Base Pairing Rules Research Articles

AI-based Prediction of Protein Corona Composition on DNA Nanostructures.

DNA nanotechnology has emerged as a powerful approach to engineering biophysical tools, therapeutics, and diagnostics because it enables the construction of designer nanoscale structures with high programmability. Based on DNA base pairing rules, nanostructure size, shape, surface functionality, and structural reconfiguration can be programmed with a degree of spatial, temporal, and energetic precision that is difficult to achieve with other methods. However, the properties and structure of DNA constructs are greatly altered in vivo due to spontaneous protein adsorption from biofluids. These adsorbed proteins, referred to as the protein corona, remain challenging to control or predict, and subsequently, their functionality and fate in vivo are difficult to engineer. To address these challenges, we prepared a library of diverse DNA nanostructures and investigated the relationship between their design features and the composition of their protein corona. We identified protein characteristics important for their adsorption to DNA nanostructures and developed a machine-learning model that predicts which proteins will be enriched on a DNA nanostructure based on the DNA structures' design features and protein properties. Our work will help to understand and program the function of DNA nanostructures in vivo for biophysical and biomedical applications.

Inosine-Induced Base Pairing Diversity during Reverse Transcription.

A-to-I editing catalyzed by adenosine deaminase acting on RNAs impacts numerous physiological and biochemical processes that are essential for cellular functions and is a big contributor to the infectivity of certain RNA viruses. The outcome of this deamination leads to changes in the eukaryotic transcriptome functionally resembling A-G transitions since inosine preferentially pairs with cytosine. Moreover, hyper-editing or multiple A to G transitions in clusters were detected in measles virus. Inosine modifications either directly on viral RNA or on cellular RNA can have antiviral or pro-viral repercussions. While many of the significant roles of inosine in cellular RNAs are well understood, the effects of hyper-editing of A to I on viral polymerase activity during RNA replication remain elusive. Moreover, biological strategies such as molecular cloning and RNA-seq for transcriptomic interrogation rely on RT-polymerase chain reaction with little to no emphasis placed on the first step, reverse transcription, which may reshape the sequencing results when hypermodification is present. In this study, we systematically explore the influence of inosine modification, varying the number and position of inosines, on decoding outcomes using three different reverse transcriptases (RTs) followed by standard Sanger sequencing. We find that inosine alone or in clusters can differentially affect the RT activity. To gain structural insights into the accommodation of inosine in the polymerase site of HIV-1 reverse transcriptase (HIV-1-RT) and how this structural context affects the base pairing rules for inosine, we performed molecular dynamics simulations of the HIV-1-RT. The simulations highlight the importance of the protein-nucleotide interaction as a critical factor in deciphering the base pairing behavior of inosine clusters. This effort sets the groundwork for decrypting the physiological significance of inosine and linking the fidelity of reverse transcriptase and the possible diverse transcription outcomes of cellular RNAs and/or viral RNAs where hyper-edited inosines are present in the transcripts.

Self-assembled methodologies for the construction of DNA nanostructures and biological applications.

Over the past decades, deoxyribonucleic acid (DNA), as a versatile building block, has been widely employed to construct functionalized nanostructures. Among the diverse types of materials, DNA related nanostructures have gained growing attention due to their intrinsic programmability, favorable biocompatibility, and strong molecular recognition capability. The conventional construction strategy for building DNA structures is based on Watson-Crick base-pairing rules, which are mainly driven by the hydrogen bonding of bases. However, hydrogen bonding-based DNA nanostructures cannot meet the requirements of specific morphology and multifunctionality. Currently, various functional elements have been introduced to expand the synthetic methodologies for constructing the DNA hybrid nanostructures, including small molecules, peptide polymers, organic ligands and transition metal ions. Besides, the potential applications for these DNA hybrid nanostructures have also been explored. It has been demonstrated that DNA hybrid structures with various properties can be extensively applied in the fields of magnetic resonance, luminescence imaging, biomedical detection, and drug delivery systems. In this review, we highlight the pioneering contributions to the methodologies of DNA-based nanostructure assembly. Furthermore, the recent advances in drug delivery systems and biomedical diagnosis based on DNA hybrid nanostructures are briefly summarized.

Simulation-guided tunable DNA probe design for mismatch tolerant hybridization.

The ability to both sensitively and specifically assess the sequence composition of a nucleic acid strand is an ever-growing field. Designing a detection scheme that can perform this function when the sequence of the target being detected deviates significantly from the canonical sequence however is difficult in part because probe/primer design is based on established Watson-Crick base-pairing rules. We present here a robust and tunable toehold-based exchange probe that can detect a sequence with a variable number of SNPs of unknown identity by inserting a series of controlled, sequential mismatches into the protector seal of the toehold probe, in an effort to make the protector seal "sloppy". We show that the mismatch-tolerant system follows predicted behavior closely even with targets containing up to four mismatches that thermodynamically deviate from the canonical sequence by up to 15 kcal/mole. The system also performs faithfully regardless of the global mismatch position on either the protector seal or target. Lastly, we demonstrate the generalizability of the approach by testing the increasingly mismatch-tolerant protectors on HIV clinical samples to show that the system is capable of resolving multiple, iteratively mutated sequences derived from numerous HIV sub-populations with remarkable precision.

Harnessing non-Watson-Crick's base pairing to enhance CRISPR effectors cleavage activities and enable gene editing in mammalian cells.

Genomic DNA of the cyanophage S-2L virus is composed of 2-aminoadenine (Z), thymine (T), guanine (G), and cytosine (C), forming the genetic alphabet ZTGC, which violates Watson-Crick base pairing rules. The Z-base has an extra amino group on the two position that allows the formation of a third hydrogen bond with thymine in DNA strands. Here, we explored and expanded applications of this non-Watson-Crick base pairing in protein expression and gene editing. Both ZTGC-DNA (Z-DNA) and ZUGC-RNA (Z-RNA) produced in vitro show detectable compatibility and can be decoded in mammalian cells, including Homo sapiens cells. Z-crRNA can guide CRISPR-effectors SpCas9 and LbCas12a to cleave specific DNA through non-Watson-Crick base pairing and boost cleavage activities compared to A-crRNA. Z-crRNA can also allow for efficient gene and base editing in human cells. Together, our results help pave the way for potential strategies for optimizing DNA or RNA payloads for gene editing therapeutics and give insights to understanding the natural Z-DNA genome.

Potential therapeutic strategies for osteoarthritis via CRISPR/Cas9 mediated gene editing.

Osteoarthritis (OA) is an incapacitating and one of the most common physically degenerative conditions with an assorted etiology and a highly complicated molecular mechanism that to date lacks an efficient treatment. The capacity to design biological networks and accurately modify existing genomic sites holds an apt potential for applications across medical and biotechnological sciences. One of these highly specific genomes editing technologies is the CRISPR/Cas9 mechanism, referred to as the clustered regularly interspaced short palindromic repeats, which is a defense mechanism constituted by CRISPR associated protein 9 (Cas9) directed by small non-coding RNAs (sncRNA) that bind to target DNA through Watson-Crick base pairing rules where subsequent repair of the target DNA is initiated. Up-to-date research has established the effectiveness of the CRISPR/Cas9 mechanism in targeting the genetic and epigenetic alterations in OA by suppressing or deleting gene expressions and eventually distributing distinctive anti-arthritic properties in both in vitro and in vivo osteoarthritic models. This review aims to epitomize the role of this high-throughput and multiplexed gene editing method as an analogous therapeutic strategy that could greatly facilitate the clinical development of OA-related treatments since it's reportedly an easy, minimally invasive technique, and a comparatively less painful method for osteoarthritic patients.

Watson-Crick Base Pairing of N-Methoxy-1,3-Oxazinane (MOANA) Nucleoside Analogues within Double-Helical DNA.

Hairpin oligodeoxynucleotides incorporating a (2R,3S)-4-(methoxyamino)butane-1,2,3-triol residue in the middle of the double-helical stem and opposite to either one of the canonical nucleobases or an abasic 2-(hydroxymethyl)tetrahydrofuran-3-ol spacer were synthesized. Under mildly acidic conditions, aromatic aldehydes reacted reversibly with these oligonucleotides, converting the (2R,3S)-4-(methoxyamino)butane-1,2,3-triol unit into a 2-aryl-N-methoxy-1,3-oxazinane nucleoside analogue. The equilibrium of this reaction was found to be dependent on both the aldehyde and the nucleobase opposite to the modified residue. 9-Formyl-9-deazaadenine, combining a large stacking surface with an array of hydrogen bond donors and acceptors, showed the highest affinity as well as selectivity consistent with the rules of Watson-Crick base pairing. 5-Formyluracil or indole-3-carbaldehyde, lacking in either stacking or hydrogen bonding ability, were incorporated with a much lower affinity and selectivity.

Dynamic DNA nanomachines for amplification imaging of diseased cells based on stimuli-responsive mechanism

As a basic functional unit, living cell with sophisticated structures play an indispensable role in life activities. Since the abnormality of important molecules inside cells is closely related to diseases, the dynamic analysis and spatio-temporal monitoring of specific molecules in living cells can provide precious information for the diagnosis and treatment of diseases. More recently, DNA has not only been recognized as the carrier of genetic information, but has also used as a robust building block for the assembly of multitudinous nanoscale structures due to the intrinsic advantages of high programmability of classic Watson–Crick base-pairing rule. Intensive study promotes the rapid progress of nanotechnology in various fields, such as bioimaging, diagnosis, and therapeutics. Among numerous well-defined DNA nanomaterials, DNA nanomachines have been widely exploited in cell imaging owing to their desirable ability to achieve high-resolution temporal and spatial images in response to endogenous or exogenous stimuli. In brief, elaborate DNA nanomachines can undergo structural changes upon the stimuli of target analytes or environmental factors, resulting in rapid increase or reduction of output signals and thereby indirectly reflecting the expression level of targets. DNA nanomachines with high sensitivity and specificity contribute to the recognition of diseased tissues. In this review, we introduce the basic assembly modules of DNA nanomachines and summarize the recent advances in dynamic DNA nanomachines for diseased-cell imaging. Finally, the current challenges and future directions of DNA nanomachines for bioimaging are discussed.

DNA Nanoclusters Combined with One-Shot Radiotherapy Augment Cancer Immunotherapy Efficiency.

Immunotherapy shows immense promise for improving cancer treatment. Combining immunotherapy with radiotherapy provides a conspicuous advantage due to its enhanced abscopal effect. However, established immune tolerance mechanisms in the tumor microenvironment can hamper the generation of a sufficient abscopal effect. Herein, a type of DNA nanocluster (DNAnc) that is self-assembled by a CpG-ODNs-loaded Y-shaped double-stranded DNA vector based on the unique complementary base-pairing rules is designed. The unique structure of DNAnc makes it load more than ≈8125.5 ± 822.5 copies of CpG ODNs within one single nanostructure, which effectively increases resistance to nuclease degradation and elevates the efficiency of repolarizing macrophages to an M1-like phenotype. Mechanistic studies reveal that more DNAncs are endocytosed by macrophages in the cancer tissue and repolarized macrophages to elicit a robust abscopal effect with the accumulation of macrophages induced by radiotherapy, generating potent, long-term, and durable antitumor immunity for the inhibition of tumor metastasis and the prevention of tumor recurrence, which provides a novel strategy to boost cancer immunotherapy.

Translating the Genome into Drugs.

The Human Genome Project ultimately aimed to translate DNA sequence into drugs. With the draft in hand, the Molecular Libraries Program set out to prosecute all genome-encoded proteins for drug discovery with automated high-throughput screening (HTS). This ambitious vision remains unfulfilled, even while innovations in sequencing technology have fully democratized access to genome-scale sequencing. Why? While the central dogma of biology allows us to chart the entirety of cellular metabolism through sequencing, there is no direct coding for chemistry. The rules of base pairing that relate DNA gene to RNA transcript and amino acid sequence do not exist for relating small-molecule structure with macromolecular binding partners and subsequently cellular function. Obtaining such relationships genome-wide is unapproachable via state-of-the-art HTS, akin to attempting genome-wide association studies using turn-of-the-millennium Sanger DNA sequencing.Our laboratory has been engaged in a multipronged technology development campaign to revolutionize molecular screening through miniaturization in pursuit of genome-scale drug discovery capabilities. The compound library was ripe for miniaturization: it clearly needed to become a consumable. We employed DNA-encoded library (DEL) synthesis principles in the development of solid-phase DELs prepared on microscopic beads, each harboring 100 fmol of a single library member and a DNA tag whose sequence describes the structure of the library member. Loading these DEL beads into 100 pL microfluidic droplets followed by online photocleavage, incubation, fluorescence-activated droplet sorting, and DNA sequencing of the sorted DEL beads reveals the chemical structures of bioactive compounds. This scalable library synthesis and screening platform has proven useful in several proof-of-concept projects involving current clinical targets.Moving forward, we face the problem of druggability and proteome-scale assay development. Developing biochemical or cellular assays for all genome-encoded targets is not scalable and likely impossible as most proteins have ill-defined or unknown activity and may not function outside of their native contexts. These are the dark undruggable expanses, and charting them will require advanced synthesis and analytical technologies that can generalize probe discovery, irrespective of mature protein function, to fulfill the Genome Project's vision of proteome-wide control of cellular pharmacology.

I-DNAN6mA: Accurate Identification of DNA N6-Methyladenine Sites Using the Base-Pairing Map and Deep Learning.

The recent discovery of numerous DNA N6-methyladenine (6mA) sites has transformed our perception about the roles of 6mA in living organisms. However, our ability to understand them is hampered by our inability to identify 6mA sites rapidly and cost-efficiently by existing experimental methods. Developing a novel method to quickly and accurately identify 6mA sites is critical for speeding up the progress of its function detection and understanding. In this study, we propose a novel computational method, called I-DNAN6mA, to identify 6mA sites and complement experimental methods well, by leveraging the base-pairing rules and a well-designed three-stage deep learning model with pairwise inputs. The performance of our proposed method is benchmarked and evaluated on four species, i.e., Arabidopsis thaliana, Drosophila melanogaster, Rice, and Rosaceae. The experimental results demonstrate that I-DNAN6mA achieves area under the receiver operating characteristic curve values of 0.967, 0.963, 0.947, 0.976, and 0.990, accuracies of 91.5, 92.7, 88.2, 0.938, and 96.2%, and Mathew's correlation coefficient values of 0.855, 0.831, 0.763, 0.877, and 0.924 on five benchmark data sets, respectively, and outperforms several existing state-of-the-art methods. To our knowledge, I-DNAN6mA is the first approach to identify 6mA sites using a novel image-like representation of DNA sequences and a deep learning model with pairwise inputs. I-DNAN6mA is expected to be useful for locating functional regions of DNA.

Predicting RNA secondary structure by a neural network: what features may be learned?

Deep learning is a class of machine learning techniques capable of creating internal representation of data without explicit preprogramming. Hence, in addition to practical applications, it is of interest to analyze what features of biological data may be learned by such models. Here, we describe PredPair, a deep learning neural network trained to predict base pairs in RNA structure from sequence alone, without any incorporated prior knowledge, such as the stacking energies or possible spatial structures. PredPair learned the Watson-Crick and wobble base-pairing rules and created an internal representation of the stacking energies and helices. Application to independent experimental (DMS-Seq) data on nucleotide accessibility in mRNA showed that the nucleotides predicted as paired indeed tend to be involved in the RNA structure. The performance of the constructed model was comparable with the state-of-the-art method based on the thermodynamic approach, but with a higher false positives rate. On the other hand, it successfully predicted pseudoknots. t-SNE clusters of embeddings of RNA sequences created by PredPair tend to contain embeddings from particular Rfam families, supporting the predictions of PredPair being in line with biological classification.

A universal model of RNA.DNA:DNA triplex formation accurately predicts genome-wide RNA-DNA interactions.

RNA.DNA:DNA triple helix (triplex) formation is a form of RNA-DNA interaction which regulates gene expression but is difficult to study experimentally in vivo. This makes accurate computational prediction of such interactions highly important in the field of RNA research. Current predictive methods use canonical Hoogsteen base pairing rules, which whilst biophysically valid, may not reflect the plastic nature of cell biology. Here, we present the first optimization approach to learn a probabilistic model describing RNA-DNA interactions directly from motifs derived from triplex sequencing data. We find that there are several stable interaction codes, including Hoogsteen base pairing and novel RNA-DNA base pairings, which agree with in vitro measurements. We implemented these findings in TriplexAligner, a program that uses the determined interaction codes to predict triplex binding. TriplexAligner predicts RNA-DNA interactions identified in all-to-all sequencing data more accurately than all previously published tools in human and mouse and also predicts previously studied triplex interactions with known regulatory functions. We further validated a novel triplex interaction using biophysical experiments. Our work is an important step towards better understanding of triplex formation and allows genome-wide analyses of RNA-DNA interactions.

An aptasensor for cadmium ions detection based on PEI-MoS2@Au NPs 3D flower-like nanocomposites and Thi-PtPd NPs core-shell sphere

A novel ultrasensitive electrochemical aptasensor was proposed for quantitative detection of Cd2+. To this end, flower-like polyethyleneimine-functionalized molybdenum disulfide-supported gold nanoparticles (PEI-MoS2 NFs@Au NPs) were used as substrates for the modification of bare gold electrodes (AuE). PEI-MoS2 NFs@Au NPs not only possessed excellent biocompatibility and large specific surface area to enhance the cDNA loading capacity, but also possessed good conductivity to accelerate the electron transfer rate. Furthermore, the preparation of dendritic platinum-palladium nanoparticles (PtPd NPs) can effectively load Cd2+-aptamer. Thionine and aptamers were loaded onto PtPd NPs to construct Thi-PtPd NPs-aptamer signal probes. The signal probes were captured by the cDNA immobilized on the electrode via base-pairing rule, and the signal of Thi was detected by differential pulse voltammetry (DPV). In the presence of Cd2+, aptamer-cDNA unwinded, and the combination of aptamer and Cd2+ caused the signal probes to fall off the electrode and the electrical signal decreases. Under optimal conditions, the proposed aptasensor exhibited a linear relationship between the logarithm of Cd2+ concentration and the current response over a wide range of 1 × 10−3 nM to 1 × 102 nM, with a detection limit of 2.34 × 10−4 nM. At the same time, the aptasensor was used to detect Cd2+ in tap water with satisfactory results. In addition, it has good reproducibility, selectivity and stability, and has broad application prospects in heavy metal analysis.

Watson-Crick Base Pairing-Inspired Laser/GSH Activatable miRNA-Coordination Polymer Nanoplexes for Combined Cancer Chemo-Immuno-Photothermal Therapy.

The tumor immunosuppressive microenvironment (TIM) greatly hindered the efficacy of cancer immunotherapy. Overexpressed indoleamine 2,3-dioxygenase-1 (IDO1) in tumor tissues plays a vital role in TIM generation, and downregulation of IDO1 expression may reverse TIM. Inspired by the Watson-Crick base-pairing rule, a versatile noncationic miRNA vector (miDAC@PDA) is developed for cancer immunotherapy. Doxorubicin (DOX), adenosine triphosphate (ATP), and copper ions (Cu2+) are coassembled into coordination polymer nanoparticles (DAC) and bind miRNA via the hydrogen bond interaction (miDAC) between adenine residues (ATP) and uracil residues (miRNA). Polydopamine (PDA) is deposited onto the surface of miDAC for photothermal therapy. miDAC@PDA can efficiently accumulate into tumor tissues for cellular uptake. Under laser irradiation and high intracellular GSH levels, the PDA shell of miDAC@PDA can dissociate from miDAC for miRNA release due to local hyperthermia. Cu2+-mediated GSH consumption and intracellular ATP release can amplify the DOX-based immunogenic cell death (ICD) cascade, together with miR-448-mediated IDO1 inhibition, and these versatile nanoplexes will not only restrain primary tumor growth but also display a remarkable abscopal effect on distant tumors. Collectively, our study provides a unique strategy for intracellular gene delivery and an inspirational approach for multimechanism cancer management.

Nucleic Acid Nanotechnology for Diagnostics and Therapeutics in Acute Kidney Injury.

Acute kidney injury (AKI) has impacted a heavy burden on global healthcare system with a high morbidity and mortality in both hospitalized and critically ill patients. However, there are still some shortcomings in clinical approaches for the disease to date, appealing for an earlier recognition and specific intervention to improve long-term outcomes. In the past decades, owing to the predictable base-pairing rule and highly modifiable characteristics, nucleic acids have already become significant biomaterials for nanostructure and nanodevice fabrication, which is known as nucleic acid nanotechnology. In particular, its excellent programmability and biocompatibility have further promoted its intersection with medical challenges. Lately, there have been an influx of research connecting nucleic acid nanotechnology with the clinical needs for renal diseases, especially AKI. In this review, we begin with the diagnostics of AKI based on nucleic acid nanotechnology with a highlight on aptamer- and probe-functionalized detection. Then, recently developed nanoscale nucleic acid therapeutics towards AKI will be fully elucidated. Furthermore, the strengths and limitations will be summarized, envisioning a wiser and wider application of nucleic acid nanotechnology in the future of AKI.

Advancements in 3WJ-based RNA nanotechnology and its application for cancer diagnosis and therapy.

This attractive and intriguing Ribonucleic acid (RNA) nanotechnology has been conceptualized over the last two decades and with our increasing understanding of RNA structure and function and improvements of RNA nanotechnology it is now possible to use this in clinical settings. Here we review the unique properties and the recent advances in RNA nanotechnology and then look at its scientific and preclinical applications for tumor diagnosis and targeted delivery and RNA-based therapy using RNA nanoparticles with diverse structures and functions. Finally, we discuss the future perspectives and challenges to RNA nanotechnology. RNA can be designed and manipulated in a similar way to DNA while having different rules for base-pairing and displaying functions similar to proteins. Rationally designed RNA nanoparticles based on the three-way junction (3WJ) motif as the core scaffold have been extensively explored in the field of nanomedicine and targeted cancer diagnosis and therapy. RNA nanostructures based on 3WJs demonstrate promising future applications due to their thermal stability, molecular-level plasticity, multifunctional chemotherapeutic drug delivery and other intrinsic characteristics, which will greatly improve the treatment of cancer and promote further major breakthroughs in this field.

Catalytic hairpin assembly coupled with Ir (III) complex sensitization strategy for split-type photoelectrochemical biosensing

High-performance photoelectrochemical (PEC) assay of low-abundance microRNAs (miRNAs) still suffers from a challenge. Herein, a skillful magnetic extraction-assisted PEC biosensor has been developed based on miRNA-20b induced catalytic hairpin assembly (CHA) quenching the enhanced photoelectric signal of CdSe QDs and [Ir(ppy)2(phen-NH2)]+PF6−. A small amount of miRNA-20b can initiate the CHA amplification, and the separated Fe3O4 @SiO2-H1-H2 composites incorporation with CdSe QDs-S1 and the iridium (III) complex cause a decrease in the number of organic-inorganic hybrid nanomaterials in the liquid phase, thereby bringing about a proportional decrease in photoelectric signal. This flexible design and one-step electrode assembly contribute to the PEC biosensor with advantages of stability, reproducibility, as well as short-time consumption. Moreover, CHA amplification technique and organic-inorganic hybrid nanomaterial sensitization strategy make the detection more sensitive. In addition, the specific Watson-Crick base-pairing rules coupled with bio-functionalized superparamagnetic Fe3O4@SiO2 particles enable the PEC biosensor with excellent selectivity in serum samples. Particularly, instead of the iridium (III) complex as an intercalator in duplex DNAs for photocurrent readout, this split-type strategy endows a robust PEC biosensor and provides an alternative perspective for general detection of miRNAs.

Framework Nucleic Acids in Nuclear Medicine Imaging: Shedding Light on Nano-Bio Interactions.

Framework nucleic acids (FNAs) represent nanoscale oligonucleotide assemblies with unique physical, chemical, and biological properties that are different from their building blocks. Following simple Watson-Crick base-pairing rules, arbitrary DNA frameworks with diverse shapes, sizes, and dimensions can be prepared with high reproducibility and stability. The programmable assembly of nucleic acids into FNAs presents a highly controllable model for studies on nano-bio interactions and allows scrutiny of "nanostructure-activity relationships." Herein, we present an overview of recent progress with FNAs in the hope of deepening our understanding of nano-bio interfacing. We summarize the biological profiles and immune responses of various FNAs as functions of their shape, sizes, and surface charges. We then highlight recent efforts to apply FNAs for biomedical applications and discuss the challenges of FNAs for potential clinical translation. We believe that this Minireview can bring up-to-date information on FNAs and shed light on how their design may be harnessed for selective biomedical applications.

UFold: fast and accurate RNA secondary structure prediction with deep learning.

For many RNA molecules, the secondary structure is essential for the correct function of the RNA. Predicting RNA secondary structure from nucleotide sequences is a long-standing problem in genomics, but the prediction performance has reached a plateau over time. Traditional RNA secondary structure prediction algorithms are primarily based on thermodynamic models through free energy minimization, which imposes strong prior assumptions and is slow to run. Here, we propose a deep learning-based method, called UFold, for RNA secondary structure prediction, trained directly on annotated data and base-pairing rules. UFold proposes a novel image-like representation of RNA sequences, which can be efficiently processed by Fully Convolutional Networks (FCNs). We benchmark the performance of UFold on both within- and cross-family RNA datasets. It significantly outperforms previous methods on within-family datasets, while achieving a similar performance as the traditional methods when trained and tested on distinct RNA families. UFold is also able to predict pseudoknots accurately. Its prediction is fast with an inference time of about 160 ms per sequence up to 1500 bp in length. An online web server running UFold is available at https://ufold.ics.uci.edu. Code is available at https://github.com/uci-cbcl/UFold.