RNA Primary Sequence Research Articles

Ac4C-AFL: A high-precision identification of human mRNA N4-acetylcytidine sites based on adaptive feature representation learning

RNA N4-acetylcytidine (ac4C) is a highly conserved RNA modification that plays a crucial role in controlling mRNA stability, processing, and translation. Consequently, accurate identification of ac4C sites across the genome is critical for understanding gene expression regulation mechanisms. In this study, we have developed ac4C-AFL, a bioinformatics tool that precisely identifies ac4C sites from primary RNA sequences. In ac4C-AFL, we identified the optimal sequence length for model building and implemented an adaptive feature representation strategy that is capable of extracting the most representative features from RNA. To identify the most relevant features, we proposed a novel ensemble feature importance scoring strategy to rank features effectively. We then used this information to conduct the sequential forward search, which individually determine the optimal feature set from the 16 sequence-derived feature descriptors. Utilizing these optimal feature descriptors, we constructed 176 baseline models using 11 popular classifiers. The most efficient baseline models were identified using the two-step feature selection approach, whose predicted scores were integrated and trained with the appropriate classifier to develop the final prediction model. Our rigorous cross-validations and independent tests demonstrate that ac4C-AFL surpasses contemporary tools in predicting ac4C sites. Moreover, we have developed a publicly accessible web server at https://balalab-skku.org/ac4C-AFL/.

Revealing the grammar of small RNA secretion using interpretable machine learning

Small non-coding RNAs can be secreted through a variety of mechanisms, including exosomal sorting, in small extracellular vesicles, and within lipoprotein complexes. However, the mechanisms that govern their sorting and secretion are not well understood. Here, we present ExoGRU, a machine learning model that predicts small RNA secretion probabilities from primary RNA sequences. We experimentally validated the performance of this model through ExoGRU-guided mutagenesis and synthetic RNA sequence analysis. Additionally, we used ExoGRU to reveal cis and trans factors that underlie small RNA secretion, including known and novel RNA-binding proteins (RBPs), e.g., YBX1, HNRNPA2B1, and RBM24. We also developed a novel technique called exoCLIP, which reveals the RNA interactome of RBPs within the cell-free space. Together, our results demonstrate the power of machine learning in revealing novel biological mechanisms. In addition to providing deeper insight into small RNA secretion, this knowledge can be leveraged in therapeutic and synthetic biology applications.

Detection, Clinical Application, and Manipulation of RNA Modifications.

ConspectusWith increasing research interest, more than 170 types of chemical modifications of RNA have been characterized. The epigenetic modifications of RNA do not alter the primary sequence of RNA but modulate the gene activity. Increasing numbers of regulatory functions of these RNA modifications, particularly in controlling mRNA fate and gene expression, are being discovered. To gain a deeper understanding of the biological significance and clinical prospects of RNA modifications, the development of innovative labeling and detection methodologies is of great importance. Owing to the dynamic features of RNA modifications and the fact that only a portion of genes are modified, detection methods should accurately reveal the precise distribution and modification level of RNA modifications. In general, detection methodologies identify specific RNA modifications in two ways: (1) enriching modification-containing RNAs; and (2) altering the Watson-Crick base pairing pattern to produce truncation or mutation signatures. Additionally, it is important to develop flexible and accurate manipulation tools that enable the installation or removal of RNA modifications at specific positions to investigate the biological functions of a single site. With the development of detection and manipulation methods, the scientific understanding of the biological functions of RNA modifications has increased, paving the way for applications of RNA modifications in disease diagnosis and treatments.In this Account, we provide a brief summary of recent efforts to develop methodologies for detecting RNA modifications. Through the evolution of these detection techniques, our team has uncovered the potential biological roles of RNA modifications in diseases such as diabetic cardiovascular complications, viral infections, and hematologic malignancies. We mainly summarize the recently developed strategies for manipulating RNA modifications. The advent of these programmable editing tools allows for the precise installation or removal of RNA modifications at specific positions. As a result, the biological functions of RNA modifications at these specific loci could be identified, further advancing our knowledge in this field.With this Account, we anticipate providing chemical and biological researchers with comprehensive strategies to discover the underlying mechanisms of RNA modification-mediated biological processes. Although the field of RNA modifications has undergone rapid progress in recent years, our understanding of most of these RNA modifications remains incomplete. We hope to inspire efforts to expand the toolbox for investigating RNA modifications and promote translational research on epigenetics in clinical diagnosis and treatment.

Self-attention enabled deep learning of dihydrouridine (D) modification on mRNAs unveiled a distinct sequence signature from tRNAs

Dihydrouridine (D)is a modified pyrimidine nucleotide universally found in viral, prokaryotic, and eukaryotic species. It serves as a metabolic modulator for various pathological conditions, and its elevated levels in tumors are associated with a series of cancers. Precise identification of D sites on RNA is vital for understanding its biological function. A number of computational approaches have been developed for predicting D sites on tRNAs; however, none have considered mRNAs. We present here DPred, the first computational tool for predicting D on mRNAs in yeast from the primary RNA sequences. Built on a local self-attention layer and a convolutional neural network (CNN) layer, the proposed deep learning model outperformed classic machine learning approaches (random forest, support vector machines, etc.) and achieved reasonable accuracy and reliability with areas under the curve of 0.9166 and 0.9027 in jackknife cross-validation and on an independent testing dataset, respectively. Importantly, we showed that distinct sequence signatures are associated with the D sites on mRNAs and tRNAs, implying potentially different formation mechanisms and putative divergent functionality of this modification on the two types of RNA. DPred is available as a user-friendly Web server.

Prediction and Motif Analysis of 2’-O-methylation Using a Hybrid Deep Learning Model from RNA Primary Sequence and Nanopore Signals

Background: 2’-O-Methylation (2’-O-Me) is a post-transcriptional RNA modification that occurs in the ribose sugar moiety of all four nucleotides and is abundant in both coding and non-coding RNAs. Accurate prediction of each subtype of 2’-O-Me (Am, Cm, Gm, Um) helps understand their role in RNA metabolism and function. Objective: This study aims to build models that can predict each subtype of 2’-O-Me from RNA sequence and nanopore signals and exploit the model interpretability for sequence motif mining. Methods: We first propose a novel deep learning model DeepNm to better capture the sequence features of each subtype with a multi-scale framework. Based on DeepNm, we continue to propose HybridNm, which combines sequences and nanopore signals through a dual-path framework. The nanopore signalderived features are first passed through a convolutional layer and then merged with sequence features extracted from different scales for final classification. Results: A 5-fold cross-validation process on Nm-seq data shows that DeepNm outperforms two stateof- the-art 2’-O-Me predictors. After incorporating nanopore signal-derived features, HybridNm further achieved significant improvements. Through model interpretation, we identified not only subtypespecific motifs but also revealed shared motifs between subtypes. In addition, Cm, Gm, and Um shared motifs with the well-studied m6A RNA methylation, suggesting a potential interplay among different RNA modifications and the complex nature of epitranscriptome regulation. Conclusion: The proposed frameworks can be useful tools to predict 2’-O-Me subtypes accurately and reveal specific sequence patterns.

Primary sequence-assisted prediction of m6A RNA methylation sites from Oxford nanopore direct RNA sequencing data.

Traditional epitranscriptome profiling approach relies on specific antibodies or chemical treatments to capture modified RNA molecules and then applies high throughput sequencing to identify their transcriptomic locations. However, due to the lack of suitable or high-quality antibodies, only a small proportion of the 170 documented RNA modifications were profiled with those approaches. Direct sequencing of native RNA molecules using Oxford Nanopore Technologies (ONT) enabled straight inspection of RNA modifications and offered a promising alternative solution. N6-methyladenosine (m6A) is known to cause characteristic changes and increased base call errors of ONT signals compared with non-modified adenosines, based on which, the m6A sites can be identified directly from ONT signals. Meanwhile, a number of studies have shown that it is possible to predict m6A sites from RNA primary sequences. Using the m6A sites revealed by Illumina technology as benchmark, we showed that, the accuracy of ONT-based m6A site prediction can be further increased by integrating additional information from the primary sequences of RNA (AUROC of 0.918), compared with using ONT signals only (AUROC 0.878 using Base call error features, and 0.804 using Tombo features), providing a new perspective for more reliable mining of the relatively noisy ONT signals.

WHISTLE server: A high-accuracy genomic coordinate-based machine learning platform for RNA modification prediction.

The primary sequences of DNA, RNA and protein have been used as the dominant information source of existing machine learning tools, especially for contexts not fully explored by wet-experimental approaches. Since molecular markers are profoundly orchestrated in the living organisms, those markers that cannot be unambiguously recovered from the primary sequence often help to predict other biological events. To the best of our knowledge, there is no current tool to build and deploy machine learning models that consider genomic evidence. We therefore developed the WHISTLE server, the first machine learning platform based on genomic coordinates. It features convenient covariate extraction and model web deployment with 46 distinct genomic features integrated along with the conventional sequence features. We showed that, when predicting m6A sites from SRAMP project, the model integrating genomic features substantially outperformed those based on only sequence features. The WHISTLE server should be a useful tool for studying biological attributes specifically associated with genomic coordinates, and is freely accessible at: www.xjtlu.edu.cn/biologicalsciences/whi2.

Inhibition of SARS-CoV-2 by Targeting Conserved Viral RNA Structures and Sequences.

The ongoing COVID-19/Severe Acute Respiratory Syndrome CoV-2 (SARS-CoV-2) pandemic has become a significant threat to public health and has hugely impacted societies globally. Targeting conserved SARS-CoV-2 RNA structures and sequences essential for viral genome translation is a novel approach to inhibit viral infection and progression. This new pharmacological modality compasses two classes of RNA-targeting molecules: 1) synthetic small molecules that recognize secondary or tertiary RNA structures and 2) antisense oligonucleotides (ASOs) that recognize the RNA primary sequence. These molecules can also serve as a “bait” fragment in RNA degrading chimeras to eliminate the viral RNA genome. This new type of chimeric RNA degrader is recently named ribonuclease targeting chimera or RIBOTAC. This review paper summarizes the sequence conservation in SARS-CoV-2 and the current development of RNA-targeting molecules to combat this virus. These RNA-binding molecules will also serve as an emerging class of antiviral drug candidates that might pivot to address future viral outbreaks.

The HIV 5' Gag Region Displays a Specific Nucleotide Bias Regulating Viral Splicing and Infectivity.

Alternative splicing and the expression of intron-containing mRNAs is one hallmark of HIV gene expression. To facilitate the otherwise hampered nuclear export of non-fully processed mRNAs, HIV encodes the Rev protein, which recognizes its intronic response element and fuels the HIV RNAs into the CRM-1-dependent nuclear protein export pathway. Both alternative splicing and Rev-dependency are regulated by the primary HIV RNA sequence. Here, we show that these processes are extremely sensitive to sequence alterations in the 5’coding region of the HIV genomic RNA. Increasing the GC content by insertion of either GFP or silent mutations activates a cryptic splice donor site in gag, entirely deregulates the viral splicing pattern, and lowers infectivity. Interestingly, an adaptation of the inserted GFP sequence toward an HIV-like nucleotide bias reversed these phenotypes completely. Of note, the adaptation yielded completely different primary sequences although encoding the same amino acids. Thus, the phenotypes solely depend on the nucleotide composition of the two GFP versions. This is a strong indication of an HIV-specific mRNP code in the 5′ gag region wherein the primary RNA sequence bias creates motifs for RNA-binding proteins and controls the fate of the HIV-RNA in terms of viral gene expression and infectivity.

M6AmPred: Identifying RNA N6, 2'-O-dimethyladenosine (m6Am) sites based on sequence-derived information.

N6,2'-O-dimethyladenosine (m6Am) is a reversible modification widely occurred on varied RNA molecules. The biological function of m6Am is yet to be known though recent studies have revealed its influences in cellular mRNA fate. Precise identification of m6Am sites on RNA is vital for the understanding of its biological functions. We present here m6AmPred, the first web server for in silico identification of m6Am sites from the primary sequences of RNA. Built upon the eXtreme Gradient Boosting with Dart algorithm (XgbDart) and EIIP-PseEIIP encoding scheme, m6AmPred achieved promising prediction performance with the AUCs greater than 0.954 when tested by 10-fold cross-validation and independent testing datasets. To critically test and validate the performance of m6AmPred, the experimentally verified m6Am sites from two data sources were cross-validated. The m6AmPred web server is freely accessible at: https://www.xjtlu.edu.cn/biologicalsciences/m6am, and it should make a useful tool for the researchers who are interested in N6,2'-O-dimethyladenosine RNA modification.

Deep neural networks for inferring binding sites of RNA-binding proteins by using distributed representations of RNA primary sequence and secondary structure

BackgroundRNA binding proteins (RBPs) play a vital role in post-transcriptional processes in all eukaryotes, such as splicing regulation, mRNA transport, and modulation of mRNA translation and decay. The identification of RBP binding sites is a crucial step in understanding the biological mechanism of post-transcriptional gene regulation. However, the determination of RBP binding sites on a large scale is a challenging task due to high cost of biochemical assays. Quite a number of studies have exploited machine learning methods to predict binding sites. Especially, deep learning is increasingly used in the bioinformatics field by virtue of its ability to learn generalized representations from DNA and protein sequences.ResultsIn this paper, we implemented a novel deep neural network model, DeepRKE, which combines primary RNA sequence and secondary structure information to effectively predict RBP binding sites. Specifically, we used word embedding algorithm to extract features of RNA sequences and secondary structures, i.e., distributed representation of k-mers sequence rather than traditional one-hot encoding. The distributed representations are taken as input of convolutional neural networks (CNN) and bidirectional long-term short-term memory networks (BiLSTM) to identify RBP binding sites. Our results show that deepRKE outperforms existing counterpart methods on two large-scale benchmark datasets.ConclusionsOur extensive experimental results show that DeepRKE is an efficacious tool for predicting RBP binding sites. The distributed representations of RNA sequences and secondary structures can effectively detect the latent relationship and similarity between k-mers, and thus improve the predictive performance. The source code of DeepRKE is available at https://github.com/youzhiliu/DeepRKE/.

M5UPred: A Web Server for the Prediction of RNA 5-Methyluridine Sites from Sequences

As one of the widely occurring RNA modifications, 5-methyluridine (m5U) has recently been shown to play critical roles in various biological functions and disease pathogenesis, such as under stress response and during breast cancer development. Precise identification of m5U sites on RNA is vital for the understanding of the regulatory mechanisms of RNA life. We present here m5UPred, the first web server for in silico identification of m5U sites from the primary sequences of RNA. Built upon the support vector machine (SVM) algorithm and the biochemical encoding scheme, m5UPred achieved reasonable prediction performance with the area under the receiver operating characteristic curve (AUC) greater than 0.954 by 5-fold cross-validation and independent testing datasets. To critically test and validate the performance of our newly proposed predictor, the experimentally validated m5U sites were further separated by high-throughput sequencing techniques (miCLIP-Seq and FICC-Seq) and cell types (HEK293 and HAP1). When tested on cross-technique and cross-cell-type validation using independent datasets, m5UPred achieved an average AUC of 0.922 and 0.926 under mature mRNA mode, respectively, showing reasonable accuracy and reliability. The m5UPred web server is freely accessible now and it should make a useful tool for the researchers who are interested in m5U RNA modification.

HPC-REDItools: a novel HPC-aware tool for improved large scale RNA-editing analysis

BackgroundRNA editing is a widespread co-/post-transcriptional mechanism that alters primary RNA sequences through the modification of specific nucleotides and it can increase both the transcriptome and proteome diversity. The automatic detection of RNA-editing from RNA-seq data is computational intensive and limited to small data sets, thus preventing a reliable genome-wide characterisation of such process.ResultsIn this work we introduce HPC-REDItools, an upgraded tool for accurate RNA-editing events discovery from large dataset repositories. Availability: https://github.com/BioinfoUNIBA/REDItools2.ConclusionsHPC-REDItools is dramatically faster than the previous version, REDItools, enabling big-data analysis by means of a MPI-based implementation and scaling almost linearly with the number of available cores.

Emergence of New Disease: How Can Artificial Intelligence Help?

Emergence of new disease remains a critical parameter in human health and society. Advances in artificial intelligence (AI) allow for rapid processing and analysis of massive and complex data. In this forum article, the recent applications across disease prediction and drug development in relation to the COVID-19 pandemic are reviewed.

IRNA5hmC: The First Predictor to Identify RNA 5-Hydroxymethylcytosine Modifications Using Machine Learning.

RNA 5-hydroxymethylcytosine (5hmC) modification plays an important role in a series of biological processes. Characterization of its distributions in transcriptome is fundamentally important to reveal the biological functions of 5hmC. Sequencing-based technologies allow the high-throughput identification of 5hmC; however, they are labor-intensive, time-consuming, as well as expensive. Thus, there is an urgent need to develop more effective and efficient computational methods, at least complementary to the high-throughput technologies. In this study, we developed iRNA5hmC, a computational predictive protocol to identify RNA 5hmC sites using machine learning. In this predictor, we introduced a sequence-based feature algorithm consisting of two feature representations, (1) k-mer spectrum and (2) positional nucleotide binary vector, to capture the sequential characteristics of 5hmC sites. Afterward, we utilized a two-stage feature space optimization strategy to improve the feature representation ability, and trained a predictive model using support vector machine (SVM). Our feature analysis results showed that feature optimization can help to capture the most discriminative features. As compared to well-known existing feature descriptors, our proposed representations can more accurately separate true 5hmC from non-5hmC sites. To the best of our knowledge, iRNA5hmC is the first RNA 5hmC predictor that enables to make predictions based on RNA primary sequences only, without any need of prior experimental knowledge. Importantly, we have established an easy-to-use webserver which is currently available at http://server.malab.cn/iRNA5hmC. We expect it has potential to be a useful tool for the prediction of 5hmC sites.

New opportunities for designing effective small interfering RNAs

Small interfering RNAs (siRNAs) that silence genes of infectious diseases are potentially potent drugs. A continuing obstacle for siRNA-based drugs is how to improve their efficacy for adequate dosage. To overcome this obstacle, the interactions of antiviral siRNAs, tested in vivo, were computationally examined within the RNA-induced silencing complex (RISC). Thermodynamics data show that a persistent RISC cofactor is significantly more exothermic for effective antiviral siRNAs than their ineffective counterparts. Detailed inspection of viral RNA secondary structures reveals that effective antiviral siRNAs target hairpin or pseudoknot loops. These structures are critical for initial RISC interactions since they partially lack intramolecular complementary base pairing. Importing two temporary RISC cofactors from magnesium-rich hairpins and/or pseudoknots then kickstarts full RNA hybridization and hydrolysis. Current siRNA design guidelines are based on RNA primary sequence data. Herein, the thermodynamics of RISC cofactors and targeting magnesium-rich RNA secondary structures provide additional guidelines for improving siRNA design.

A Polymer Physics Framework for the Entropy of Arbitrary Pseudoknots

The accurate prediction of RNA secondary structure from primary sequence has had enormous impact on research from the past 40 years. Although many algorithms are available to make these predictions, the inclusion of non-nested loops, termed pseudoknots, still poses challenges arising from two main factors: 1) no physical model exists to estimate the loop entropies of complex intramolecular pseudoknots, and 2) their NP-complete enumeration has impeded their study. Here, we address both challenges. First, we develop a polymer physics model that can address arbitrarily complex pseudoknots using only two parameters corresponding to concrete physical quantities—over an order of magnitude fewer than the sparsest state-of-the-art phenomenological methods. Second, by coupling this model to exhaustive enumeration of the set of possible structures, we compute the entire free energy landscape of secondary structures resulting from a primary RNA sequence. We demonstrate that for RNA structures of ∼80 nucleotides, with minimal heuristics, the complete enumeration of possible secondary structures can be accomplished quickly despite the NP-complete nature of the problem. We further show that despite our loop entropy model’s parametric sparsity, it performs better than or on par with previously published methods in predicting both pseudoknotted and non-pseudoknotted structures on a benchmark data set of RNA structures of ≤80 nucleotides. We suggest ways in which the accuracy of the model can be further improved.

Chemical reaction optimization for RNA structure prediction

RNA Structure Prediction (RSP) is an optimization problem, where a stable secondary structure is obtained from an RNA primary sequence. To solve the RSP problem, many exact and metaheuristic algorithms were established in recent years. We have proposed an approach based on metaheuristic algorithm named Chemical Reaction Optimization (CRO) to solve the RSP problem. CRO is a population-based metaheuristic which has been employed in different optimization problems and works better than all other related existing algorithms. We have redesigned the reaction operators of CRO algorithm and calculated the minimum free energy of the RNA structure to solve RSP problem. The operators spread out the population entirely on the solution space using both local and global searches and find the better structure, which makes the proposed algorithm more unique. We have designed a novel operator called Repair function to verify and remove the repeated stem from the solution of an RNA sequence, which makes the process more time efficient. Both the quality of solutions and execution time are considered in designing the basic operators and the repair function. Thus, the proposed methodology gives robustness, efficiency, and effectiveness in solving the problem. The results of the proposed CRO based algorithm for RSP problem are compared with genetic algorithm (RNAPredict), simulated annealing algorithm (SARNA-Predict), coincidence algorithm (COIN), two-level particle swarm optimization algorithm (TL-PSOfold) and Changing Range Bat Algorithm (CRBA) to present that, the proposed work gives better results than those. The significance testing using Kruskal-Wallis test followed by post-hoc analysis also proves that the proposed work outperforms the five related methods.

Determination of secondary and tertiary structures of cervical cancer lncRNA diagnostic and siRNA therapeutic biomarkers

Cervical cancer is one of the primary causes of mortality in women due to human papilloma virus (HPV) infection. The fingerprint of an HPV infection could be detected using a long non-coding RNA (lncRNA) biomarker, enabling it to be utilized in molecular diagnostics. The primary structure or sequences of RNA should be annotated within conventional bioinformatics tools. Therefore, this study aimed to determine the fine-grained 2D and 3D structures of lncRNA PVT1 and its respective siRNA inhibitors. lncRNA PVT1 sequences from Homo sapiens, Mus musculus, and Rattus norvegicus were retrieved from Genbank (NCBI). Prediction of the 2D structure and analysis of the interactions of the lncRNA and siRNA were performed using the Vienna RNA package. The 3D structure of the RNA was computed using the SimRNA and ModeRNA software programs. The results showed that lncRNA PVT1 from H. sapiens and M. musculus had a conserved region. However, the lncRNA from both H. sapiens and M. musculus showed a low conserved region, and the 2D structure could not be determined; thus, the annotation and 2D model focused only on H. sapiens. Both of their lncRNA PVT1 also had a short half-life in the cell. Based on the 3D modeling pipeline, the 3D model of lncRNA PVT1 showed the stability and possible function as molecules, while the PVT1 siRNA-lncRNA interaction analysis revealed that the molecules could bind well. Based on these findings, the structures of both lncRNA PVT1 and its siRNA have the potential to be utilized as biomarkers.

Functional characterization of the bovine foamy virus miRNA expression cassette and its dumbbell-shaped pri-miRNA.

Foamy viruses are unconventional and complex retroviruses distinct from the other members of the Retroviridae family. Currently, no disease has been firmly linked to persistent foamy virus infection of their cognate host including simians, bovines, felines, and equines or upon zoonotic transmission of different simian foamy viruses to humans. Bovine and simian foamy viruses have been recently shown to encode a RNA polymerase-III-driven micro RNA cluster which likely modulates and regulates host-virus interactions at different levels. Using sub-genomic bovine foamy virus micro RNA expression plasmids and dual luciferase reporter assays as readout, the requirements for expression and processing of the bovine foamy virus micro RNAs have been analyzed. Here, we report that the minimal BFV micro RNA cassette is significantly weaker than a U6 promoter-based construct and strongly suppressed by flanking sequences. The primary micro RNA sequence can be manipulated and chimerized as long as the dumbbell-like folding of the primary micro RNA is maintained. Since more subtle changes are associated with reduced functionality, the overall structure and shape, but possibly individual elements and residues also, are important for the expression and processing of the bovine foamy virus micro RNAs.