SQUARNA - an RNA secondary structure prediction method based on a greedy stem formation model

Abstract

Non-coding RNAs play a great variety of roles in many cellular processes with their spatial structure known to dictate the functioning. RNA secondary structure largely determines the molecule’s global fold, which together with the paucity of experimentally determined RNA 3D structures makes its knowledge crucial for determining the function of the molecule. Currently, there is no one good solution for de novo RNA secondary structure prediction, with the existing methods getting more and more complicated without substantial progress in accuracy and being subject to drastic limitations such as ignoring pseudoknots. In this work, we present SQUARNA, a new approach for de novo RNA secondary structure prediction based on a straightforward greedy stem formation model overcoming many limitations of the existing tools. The benchmarks show that SQUARNA is on par with state-of-the-art methods for a single sequence input and significantly outperforms existing tools for a sequence alignment input.