RNAdemocracy: a Consensus Scoring Approach for Computational Prediction of RNA Secondary Structures

Computational RNA secondary structure prediction is an important tool for the characterization of nucleic acid. If no sequence homologues are available, the prediction of accurate structure becomes harder to achieve. Presently, popular methods are able to produce accuracies of 70% but struggle on long nucleic acid sequences. The improvement of established methods is slow and often relies on redundant methodology. With this in mind, a novel consensus scoring approach was created to incorporate the outputs of several of these established methods into consensus models. The RNAdemocracy program is a collection of python3 scripts implementing this consensus approach. This method allows users the ability to customize input options to best suit their sequence and can be operated in a variety of UNIX environments. RNAdemocracy utilizes a majority rules system to break disagreements between input structures, implementing those structures identified in more inputs into a consensus model. This consensus model is utilized as a constraint for a second round of secondary structure prediction that fills in remaining sequence space. The resulting outputs are able to capture important functional RNA motifs and the modular nature of the program allows it to be customized for specific structure identification. The consensus scoring approach is currently competitive with established methods and it has been determined that the improvement of input reliability may further the applicability. Furthermore, the novelty of the consensus approach provides a future opportunity for its improvement, through modular or algorithmic modifications.