Biology Ph.D. Dissertations


Predicting the Structure of RNA 3D Motifs

Date of Award


Document Type


Degree Name

Doctor of Philosophy (Ph.D.)


Biological Sciences

First Advisor

Neocles B. Leontis, PhD

Second Advisor

Craig L. Zirbel, PhD (Committee Member)

Third Advisor

Scott O. Rogers, PhD (Committee Member)

Fourth Advisor

Alexei Fedorov, PhD (Committee Member)

Fifth Advisor

R. Marshall Wilson, PhD (Committee Member)


RNA molecules participate actively in all aspects of gene expression, as information carriers, catalysts and regulators. Most RNA molecules form specific 3D structures to carry out their functions. Structured RNAs often contain modular 3D motifs, many of which occur in different RNA molecules where they play similar roles. For example, some RNA 3D motifs form specific interactions that direct folding or stabilize the folded structure. Others catalyze chemical reactions or bind small molecules such as metabolites, drugs and specific proteins. Therefore, identifying recurrent 3D motifs in RNA sequences and secondary structures provides valuable information about RNA function. Different sequences can form the same 3D motif, so motif definitions are needed that can account for the sequence variations of 3D motifs. We propose definitions based on decomposing 3D motifs into pairwise interactions, focusing on base-pairing, base-stacking, and base-phosphate interactions, which are the most important. Definitions for each type of interaction were developed for the FR3D (“Find RNA 3D”) program suite, manually evaluated and refined. Basepairs were classified into twelve geometric families, base-stacks into four basic types and base-phosphate interactions into ten types. These definitions were used with FR3D to construct symbolic 3D searches for common motifs and shown to give reliable results. FR3D was used to construct 3D structural alignments of representative 3D structures of the 5S, 16S, and 23S ribosomal RNAs of E. coli and the distantly related bacterium T. thermophilus, by aligning corresponding nucleotides that form equivalent pairwise interactions. To evaluate the structural conservation of basepairs, we defined three qualitative criteria for basepair isostericity and translated these into a quantitative measure, the IsoDiscrepancy Index (IDI). The IDI was applied to evaluate the usefulness of basepair isostericity to understand sequence variations in structurally conserved 3D motifs. The IDI shows that basepairs that are isosteric belong to the same geometric family, and groups basepairs in the same family in isosteric or near isosteric sub-groups. Applying the IDI to rRNA structural alignments showed that ~98% of aligned basepairs are isosteric or near isosteric to each other. This work shows that basepair isostericity is a valuable concept for understanding RNA 3D motif sequence variations.