Abstract
The structures of RNA exhibit recurrent patterns, which are defined as motifs. Motifs are crucial for the biological functions of RNA molecules. A 3D RNA motif library includes 3D RNA structures, providing a resource for motif-based 3D RNA structure prediction and design. Existing motif libraries face challenges in systematically extracting and classifying highly complex topologies with crossing interactions and non-Watson-Crick interactions. To overcome this limitation, we built RNAdex, an RNA motif library that stores 3D atomic coordinates and incorporates non-Watson-Crick interactions and crossing interactions from the Protein Data Bank. Leveraging motifs from RNAdex, we developed a measure of motif-content similarity between RNAs and observed that the pairwise similarity scores resemble a power-law distribution. To explore this macroscopic statistical pattern, we present a physics-inspired, hypothesis-generating framework for RNA structural evolution. RNAdex motif graphs were used to compare motif content and formulate testable hypotheses about RNA structural evolution. RNAdex is available athttps://rnadex.dokhlab.org.