Structured RNA molecules play essential roles in a variety of cellular processes. This wide range of function is made possible by the diversity of three-dimensional structures that RNA can assume. Structural studies of RNA, however, present a large number of challenges and are frequently hampered by the complexity of the RNA backbone, the limited resolution typical of RNA crystallography, and the lack of computational tools for RNA modeling. These difficulties lead to errors in RNA structure determination, particularly in modeling of the RNA backbone, which is highly error prone due to the large number of variable torsion angles per nucleotide.

We have developed a method for accurately building the RNA backbone into electron density maps of intermediate or low resolution [1]. This method is semi-automated, as it requires a crystallographer to first locate phosphates and bases in an electron density map. After this initial trace of the molecule, however, an accurate backbone structure can be built without further input from the crystallographer. To accomplish this, three disparate approaches to RNA structure analysis are used: pseudotorsions, which simplify the backbone of each nucleotide to two virtual bonds; the consensus backbone conformers, which enumerate a limited number of allowed configurations for the RNA backbone; and the base-phosphate perpendicular distance, a heuristic for determining the pucker of the ribose sugar. To build the RNA backbone, conformers are first predicted using pseudotorsions and the base-phosphate perpendicular distance, and detailed backbone coordinates are then calculated to conform both to the predicted conformer and to the previously located phosphates and bases. This technique can produce accurate backbone structure even when starting from imprecise phosphate and base coordinates. We have implemented this technique as a Coot plugin, and are beginning work to incorporate this algorithm into Phenix.