Technical Highlight - July 2011
Short description: A new computational tool will allow researchers to easily predict RNA structural modules from the sequence alone.
It is no secret that tertiary structures in RNA are almost, if not equally, as important as the RNA sequence itself. These structures are formed by short- and long-range interactions between the bases in the RNA and often take the form of specific motifs, also referred to as modules. These modules include G-bulges, kink-turns, C-loops and tandem GA/AG loops (tandem GAs). However, although there are tools to predict RNA secondary structure motifs, they are limited by their ineffectiveness to treat the whole range of sequence variations. Three-dimensional structure prediction tools, which can also help to identify structural modules, are often not user-friendly in terms of the expertise and computer time they require to be used effectively.
Now, Cruz and Westhof have generated a new computational tool for detecting RNA structural modules on the basis of the RNA sequence alone. They name their tool RMDetect, short for RNA three-dimensional modules detection. Using Bayesian network models, base-pair probability prediction and positional clustering of candidates, RMDetect can predict the presence of each of the above-mentioned structural modules.
To validate the utility of RMDetect, the authors tested it against single target sequences as well as multiple sequence alignments. RMDetect performed well on both, although the false discovery rate (FDR) for tandem GAs in the single sequence analysis was higher than for the other modules. To counteract this, the authors caution that the training data set used for each model needs to be as complete and representative of the population as possible. RMDetect performed better on multiple sequence alignments (even when the same data sets tested in the single sequence analysis were used), demonstrating that the additional information provided by the alignments added to the robustness of the tool.
When RMDetect was applied to sequence alignments from the available databases and published data (including bacterial data), the authors uncovered several new structural modules of each type tested (G-bulges, kink-turns, C-loops and tandem GAs), with a particular prevalence of tandem GAs in some cases. This demonstrates that the RMDetect is a robust tool that will be integral to the hunt for tertiary RNA structural elements. The authors have also made available RMBuild, a tool to allow additional modules to be added to RMDetect, which will undoubtedly increase its utility and appeal to other researchers. This will then pave the way for the detection of RNA structural modules in many species, allowing greater insight into their biological significance.
J.A. Cruz, E. Westhof Sequence-based identification of 3D structural modules in RNA with RMDetect.
Nat Methods 8, 513-519 (2011). doi:10.1038/nmeth.1603