical properties of the new distance and its performance in the two clustering evaluations described. The authors present SERNA and SERNAlign as an efficient addition to methods for comparing RNA secondary structures, particularly in settings involving complex pseudoknots.

Subject of Research: Sequence-based alignment of RNA secondary structures containing arbitrary pseudoknots

Article Title: Alignment of RNA secondary structures with arbitrary pseudoknots using structural sequences

Article References: Tesei, L., Levi, F., Quadrini, M., & Merelli, E. (2026). Alignment of RNA secondary structures with arbitrary pseudoknots using structural sequences. BMC Bioinformatics. https://doi.org/10.1186/s12859-026-06687-7

Image Credits: AI Generated

DOI: 10.1186/s12859-026-06687-7

Keywords: RNA, pseudoknots, secondary structure, sequence alignment, edit distance, SERNA, SERNAlign, dynamic programming, computational biology, phylogenetic clustering, Alignment, secondary