Computational Tools for Investigating RNA-Protein Interaction Partners

Abstract

RNA-protein interactions are important in a wide variety of cellular and developmental processes. Recently, high-throughput experiments have begun to provide valuable information about RNA partners and binding sites for many RNA-binding proteins (RBPs), but these experiments are expensive and time consuming. Thus, computational methods for predicting RNA-Protein interactions (RPIs) can be valuable tools for identifying potential interaction partners of a given protein or RNA, and for identifying likely interfacial residues in RNAprotein complexes. This review focuses on the "partner prediction" problem and summarizes available computational methods, web servers and databases that are devoted to it. New computational tools for addressing the related "interface prediction" problem are also discussed. Together, these computational methods for investigating RNA-protein interactions provide the basis for new strategies for integrating RNAprotein interactions into existing genetic and developmental regulatory networks, an important goal of future research.

RNA–LIM: A novel procedure for analyzing protein/single-stranded RNA propensity data with concomitant estimation of interface structure

“…Although it is capable of providing only relatively coarse structural predictions (individual ribonucleotide location P 3 Å), we believe that, due to its systematic nature, the RNA-LIM procedure is particularly well-suited to a role as the first step in a structural biology pipeline with layered levels of sophistication for investigating ssRNA-protein interactions. Refinement and optimization of the characteristic parameters used to define the RNA and protein interaction in the RNA-LIM procedure by its application to training data held in a number of curated RNA-protein structural databases [33] will facilitate this future goal. The approach may also show merit in examining the interaction of other classes of unfolded polymers in cases of chaperone function [36] and amyloid-polymer interaction [37].…”

“…Such a refinement could be included through reconfiguration of the statistics derived from the structural library of ssRNA-protein structures. Although this was not attempted in the current case, it may form the basis of a future improvement to the routine as more structures are included in the relevant protein-RNA databases [33]. Another disadvantage of the current routine is the inherent coarse-grained limitation to the structural accuracy (P3 Å).…”

“…For values of V CONSTANT P 0.5, the method can consistently produce many thousands of possible coarse/fine solutions, of which the top 10,000 were kept for further examination. Of these best estimates, those clustered into the left-hand side yield an RMSD of 15 Å or better-a value useful in the assignment of ssRNA-protein binding sites [33]. For values of V CONSTANT P 0.75, the RMSD in this preferred solution space (bottom left-hand corner) is typically 610 Å-a finding of significance due to the fact that success in assigning ssRNA binding residues by modern-day machinelearning approaches is consistently P0.85 [1][2][3].…”

“…9 A similar limitation is faced when considering very long nucleotide sequences for which N > 6. All three of these disadvantages can be offset by using the RNA-LIM procedure as the first leg of a structure evaluation pipeline to produce coarsegrained candidate structures that can be subsequently used for more detailed analysis using higher order approaches [1][2][3][12][13][14][15][20][21][22][23][24]33]. Despite the less common occurrence of very long segments of ssRNA being involved in binding [1,31], the ssRNA sequence size limitation (N 6 6) could be potentially offset by (i) consideration of longer sequences as a set of smaller sequence fragments or (ii) reconfiguration of the ssRNA homopolymer statistics in terms of every second ribonucleotide.…”

A Motif-Based Method for Predicting Interfacial Residues in Both the Rna and Protein Components of Protein-Rna Complexes

“…Recently, high-throughput (HTP) methods for identifying the in vivo targets of specific RNA binding proteins - and the RNA motifs they bind - have provided a wealth of information about the determinants of sequence recognition in protein-RNA complexes [4-6]. Data from both the PDB and HTP experiments have been exploited to develop several computational methods for predicting interfacial residues in protein-RNA complexes [reviewed in 7-10] as well as a few methods for predicting interaction partners in protein-RNA complexes and interaction networks [reviewed in 11-13]. …”