Representation, searching and discovery of patterns of bases in complex RNA structures

Abstract

SummaryWe describe a graph theoretic method designed to perform efficient searches for substructural patterns in nucleic acid structural coordinate databases using a simplified vectorial representation. Two vectors represent each nucleic acid base and the relative positions of bases with respect to one another are described in terms of distances between the defined start and end points of the vectors on each base. These points comprise the nodes and the distances the edges of a graph, and a pattern search can then be performed using a subgraph isomorphism algorithm.The minimal representation was designed to facilitate searches for complex patterns but was first tested on simple, well-characterised arrangements of bases such as base pairs and GNRA-tetraloop receptor interactions. The method performed very well for these interaction types. A survey of side-by-side base interactions, of which the adenosine platform is the best known example also locates examples of similar base rearrangements that we consider to be 2 important in structural regulation. A number of examples were found, with GU platforms being particularly prevalent. A GC platform in the RNA of the Thermus thermophilus small ribosomal subunit is in an analogous position to an adenosine platform in other species. An unusual GG platform is also observed close to one of the substrate binding sites in Haloarcula marismortui large ribosomal subunit RNA.Abbreviations PDB Protein Databank 3D three-dimensional GNRA Guanine, N = any base, R = pyrimidine base, Adenine 3

FR3D: finding local and composite recurrent structural motifs in RNA 3D structures

“…Now we reject some of these partial candidates by applying the subset screening criterion (10) with I = {1, 2, 3}; we reject (p, q, r) triples for which (15) because we can be certain that any m-nucleotide candidate for which (p, q, r) correspond to ( (p 1 ,…, p k , q) for (1, 2,…, k, k + 1) provided that (16) so that all pairwise constraints between the k existing nucleotides and the one new nucleotide are met. We then use the subset screening criterion; we reject those candidates for which: (17) Here S * is the corresponding sum for (p 1 ,…, p k ). Thus, when adding the next nucleotide, k pairwise constraints must be checked, and k additional terms must be summed to apply the subset screening criterion.…”

“…Harrison and co-workers also apply subgraph isomorphisms to search for motifs in graphs representing RNA 3D structure [17], using methods that were first developed for substructure searching libraries of small molecule structures and then applied to proteins, carbohydrates, and most recently, RNA [4]. For RNA structure searching, each base is represented by two vectors and the whole RNA structure as a labeled graph so the search problem is reduced to finding subgraph isomorphisms representing query motifs in graphs representing RNA 3D structures.…”

“…For RNA structure searching, each base is represented by two vectors and the whole RNA structure as a labeled graph so the search problem is reduced to finding subgraph isomorphisms representing query motifs in graphs representing RNA 3D structures. This approach was applied to search for non-Watson-Crick basepairs and other small motifs in RNA structures [17]. As the Ullman algorithm used for subgraph searching scales with n factorial (n!…”

RNA structural motifs: building blocks of a modular biomolecule

“…Cluster analysis of root-mean-square deviations between pairs of RNA fragments has been used as the basis for cluster analysis of RNA loop structures to group tetraloops into various families (Huang et al 2005). A graph theoretic approach has been used to search for substructure patterns in RNA structures using a vectorial approach (Harrison et al 2003). Finally, the Structural Classification of RNA (SCOR), database has been developed to organize and classify RNA structural motifs (Klosterman et al 2002;Tamura et al 2004).…”

Chemoinformatics at the University of Sheffield 2002–2014

“…However, we have also made other contributions, e.g., in the application of chemoinformatics techniques to the representation and searching of biological macromolecules such as protein and RNA structures, [71][72][73][74][75] in the analysis of matched molecular pairs [76] and in charting the historical development of chemoinformatics and of its associated literature. [77][78][79][80] We have also sought to influence the development of the field by means of conferences and educational programmes.…”