Protein-specific prediction of mRNA binding using RNA sequences, binding motifs and predicted secondary structures

Abstract

BackgroundRNA-binding proteins interact with specific RNA molecules to regulate important cellular processes. It is therefore necessary to identify the RNA interaction partners in order to understand the precise functions of such proteins. Protein-RNA interactions are typically characterized using in vivo and in vitro experiments but these may not detect all binding partners. Therefore, computational methods that capture the protein-dependent nature of such binding interactions could help to predict potential binding partners in silico.ResultsWe have developed three methods to predict whether an RNA can interact with a particular RNA-binding protein using support vector machines and different features based on the sequence (the Oli method), the motif score (the OliMo method) and the secondary structure (the OliMoSS method). We applied these approaches to different experimentally-derived datasets and compared the predictions with RNAcontext and RPISeq. Oli outperformed OliMoSS and RPISeq, confirming our protein-specific predictions and suggesting that tetranucleotide frequencies are appropriate discriminative features. Oli and RNAcontext were the most competitive methods in terms of the area under curve. A precision-recall curve analysis achieved higher precision values for Oli. On a second experimental dataset including real negative binding information, Oli outperformed RNAcontext with a precision of 0.73 vs. 0.59.ConclusionsOur experiments showed that features based on primary sequence information are sufficiently discriminating to predict specific RNA-protein interactions. Sequence motifs and secondary structure information were not necessary to improve these predictions. Finally we confirmed that protein-specific experimental data concerning RNA-protein interactions are valuable sources of information that can be used for the efficient training of models for in silico predictions. The scripts are available upon request to the corresponding author.

RNA‐protein interactions in an unstructured context

“…Most modern methods for predicting RNA–protein binding from sequence information, physicochemical properties of individual RNA and protein building blocks, or global RNA–protein characteristics are based on machine learning strategies. In particular, predictive models are trained on the known interactors by using features such as composition, hydrophobicity, and evolutionary information . An early such approach, developed by Pancaldi and Bahler, is based on support vector machine (SVM) and random forest (RF) formalisms and uses the known RNA and protein features as predictors, including inter alia GO terms, protein localization, and chromosome position information .…”

“…Although they tackle an important and difficult problem, the above methods still frequently suffer from a limited accuracy, a lack of general applicability, and the fact that only a few of them are fundamentally steeped in basic physicochemical principles. The latter criticism is probably the most important downside of the machine learning approaches: they frequently do not allow for a deeper insight into the physicochemical underpinnings of RNA–protein interactions.…”

“…An improvement in our understanding of the RNA–protein interactions in the unstructured context, including the ability to predict binding sites and relative affinities, would represent a major step forward from both fundamental and practical perspectives. Currently, most of our knowledge of RNA–protein interactions concerns structured RNA‐binding motifs and, consequently, computational methods for predicting RNA‐protein interactions in a physicochemically and structurally realistic manner are necessarily limited and biased by such knowledge []. Similarly problematic, the sequence‐based computational approaches for predicting RNA–protein interactions typically involve machine learning strategies and only indirectly rely on or help further understand the microscopic physicochemical principles behind such interactions .…”

“…Currently, most of our knowledge of RNA–protein interactions concerns structured RNA‐binding motifs and, consequently, computational methods for predicting RNA‐protein interactions in a physicochemically and structurally realistic manner are necessarily limited and biased by such knowledge []. Similarly problematic, the sequence‐based computational approaches for predicting RNA–protein interactions typically involve machine learning strategies and only indirectly rely on or help further understand the microscopic physicochemical principles behind such interactions . While experimental approaches toward enumerating and characterizing RNA interactions with IDPs are making significant advances [], they are limited by the fact that the microscopic, high‐resolution features of such complexes are largely beyond the reach of modern structural biology methods.…”

Prediction of RNA-protein sequence and structure binding preferences using deep convolutional and recurrent neural networks

“…There are many computational methods developed for predicting RNA-protein binding sites [1,25,26,35]. In this study, we compare iDeepS with the state-ofthe-art sequence-based methods DeepBind [1], Oli [25], iONMF [35] and Graph-Prot [26]. DeepBind, uses a sequence CNN with the same architecture as iDeepS to predict RBP binding sites.…”