Automated, customizable and efficient identification of 3D base pair modules with BayesPairing

Abstract

RNA structures possess multiple levels of structural organization. A secondary structure, made of Watson–Crick helices connected by loops, forms a scaffold for the tertiary structure. The 3D structures adopted by these loops are therefore critical determinants shaping the global 3D architecture. Earlier studies showed that these local 3D structures can be described as conserved sets of ordered non-Watson–Crick base pairs called RNA structural modules. Unfortunately, the computational efficiency and scope of the current 3D module identification methods are too limited yet to benefit from all the knowledge accumulated in the module databases. We present , an automated, efficient and customizable tool for (i) building Bayesian networks representing RNA 3D modules and (ii) rapid identification of 3D modules in sequences. uses a flexible definition of RNA 3D modules that allows us to consider complex architectures such as multi-branched loops and features multiple algorithmic improvements. We benchmarked our methods using cross-validation techniques on 3409 RNA chains and show that achieves up to ∼70% identification accuracy on module positions and base pair interactions. can handle a broader range of motifs (versatility) and offers considerable running time improvements (efficiency), opening the door to a broad range of large-scale applications.

Stochastic Sampling of Structural Contexts Improves the Scalability and Accuracy of RNA 3D Module Identification

“…The first software to tackle the specific task of identifying 3D motifs in full RNA sequences was RMDetect (2011) [8], which showed good accuracy but was severely limited in the variety of motifs it could identify. BayesPairing 1 improved on this method by adding more flexibility and improving its search efficiency [32]. Another method, JAR3D, does not undertake full sequence searches but scores hairpin and internal loops against a database of models from the RNA 3D Motif Atlas.…”

“…The good performances of BayesPairing 1 [32] relies on the assumption that the structural motif searched has a strong sequence signal. Indeed, the tool identifies motif location candidates through regular expressions.…”

“…In contrast to well-established secondary structure prediction tools [20,22], we are still lacking efficient computational methods to leverage the information accumulated in the module databases. Software such as RMDetect [8], JAR3D [34] and our previous contribution BayesPairing 1 [32] have been released, but their precision and scalability remains a major bottleneck.…”

“…Candidate modules found in an input sequence are then scored with Bayesian probabilities. However, while showing excellent accuracy, RMDetect suffers from high computational costs, and minimal structure diversity among modules predicted [32]. Another option is JAR3D [34], which refined the graphical model-based scoring approach introduced by RMDetect and represents the state of the art for module scoring.…”

“…However, it was not designed to maximize input sequence scanning efficiency and is limited in module diversity, only being applied to hairpin and internal loops. Finally, BayesPairing 1 [32], a recently introduced tool combining the Bayesian scoring of RMDetect to a regular expression based sequence parsing, is able to identify junction modules in input sequences and showed improved computational costs compared to RMDetect, which it was inspired from. Unfortunately, none of the aforementioned software can be used for the discovery of many RNA 3D modules in new sequences at the genome scale.…”