A novel NLP-based method and algorithm to discover RNA-binding protein (RBP) motifs, contexts, binding preferences,...

Area of Science:

• Computational Biology
• Molecular Biology
• Genomics

Background:

• RNA-binding proteins (RBPs) regulate mRNA processing, but their binding patterns and functions are poorly understood.
• Existing computational methods for predicting RBP binding lack interpretability and fail to adequately address motif context and RBP-RBP interactions.
• There is a need for interpretable models to clarify the contextual factors influencing RBP binding specificity in vivo.

Purpose of the Study:

• To develop a novel computational pipeline for analyzing RBP binding specificity.
• To characterize RBP binding motifs and contexts for a large number of RBPs.
• To identify novel RBP-RBP interactions and their potential regulatory functions.

Main Methods:

• Utilized a Natural Language Processing-based decomposition to represent RNA sequences.
• Formulated RBP binding prediction as a weakly supervised Multiple Instance Learning problem.
• Developed a deterministic motif discovery algorithm for interpretable predictions.

Main Results:

• Successfully recapitulated known RBP motifs and characterized novel binding motifs and contexts for 71 RBPs.
• Identified potential cooperative and competitive RBP-RBP interaction partners using feature integration and cross-prediction.
• Proposed hypotheses for the regulatory functions of identified RBP interactions.

Conclusions:

• The developed computational strategy effectively investigates contextual determinants of RBP binding.
• The findings significantly advance the understanding of RBP binding patterns, interactions, and functions.
• This work provides a framework for future research into RBP-mediated gene regulation.