Network-Based RNA Structural Alignment Through Optimal Local Neighborhood Matching

Abstract

Structural alignment of RNAs provides an effective way to compare and study RNAs, especially for those that have low sequence similarity yet retain high similarity in terms of their secondary structure. In this paper, we propose an enhanced network-based algorithm for accurate pairwise structural alignment of RNA sequences. Different from many traditional RNA structural alignment approaches that align RNA sequences while explicitly predicting the global consensus secondary structure conserved in different RNAs, our proposed algorithm implicitly infers the most likely alignment between bases, where the goodness of a potential base alignment is evaluated based on the local structure and composition around the bases to be aligned. For this, we take a network-based approach, recently proposed in our TOPAS algorithm, where topological networks are constructed based on the RNAs such that they capture the sequence composition as well as the underlying secondary structure of the RNAs. We align the constructed topological networks based on optimal network-based neighborhood matching. We will demonstrate the advantages of the proposed algorithm over existing schemes, in terms of both accuracy and efficiency, based on RNA structural alignment benchmarks.