RNAelem: an algorithm for discovering sequence-structure motifs in RNA bound by RNA-binding proteins

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• RNA-binding proteins (RBPs) are critical for post-transcriptional gene regulation.
• Identifying specific RNA patterns recognized by RBPs is a key bioinformatics challenge.
• Current deep learning models for RBP-binding site prediction lack interpretability.

Purpose of the Study:

• To develop an interpretable computational tool for predicting sequence-structure motifs in RBP-binding regions.
• To improve the understanding of RNA-binding protein specificity.

Main Methods:

• Developed RNAelem, integrating profile context-free grammar with the Turner energy model for RNA secondary structure.
• Applied RNAelem to predict sequence-structure motifs in RBP-binding regions.

Main Results:

• RNAelem demonstrated superior accuracy in detecting RNA sequences with structural motifs compared to existing methods.
• Identified known primary sequence motifs and discovered novel secondary structural motifs, including sequence-nonspecific regions.
• Provided interpretable insights, such as long-range base-pairing interactions for U2AF protein.

Conclusions:

• RNAelem offers an accurate and interpretable approach for analyzing RBP-RNA interactions.
• The tool facilitates the discovery of complex sequence-structure motifs governing RBP binding.
• RNAelem advances the study of post-transcriptional regulation by elucidating structural determinants of RBP specificity.