Sampling globally and locally correct RNA 3D structures using Ernwin, SPQR and experimental SAXS data.

The determination of the three-dimensional structure of large RNA macromolecules in solution is a challenging task that often requires the use of several experimental and computational techniques. Small-angle X-ray scattering can provide insight into some geometrical properties of the probed molecule, but this data must be properly interpreted in order to generate a three-dimensional model. Here, we propose a multiscale pipeline which introduces SAXS data into modelling the global shape of RNA in solution, which can be hierarchically refined until reaching atomistic precision in explicit solvent. The low-resolution helix model (Ernwin) deals with the exploration of the huge conformational space making use of the SAXS data, while a nucleotide-level model (SPQR) removes clashes and disentangles the proposed structures, leading the structure to an all-atom representation in explicit water. We apply the procedure on four different known pdb structures up to 159 nucleotides with promising results. Additionally, we predict an all-atom structure for the Plasmodium falceparum signal recognition particle ALU RNA based on SAXS data deposited in the SASBDB, which has an alternate conformation and better fit to the SAXS data than the previously published structure based on the same data but other modelling methods.

tRNA expression and modification landscapes, and their dynamics during zebrafish embryo development.

tRNA genes exist in multiple copies in the genome of all organisms across the three domains of life. Besides the sequence differences across tRNA copies, extensive post-transcriptional modification adds a further layer to tRNA diversification. Whilst the crucial role of tRNAs as adapter molecules in protein translation is well established, whether all tRNAs are actually expressed, and whether the differences across isodecoders play any regulatory role is only recently being uncovered. Here we built upon recent developments in the use of NGS-based methods for RNA modification detection and developed tRAM-seq, an experimental protocol and in silico analysis pipeline to investigate tRNA expression and modification. Using tRAM-seq, we analysed the full ensemble of nucleo-cytoplasmic and mitochondrial tRNAs during embryonic development of the model vertebrate zebrafish. We show that the repertoire of tRNAs changes during development, with an apparent major switch in tRNA isodecoder expression and modification profile taking place around the start of gastrulation. Taken together, our findings suggest the existence of a general reprogramming of the expressed tRNA pool, possibly gearing the translational machinery for distinct stages of the delicate and crucial process of embryo development.

The Multiscale Ernwin/SPQR RNA Structure Prediction Pipeline.

Aside from the well-known role in protein synthesis, RNA can perform catalytic, regulatory, and other essential biological functions which are determined by its three-dimensional structure. In this regard, a great effort has been made during the past decade to develop computational tools for the prediction of the structure of RNAs from the knowledge of their sequence, incorporating experimental data to refine or guide the modeling process. Nevertheless, this task can become exceptionally challenging when dealing with long noncoding RNAs, constituted by more than 200 nucleotides, due to their large size and the specific interactions involved. In this chapter, we describe a multiscale approach to predict such structures, incorporating SAXS experimental data into a hierarchical procedure which couples two coarse-grained representations: Ernwin, a helix-based approach, which deals with the global arrangement of secondary structure elements, and SPQR, a nucleotide-centered coarse-grained model, which corrects and refines the structures predicted at the coarser level.We describe the methodology through its application on the Braveheart long noncoding RNA, starting from the SAXS and secondary structure data to propose a refined, all-atom structure.

Comparative RNA Genomics.

Over the last quarter of a century it has become clear that RNA is much more than just a boring intermediate in protein expression. Ancient RNAs still appear in the core information metabolism and comprise a surprisingly large component in bacterial gene regulation. A common theme with these types of mostly small RNAs is their reliance of conserved secondary structures. Large-scale sequencing projects, on the other hand, have profoundly changed our understanding of eukaryotic genomes. Pervasively transcribed, they give rise to a plethora of large and evolutionarily extremely flexible non-coding RNAs that exert a vastly diverse array of molecule functions. In this chapter we provide a-necessarily incomplete-overview of the current state of comparative analysis of non-coding RNAs, emphasizing computational approaches as a means to gain a global picture of the modern RNA world.

3D feasibility of 2D RNA-RNA interaction paths by stepwise folding simulations.

The structure of an RNA, and even more so its interactions with other RNAs, provide valuable information about its function. Secondary structure-based tools for RNA-RNA interaction predictions provide a quick way to identify possible interaction targets and structures. However, these tools ignore the effect of steric hindrance on the tertiary (3D) structure level, and do not consider whether a suitable folding pathway exists to form the interaction. As a consequence, these tools often predict interactions that are unrealistically long and could be formed (in three dimensions) only by going through highly entangled intermediates. Here, we present a computational pipeline to assess whether a proposed secondary (2D) structure interaction is sterically feasible and reachable along a plausible folding pathway. To this end, we simulate the folding of a series of 3D structures along a given 2D folding path. To avoid the complexity of large-scale atomic resolution simulations, our pipeline uses coarse-grained 3D modeling and breaks up the folding path into small steps, each corresponding to the extension of the interaction by 1 or 2 bp. We apply our pipeline to analyze RNA-RNA interaction formation for three selected RNA-RNA complexes. We find that kissing hairpins, in contrast to interactions in the exterior loop, are difficult to extend and tend to get stuck at an interaction length of 6 bp. Our tool, including source code, documentation, and sample data, is available at www.github.com/irenekb/RRI-3D.

Modified RNAs and predictions with the ViennaRNA Package.

MOTIVATION: In living organisms, many RNA molecules are modified post-transcriptionally. This turns the widely known four-letter RNA alphabet ACGU into a much larger one with currently more than 300 known distinct modified bases. The roles for the majority of modified bases remain uncertain, but many are already well-known for their ability to influence the preferred structures that an RNA may adopt. In fact, tRNAs sometimes require certain modifications to fold into their cloverleaf shaped structure. However, predicting the structure of RNAs with base modifications is still difficult due to the lack of efficient algorithms that can deal with the extended sequence alphabet, as well as missing parameter sets that account for the changes in stability induced by the modified bases. RESULTS: We present an approach to include sparse energy parameter data for modified bases into the ViennaRNA Package. Our method does not require any changes to the underlying efficient algorithms but instead uses a set of plug-in constraints that adapt the predictions in terms of loop evaluation at runtime. These adaptations are efficient in the sense that they are only performed for loops where additional parameters are actually available for. In addition, our approach also facilitates the inclusion of more modified bases as soon as further parameters become available. AVAILABILITY AND IMPLEMENTATION: Source code and documentation are available at https://www.tbi.univie.ac.at/RNA.

Local RNA folding revisited.

Most of the functional RNA elements located within large transcripts are local. Local folding therefore serves a practically useful approximation to global structure prediction. Due to the sensitivity of RNA secondary structure prediction to the exact definition of sequence ends, accuracy can be increased by averaging local structure predictions over multiple, overlapping sequence windows. These averages can be computed efficiently by dynamic programming. Here we revisit the local folding problem, present a concise mathematical formalization that generalizes previous approaches and show that correct Boltzmann samples can be obtained by local stochastic backtracing in McCaskill's algorithms but not from local folding recursions. Corresponding new features are implemented in the ViennaRNA package to improve the support of local folding. Applications include the computation of maximum expected accuracy structures from RNAplfold data and a mutual information measure to quantify the sensitivity of individual sequence positions.

Mono-valent salt corrections for RNA secondary structures in the ViennaRNA package.

BACKGROUND: RNA features a highly negatively charged phosphate backbone that attracts a cloud of counter-ions that reduce the electrostatic repulsion in a concentration dependent manner. Ion concentrations thus have a large influence on folding and stability of RNA structures. Despite their well-documented effects, salt effects are not handled consistently by currently available secondary structure prediction algorithms. Combining Debye-Hückel potentials for line charges and Manning's counter-ion condensation theory, Einert et al. (Biophys J 100: 2745-2753, 2011) modeled the energetic contributions of monovalent cations on loops and helices. RESULTS: The model of Einert et al. is adapted to match the structure of the dynamic programming recursion of RNA secondary structure prediction algorithms. An empirical term describing the salt dependence of the duplex initiation energy is added to improve co-folding predictions for two or more RNA strands. The slightly modified model is implemented in the ViennaRNA package in such way that only the energy parameters but not the algorithmic structure is affected. A comparison with data from the literature show that predicted free energies and melting temperatures are in reasonable agreement with experiments. CONCLUSION: The new feature in the ViennaRNA package makes it possible to study effects of salt concentrations on RNA folding in a systematic manner. Strictly speaking, the model pertains only to mono-valent cations, and thus covers the most important parameter, i.e., the NaCl concentration. It remains a question for future research to what extent unspecific effects of bi- and tri-valent cations can be approximated in a similar manner. AVAILABILITY: Corrections for the concentration of monovalent cations are available in the ViennaRNA package starting from version 2.6.0.

Investigating RNA-RNA interactions through computational and biophysical analysis.

Numerous viruses utilize essential long-range RNA-RNA genome interactions, specifically flaviviruses. Using Japanese encephalitis virus (JEV) as a model system, we computationally predicted and then biophysically validated and characterized its long-range RNA-RNA genomic interaction. Using multiple RNA computation assessment programs, we determine the primary RNA-RNA interacting site among JEV isolates and numerous related viruses. Following in vitro transcription of RNA, we provide, for the first time, characterization of an RNA-RNA interaction using size-exclusion chromatography coupled with multi-angle light scattering and analytical ultracentrifugation. Next, we demonstrate that the 5' and 3' terminal regions of JEV interact with nM affinity using microscale thermophoresis, and this affinity is significantly reduced when the conserved cyclization sequence is not present. Furthermore, we perform computational kinetic analyses validating the cyclization sequence as the primary driver of this RNA-RNA interaction. Finally, we examined the 3D structure of the interaction using small-angle X-ray scattering, revealing a flexible yet stable interaction. This pathway can be adapted and utilized to study various viral and human long-non-coding RNA-RNA interactions and determine their binding affinities, a critical pharmacological property of designing potential therapeutics.

DrTransformer: heuristic cotranscriptional RNA folding using the nearest neighbor energy model.

MOTIVATION: Folding during transcription can have an important influence on the structure and function of RNA molecules, as regions closer to the 5' end can fold into metastable structures before potentially stronger interactions with the 3' end become available. Thermodynamic RNA folding models are not suitable to predict structures that result from cotranscriptional folding, as they can only calculate properties of the equilibrium distribution. Other software packages that simulate the kinetic process of RNA folding during transcription exist, but they are mostly applicable for short sequences. RESULTS: We present a new algorithm that tracks changes to the RNA secondary structure ensemble during transcription. At every transcription step, new representative local minima are identified, a neighborhood relation is defined and transition rates are estimated for kinetic simulations. After every simulation, a part of the ensemble is removed and the remainder is used to search for new representative structures. The presented algorithm is deterministic (up to numeric instabilities of simulations), fast (in comparison with existing methods), and it is capable of folding RNAs much longer than 200 nucleotides. AVAILABILITY AND IMPLEMENTATION: This software is open-source and available at https://github.com/ViennaRNA/drtransformer. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Caveats to Deep Learning Approaches to RNA Secondary Structure Prediction.

Machine learning (ML) and in particular deep learning techniques have gained popularity for predicting structures from biopolymer sequences. An interesting case is the prediction of RNA secondary structures, where well established biophysics based methods exist. The accuracy of these classical methods is limited due to lack of experimental parameters and certain simplifying assumptions and has seen little improvement over the last decade. This makes RNA folding an attractive target for machine learning and consequently several deep learning models have been proposed in recent years. However, for ML approaches to be competitive for de-novo structure prediction, the models must not just demonstrate good phenomenological fits, but be able to learn a (complex) biophysical model. In this contribution we discuss limitations of current approaches, in particular due to biases in the training data. Furthermore, we propose to study capabilities and limitations of ML models by first applying them on synthetic data (obtained from a simplified biophysical model) that can be generated in arbitrary amounts and where all biases can be controlled. We assume that a deep learning model that performs well on these synthetic, would also perform well on real data, and vice versa. We apply this idea by testing several ML models of varying complexity. Finally, we show that the best models are capable of capturing many, but not all, properties of RNA secondary structures. Most severely, the number of predicted base pairs scales quadratically with sequence length, even though a secondary structure can only accommodate a linear number of pairs.

Insights into the secondary and tertiary structure of the Bovine Viral Diarrhea Virus Internal Ribosome Entry Site.

The internal ribosome entry site (IRES) RNA of bovine viral diarrhoea virus (BVDV), an economically significant Pestivirus, is required for the cap-independent translation of viral genomic RNA. Thus, it is essential for viral replication and pathogenesis. We applied a combination of high-throughput biochemical RNA structure probing (SHAPE-MaP) and in silico modelling approaches to gain insight into the secondary and tertiary structures of BVDV IRES RNA. Our study demonstrated that BVDV IRES RNA in solution forms a modular architecture composed of three distinct structural domains (I-III). Two regions within domain III are represented in tertiary interactions to form an H-type pseudoknot. Computational modelling of the pseudoknot motif provided a fine-grained picture of the tertiary structure and local arrangement of helices in the BVDV IRES. Furthermore, comparative genomics and consensus structure predictions revealed that the pseudoknot is evolutionarily conserved among many Pestivirus species. These studies provide detailed insight into the structural arrangement of BVDV IRES RNA H-type pseudoknot and encompassing motifs that likely contribute to the optimal functionality of viral cap-independent translation element.

RNAxplorer: harnessing the power of guiding potentials to sample RNA landscapes.

MOTIVATION: Predicting the folding dynamics of RNAs is a computationally difficult problem, first and foremost due to the combinatorial explosion of alternative structures in the folding space. Abstractions are therefore needed to simplify downstream analyses, and thus make them computationally tractable. This can be achieved by various structure sampling algorithms. However, current sampling methods are still time consuming and frequently fail to represent key elements of the folding space. METHOD: We introduce RNAxplorer, a novel adaptive sampling method to efficiently explore the structure space of RNAs. RNAxplorer uses dynamic programming to perform an efficient Boltzmann sampling in the presence of guiding potentials, which are accumulated into pseudo-energy terms and reflect similarity to already well-sampled structures. This way, we effectively steer sampling toward underrepresented or unexplored regions of the structure space. RESULTS: We developed and applied different measures to benchmark our sampling methods against its competitors. Most of the measures show that RNAxplorer produces more diverse structure samples, yields rare conformations that may be inaccessible to other sampling methods and is better at finding the most relevant kinetic traps in the landscape. Thus, it produces a more representative coarse graining of the landscape, which is well suited to subsequently compute better approximations of RNA folding kinetics. AVAILABILITYAND IMPLEMENTATION: https://github.com/ViennaRNA/RNAxplorer/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Zinc-finger protein CNBP alters the 3-D structure of lncRNA Braveheart in solution.

Long non-coding RNAs (lncRNAs) constitute a significant fraction of the transcriptome, playing important roles in development and disease. However, our understanding of structure-function relationships for this emerging class of RNAs has been limited to secondary structures. Here, we report the 3-D atomistic structural study of epigenetic lncRNA, Braveheart (Bvht), and its complex with CNBP (Cellular Nucleic acid Binding Protein). Using small angle X-ray scattering (SAXS), we elucidate the ensemble of Bvht RNA conformations in solution, revealing that Bvht lncRNA has a well-defined, albeit flexible 3-D structure that is remodeled upon CNBP binding. Our study suggests that CNBP binding requires multiple domains of Bvht and the RHT/AGIL RNA motif. We show that RHT/AGIL, previously shown to interact with CNBP, contains a highly flexible loop surrounded by more ordered helices. As one of the largest RNA-only 3-D studies, the work lays the foundation for future structural studies of lncRNA-protein complexes.

IntaRNAhelix-composing RNA-RNA interactions from stable inter-molecular helices boosts bacterial sRNA target prediction.

Efficient computational tools for the identification of putative target RNAs regulated by prokaryotic sRNAs rely on thermodynamic models of RNA secondary structures. While they typically predict RNA-RNA interaction complexes accurately, they yield many highly-ranked false positives in target screens. One obvious source of this low specificity appears to be the disability of current secondary-structure-based models to reflect steric constraints, which nevertheless govern the kinetic formation of RNA-RNA interactions. For example, often - even thermodynamically favorable - extensions of short initial kissing hairpin interactions are kinetically prohibited, since this would require unwinding of intra-molecular helices as well as sterically impossible bending of the interaction helix. Another source is the consideration of instable and thus unlikely subinteractions that enable better scoring of longer interactions. In consequence, the efficient prediction methods that do not consider such effects show a high false positive rate. To increase the prediction accuracy we devise IntaRNAhelix, a dynamic programming algorithm that length-restricts the runs of consecutive inter-molecular base pairs (perfect canonical stackings), which we hypothesize to implicitly model the steric and kinetic effects. The novel method is implemented by extending the state-of-the-art tool IntaRNA. Our comprehensive bacterial sRNA target prediction benchmark demonstrates significant improvements of the prediction accuracy and enables more than 40-times faster computations. These results indicate - supporting our hypothesis - that stable helix composition increases the accuracy of interaction prediction models compared to the current state-of-the-art approach.

Updated Phylogeny of Chikungunya Virus Suggests Lineage-Specific RNA Architecture.

Chikungunya virus (CHIKV), a mosquito-borne alphavirus of the family Togaviridae, has recently emerged in the Americas from lineages from two continents: Asia and Africa. Historically, CHIKV circulated as at least four lineages worldwide with both enzootic and epidemic transmission cycles. To understand the recent patterns of emergence and the current status of the CHIKV spread, updated analyses of the viral genetic data and metadata are needed. Here, we performed phylogenetic and comparative genomics screens of CHIKV genomes, taking advantage of the public availability of many recently sequenced isolates. Based on these new data and analyses, we derive a revised phylogeny from nucleotide sequences in coding regions. Using this phylogeny, we uncover the presence of several distinct lineages in Africa that were previously considered a single one. In parallel, we performed thermodynamic modeling of CHIKV untranslated regions (UTRs), which revealed evolutionarily conserved structured and unstructured RNA elements in the 3'UTR. We provide evidence for duplication events in recently emerged American isolates of the Asian CHIKV lineage and propose the existence of a flexible 3'UTR architecture among different CHIKV lineages.

3D based on 2D: Calculating helix angles and stacking patterns using forgi 2.0, an RNA Python library centered on secondary structure elements.

We present forgi, a Python library to analyze the tertiary structure of RNA secondary structure elements. Our representation of an RNA molecule is centered on secondary structure elements (stems, bulges and loops). By fitting a cylinder to the helix axis, these elements are carried over into a coarse-grained 3D structure representation. Integration with Biopython allows for handling of all-atom 3D information. forgi can deal with a variety of file formats including dotbracket strings, PDB and MMCIF files. We can handle modified residues, missing residues, cofold and multifold structures as well as nucleotide numbers starting at arbitrary positions. We apply this library to the study of stacking helices in junctions and pseudo knots and investigate how far stacking helices in solved experimental structures can divert from coaxial geometries.

Functional RNA Structures in the 3'UTR of Tick-Borne, Insect-Specific and No-Known-Vector Flaviviruses.

Untranslated regions (UTRs) of flaviviruses contain a large number of RNA structural elements involved in mediating the viral life cycle, including cyclisation, replication, and encapsidation. Here we report on a comparative genomics approach to characterize evolutionarily conserved RNAs in the 3 ' UTR of tick-borne, insect-specific and no-known-vector flaviviruses in silico. Our data support the wide distribution of previously experimentally characterized exoribonuclease resistant RNAs (xrRNAs) within tick-borne and no-known-vector flaviviruses and provide evidence for the existence of a cascade of duplicated RNA structures within insect-specific flaviviruses. On a broader scale, our findings indicate that viral 3 ' UTRs represent a flexible scaffold for evolution to come up with novel xrRNAs.

Control of Cognate Sense mRNA Translation by cis-Natural Antisense RNAs.

Cis-Natural Antisense Transcripts (cis-NATs), which overlap protein coding genes and are transcribed from the opposite DNA strand, constitute an important group of noncoding RNAs. Whereas several examples of cis-NATs regulating the expression of their cognate sense gene are known, most cis-NATs function by altering the steady-state level or structure of mRNA via changes in transcription, mRNA stability, or splicing, and very few cases involve the regulation of sense mRNA translation. This study was designed to systematically search for cis-NATs influencing cognate sense mRNA translation in Arabidopsis (Arabidopsis thaliana). Establishment of a pipeline relying on sequencing of total polyA+ and polysomal RNA from Arabidopsis grown under various conditions (i.e. nutrient deprivation and phytohormone treatments) allowed the identification of 14 cis-NATs whose expression correlated either positively or negatively with cognate sense mRNA translation. With use of a combination of cis-NAT stable over-expression in transgenic plants and transient expression in protoplasts, the impact of cis-NAT expression on mRNA translation was confirmed for 4 out of 5 tested cis-NAT:sense mRNA pairs. These results expand the number of cis-NATs known to regulate cognate sense mRNA translation and provide a foundation for future studies of their mode of action. Moreover, this study highlights the role of this class of noncoding RNAs in translation regulation.

RNA modifications in structure prediction - Status quo and future challenges.

Chemical modifications of RNA nucleotides change their identity and characteristics and thus alter genetic and structural information encoded in the genomic DNA. tRNA and rRNA are probably the most heavily modified genes, and often depend on derivatization or isomerization of their nucleobases in order to correctly fold into their functional structures. Recent RNomics studies, however, report transcriptome wide RNA modification and suggest a more general regulation of structuredness of RNAs by this so called epitranscriptome. Modification seems to require specific substrate structures, which in turn are stabilized or destabilized and thus promote or inhibit refolding events of regulatory RNA structures. In this review, we revisit RNA modifications and the related structures from a computational point of view. We discuss known substrate structures, their properties such as sub-motifs as well as consequences of modifications on base pairing patterns and possible refolding events. Given that efficient RNA structure prediction methods for canonical base pairs have been established several decades ago, we review to what extend these methods allow the inclusion of modified nucleotides to model and study epitranscriptomic effects on RNA structures.