Computational Approaches for RNA Structural Motif Analysis: Similarities, Variations, and Spatial Relationships

The three-dimensional structures of RNA play a critical role in understanding their functions. One way to study the RNA 3D structure is by analyzing recurrent segments within the loop regions of RNAs, known as RNA structural motifs. These motifs can be grouped into families based on their structural features. However, the inherent structural diversity presents challenges in characterizing the similarities, variations, and spatial relationships within RNA motif families. This dissertation presents three computational methods that together provide a comprehensive framework for RNA structural motif analysis. First, RNAMotifComp establishes and visualizes relational networks among structural motif families. By analyzing instances of well-known families, it identifies those that are visually similar or share base interactions despite structural differences. This analysis reveals functional analogies among divergent families and highlights cases where motifs from distinct families are predicted to belong to the same one. Second, RNAMotifProfile introduces a graph-based approach to model internal diversity within motif families. Using a profile-to-profile alignment algorithm, it captures base interactions and associated nucleotides at each position, generating profiles that reflect both conserved and variable features. These profiles enable more accurate motif searches and enhance our understanding of structural plasticity. Finally, RNAMotifModule shifts the focus from individual motifs to their collective spatial behavior. While individual motifs contribute to specific biological functions, the impact of their co-occurrence in close spatial proximity remains underexplored. RNAMotifModule analyzes RNA structures using optimized atom-atom distance calculations to identify functional motif modules. These modules are statistically validated and biologically interpreted using high-throughput CLIP-seq data, revealing potential roles in coordinated RNA functions. Together, these methods advance the computational analysis of RNA 3D structures and offer novel insights into RNA architecture, with applications in RNA therapeutics and synthetic biology.