iPro2L-DG: Hybrid network based on improved densenet and global attention mechanism for identifying promoter sequences.

The promoter is a key DNA sequence whose primary function is to control the initiation time and the degree of expression of gene transcription. Accurate identification of promoters is essential for understanding gene expression studies. Traditional sequencing techniques for identifying promoters are costly and time-consuming. Therefore, the development of computational methods to identify promoters has become critical. Since deep learning methods show great potential in identifying promoters, this study proposes a new promoter prediction model, called iPro2L-DG. The iPro2L-DG predictor, based on an improved Densely Connected Convolutional Network (DenseNet) and a Global Attention Mechanism (GAM), is constructed to achieve the prediction of promoters. The promoter sequences are combined feature encoding using C2 encoding and nucleotide chemical property (NCP) encoding. An improved DenseNet extracts advanced feature information from the combined feature encoding. GAM evaluates the importance of advanced feature information in terms of channel and spatial dimensions, and finally uses a Full Connect Neural Network (FNN) to derive prediction probabilities. The experimental results showed that the accuracy of iPro2L-DG in the first layer (promoter identification) was 94.10% with Matthews correlation coefficient value of 0.8833. In the second layer (promoter strength prediction), the accuracy was 89.42% with Matthews correlation coefficient value of 0.7915. The iPro2L-DG predictor significantly outperforms other existing predictors in promoter identification and promoter strength prediction. Therefore, our proposed model iPro2L-DG is the most advanced promoter prediction tool. The source code of the iPro2L-DG model can be found in https://github.com/leirufeng/iPro2L-DG.

Radical-Induced Photochromic Silver(I) Metal-Organic Frameworks: Alternative Topology, Dynamic Photoluminescence and Efficient Photothermal Conversion Modulated by Anionic Guests.

Photogenerated radicals are an indispensable member of the state-of-the-art photochromic material family, as they can effectively modulate the photoluminescence and photothermal conversion performance of radical-induced photochromic complexes. Herein, two novel radical-induced photochromic metal‒organic frameworks (MOFs), [Ag(TEPE)](AC)⋅7/4H2O⋅5/4EtOH (1) and [Ag(TEPE)](NC)⋅3H2O⋅EtOH (2), are reported. Distinctly different topological networks can be obtained by judiciously introducing alternative π-conjugated anionic guests, including a new topological structure (named as sfm) first reported in this work, describing as 4,4,4,4-c net. EPR data and UV-Vis spectra prove the radical-induced photochromic mechanism. Dynamic photochromism exhibits tunability in a wide CIE color space, with a linear segment from yellow to red for 1, while a curved coordinate line for 2, resulting in colorful emission from blue to orange. Moreover, photogenerated TEPE* radicals effectively activate the near-infrared (NIR) photothermal conversion effect of MOFs. Under 1 W cm-2 808 nm laser irradiation, the surface temperatures of photoproducts 1* and 2* can reach ~160 ℃ and ~120 ℃, respectively, with competitive NIR photothermal conversion efficiencies η = 51.8% (1*) and 36.2% (2*). This work develops a feasible electrostatic compensation strategy to accurately introduce photoactive anionic guests into MOFs to construct multifunctional radical-induced photothermal conversion materials with tunable photoluminescence behavior.

i5mC-DCGA: an improved hybrid network framework based on the CBAM attention mechanism for identifying promoter 5mC sites.

BACKGROUND: 5-Methylcytosine (5mC) plays a very important role in gene stability, transcription, and development. Therefore, accurate identification of the 5mC site is of key importance in genetic and pathological studies. However, traditional experimental methods for identifying 5mC sites are time-consuming and costly, so there is an urgent need to develop computational methods to automatically detect and identify these 5mC sites. RESULTS: Deep learning methods have shown great potential in the field of 5mC sites, so we developed a deep learning combinatorial model called i5mC-DCGA. The model innovatively uses the Convolutional Block Attention Module (CBAM) to improve the Dense Convolutional Network (DenseNet), which is improved to extract advanced local feature information. Subsequently, we combined a Bidirectional Gated Recurrent Unit (BiGRU) and a Self-Attention mechanism to extract global feature information. Our model can learn feature representations of abstract and complex from simple sequence coding, while having the ability to solve the sample imbalance problem in benchmark datasets. The experimental results show that the i5mC-DCGA model achieves 97.02%, 96.52%, 96.58% and 85.58% in sensitivity (Sn), specificity (Sp), accuracy (Acc) and matthews correlation coefficient (MCC), respectively. CONCLUSIONS: The i5mC-DCGA model outperforms other existing prediction tools in predicting 5mC sites, and it is currently the most representative promoter 5mC site prediction tool. The benchmark dataset and source code for the i5mC-DCGA model can be found in https://github.com/leirufeng/i5mC-DCGA .

Design of Novel 2-Phenylquinazolin-4-amines as Selective CYP1B1 Inhibitors for Overcoming Paclitaxel Resistance in A549 Cells.

Cytochrome P450 1B1 (CYP1B1) contributes to the metabolic inactivation of chemotherapeutics when overexpressed in tumor cells. Selective inhibition of CYP1B1 holds promise for reversing drug resistance. In our pursuit of potent CYP1B1 inhibitors, we designed and synthesized a series of 2-phenylquinazolin-4-amines. A substantial proportion of these newly developed inhibitors demonstrated inhibitory activity against CYP1B1, accompanied by improved water solubility. Remarkably, compound 14b exhibited exceptional inhibitory efficacy and selectivity toward CYP1B1. Molecular docking studies suggested that the expansion of the π-system through aromatization, the introduction of an amine group, and iodine atom augmented the binding affinity. Furthermore, inhibitors 14a, 14b, and 14e demonstrated the ability to significantly reduce the resistance in A549 cells to paclitaxel, while also inhibiting the migration and invasion of these cells. Finally, radioiodine labeling experiments shed light on the metabolic pathway of compound 5l in mice, highlighting the potential of 125I-5l as a radioactive probe for future research endeavors.

NIR-II photothermal conversion and imaging based on a cocrystal containing twisted components.

On account of the scarcity of molecules with a satisfactory second near-infrared (NIR-II) response, the design of high-performance organic NIR photothermal materials has been limited. Herein, we investigate a cocrystal incorporating tetrathiafulvalene (TTF) and tetrachloroperylene dianhydride (TCPDA) components. A stable radical was generated through charge transfer from TTF to TCPDA, which exhibits strong and wide-ranging NIR-II absorption. The metal-free TTF-TCPDA cocrystal in this research shows high photothermal conversion capability under 1064 nm laser irradiation and clear photothermal imaging. The remarkable conversion ability-which is a result of twisted components in the cocrystal-has been demonstrated by analyses of single crystal X-ray diffraction, photoluminescence and femtosecond transient absorption spectroscopy as well as theoretical calculations. We have discovered that space charge separation and the ordered lattice in the TTF-TCPDA cocrystal suppress the radiative decay, while simultaneously strong intermolecular charge transfer enhances the non-radiative decay. The twisted TCPDA component induces rapid charge recombination, while the distorted configuration in TTF-TCPDA favors an internal non-radiative pathway. This research has provided a comprehensive understanding of the photothermal conversion mechanism and opened a new way for the design of advanced organic NIR-II photothermal materials.

4mCPred-GSIMP: Predicting DNA N4-methylcytosine sites in the mouse genome with multi-Scale adaptive features extraction and fusion.

The epigenetic modification of DNA N4-methylcytosine (4mC) is vital for controlling DNA replication and expression. It is crucial to pinpoint 4mC's location to comprehend its role in physiological and pathological processes. However, accurate 4mC detection is difficult to achieve due to technical constraints. In this paper, we propose a deep learning-based approach 4mCPred-GSIMP for predicting 4mC sites in the mouse genome. The approach encodes DNA sequences using four feature encoding methods and combines multi-scale convolution and improved selective kernel convolution to adaptively extract and fuse features from different scales, thereby improving feature representation and optimization effect. In addition, we also use convolutional residual connections, global response normalization and pointwise convolution techniques to optimize the model. On the independent test dataset, 4mCPred-GSIMP shows high sensitivity, specificity, accuracy, Matthews correlation coefficient and area under the curve, which are 0.7812, 0.9312, 0.8562, 0.7207 and 0.9233, respectively. Various experiments demonstrate that 4mCPred-GSIMP outperforms existing prediction tools.

iGly-IDN: Identifying Lysine Glycation Sites in Proteins Based on Improved DenseNet.

Lysine glycation is one of the most significant protein post-translational modifications, which changes the properties of the proteins and causes them to be dysfunctional. Accurately identifying glycation sites helps to understand the biological function and potential mechanism of glycation in disease treatments. Nonetheless, the experimental methods are ordinarily inefficient and costly, so effective computational methods need to be developed. In this study, we proposed the new model called iGly-IDN based on the improved densely connected convolutional networks (DenseNet). First, one hot encoding was adopted to obtain the original feature maps. Afterward, the improved DenseNet was adopted to capture feature information with the importance degrees during the feature learning. According to the experimental results, Acc reaches 66%, and Mathews correlation coefficient reaches 0.33 on the independent testing data set, which indicates that the iGly-IDN can provide more effective glycation site identification than the current predictors.

SNO-DCA: A model for predicting S-nitrosylation sites based on densely connected convolutional networks and attention mechanism.

Protein S-nitrosylation is a reversible oxidative reduction post-translational modification that is widely present in the biological community. S-nitrosylation can regulate protein function and is closely associated with a variety of diseases, thus identifying S-nitrosylation sites are crucial for revealing the function of proteins and related drug discovery. Traditional experimental methods are time-consuming and expensive; therefore, it is necessary to explore more efficient computational methods. Deep learning algorithms perform well in the field of bioinformatics sites prediction, and many studies show that they outperform existing machine learning algorithms. In this work, we proposed a deep learning algorithm-based predictor SNO-DCA for distinguishing between S-nitrosylated and non-S-nitrosylated sequences. First, one-hot encoding of protein sequences was performed. Second, the dense convolutional blocks were used to capture feature information, and an attention module was added to weigh different features to improve the prediction ability of the model. The 10-fold cross-validation and independent testing experimental results show that our SNO-DCA model outperforms existing S-nitrosylation sites prediction models under imbalanced data. In this paper, a web server prediction website: https://sno.cangmang.xyz/SNO-DCA/was established to provide an online prediction service for users. SNO-DCA can be available at https://github.com/peanono/SNO-DCA.

EMDL_m6Am: identifying N6,2'-O-dimethyladenosine sites based on stacking ensemble deep learning.

BACKGROUND: N6, 2'-O-dimethyladenosine (m6Am) is an abundant RNA methylation modification on vertebrate mRNAs and is present in the transcription initiation region of mRNAs. It has recently been experimentally shown to be associated with several human disorders, including obesity genes, and stomach cancer, among others. As a result, N6,2'-O-dimethyladenosine (m6Am) site will play a crucial part in the regulation of RNA if it can be correctly identified. RESULTS: This study proposes a novel deep learning-based m6Am prediction model, EMDL_m6Am, which employs one-hot encoding to expressthe feature map of the RNA sequence and recognizes m6Am sites by integrating different CNN models via stacking. Including DenseNet, Inflated Convolutional Network (DCNN) and Deep Multiscale Residual Network (MSRN), the sensitivity (Sn), specificity (Sp), accuracy (ACC), Mathews correlation coefficient (MCC) and area under the curve (AUC) of our model on the training data set reach 86.62%, 88.94%, 87.78%, 0.7590 and 0.8778, respectively, and the prediction results on the independent test set are as high as 82.25%, 79.72%, 80.98%, 0.6199, and 0.8211. CONCLUSIONS: In conclusion, the experimental results demonstrated that EMDL_m6Am greatly improved the predictive performance of the m6Am sites and could provide a valuable reference for the next part of the study. The source code and experimental data are available at: https://github.com/13133989982/EMDL-m6Am .

EMDL-ac4C: identifying N4-acetylcytidine based on ensemble two-branch residual connection DenseNet and attention.

Introduction: N4-acetylcytidine (ac4C) is a critical acetylation modification that has an essential function in protein translation and is associated with a number of human diseases. Methods: The process of identifying ac4C sites by biological experiments is too cumbersome and costly. And the performance of several existing computational models needs to be improved. Therefore, we propose a new deep learning tool EMDL-ac4C to predict ac4C sites, which uses a simple one-hot encoding for a unbalanced dataset using a downsampled ensemble deep learning network to extract important features to identify ac4C sites. The base learner of this ensemble model consists of a modified DenseNet and Squeeze-and-Excitation Networks. In addition, we innovatively add a convolutional residual structure in parallel with the dense block to achieve the effect of two-layer feature extraction. Results: The average accuracy (Acc), mathews correlation coefficient (MCC), and area under the curve Area under curve of EMDL-ac4C on ten independent testing sets are 80.84%, 61.77%, and 87.94%, respectively. Discussion: Multiple experimental comparisons indicate that EMDL-ac4C outperforms existing predictors and it greatly improved the predictive performance of the ac4C sites. At the same time, EMDL-ac4C could provide a valuable reference for the next part of the study. The source code and experimental data are available at: https://github.com/13133989982/EMDLac4C.

A polyaniline functionalized NiFeP nanosheet array-based electrochemical immunosensor using Au/Cu2O nanocubes as a signal amplifier for the detection of SARS-CoV-2 nucleocapsid protein.

Coronavirus disease 2019 (COVID-19) is an infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which emerged as a novel pathogen in 2019. The virus is responsible for a severe acute respiratory syndrome outbreak, affecting the respiratory system of infected individuals. COVID-19 is a super amplifier of basic diseases, and the disease with basic diseases is often more serious. Controlling the spread of the COVID-19 pandemic relies heavily on the timely and accurate detection of the virus. To resolve the problem, a polyaniline functionalized NiFeP nanosheet array-based electrochemical immunosensor using Au/Cu2O nanocubes as a signal amplifier is fabricated for the detection of SARS-CoV-2 nucleocapsid protein (SARS-CoV-2 NP). Polyaniline (PANI) functionalized NiFeP nanosheet arrays are synthesized as an ideal sensing platform for the first time. PANI is coated on the surface of NiFeP by electropolymerization to enhance biocompatibility, beneficial for the efficient loading of the capture antibody (Ab1). Significantly, Au/Cu2O nanocubes possess excellent peroxidase-like activity and exhibit outstanding catalytic activity for the reduction of H2O2. Therefore, Au/Cu2O nanocubes combine with a labeled antibody (Ab2) through the Au-N bond to form labeled probes, which can effectively amplify current signals. Under optimal conditions, the immunosensor for the detection of SARS-CoV-2 NP shows a wide linear range of 10 fg mL-1-20 ng mL-1 and a low detection limit of 1.12 fg mL-1 (S/N = 3). It also exhibits desirable selectivity, repeatability, and stability. Meanwhile, the excellent analytical performance in human serum samples confirms the practicality of the PANI functionalized NiFeP nanosheet array-based immunosensor. The electrochemical immunosensor based on the Au/Cu2O nanocubes as a signal amplifier demonstrates great potential for application in the personalized point-of-care (POC) clinical diagnosis.

DGA-5mC: A 5-methylcytosine site prediction model based on an improved DenseNet and bidirectional GRU method.

The 5-methylcytosine (5mC) in the promoter region plays a significant role in biological processes and diseases. A few high-throughput sequencing technologies and traditional machine learning algorithms are often used by researchers to detect 5mC modification sites. However, high-throughput identification is laborious, time-consuming and expensive; moreover, the machine learning algorithms are not so advanced. Therefore, there is an urgent need to develop a more efficient computational approach to replace those traditional methods. Since deep learning algorithms are more popular and have powerful computational advantages, we constructed a novel prediction model, called DGA-5mC, to identify 5mC modification sites in promoter regions by using a deep learning algorithm based on an improved densely connected convolutional network (DenseNet) and the bidirectional GRU approach. Furthermore, we added a self-attention module to evaluate the importance of various 5mC features. The deep learning-based DGA-5mC model algorithm automatically handles large proportions of unbalanced data for both positive and negative samples, highlighting the model's reliability and superiority. So far as the authors are aware, this is the first time that the combination of an improved DenseNet and bidirectional GRU methods has been used to predict the 5mC modification sites in promoter regions. It can be seen that the DGA-5mC model, after using a combination of one-hot coding, nucleotide chemical property coding and nucleotide density coding, performed well in terms of sensitivity, specificity, accuracy, the Matthews correlation coefficient (MCC), area under the curve and Gmean in the independent test dataset: 90.19%, 92.74%, 92.54%, 64.64%, 96.43% and 91.46%, respectively. In addition, all datasets and source codes for the DGA-5mC model are freely accessible at https://github.com/lulukoss/DGA-5mC.

High-stability spherical lanthanide nanoclusters for magnetic resonance imaging.

High-nuclear lanthanide clusters have shown great potential for the administration of high-dose mononuclear gadolinium chelates in magnetic resonance imaging (MRI). The development of high-nuclear lanthanide clusters with excellent solubility and high stability in water or solution has been challenging and is very important for expanding the performance of MRI. We used N-methylbenzimidazole-2-methanol (HL) and LnCl3·6H2O to synthesize two spherical lanthanide clusters, Ln32 (Ln = Ho, Ho32; and Ln = Gd, Gd32), which are highly stable in solution. The 24 ligands L- are all distributed on the periphery of Ln32 and tightly wrap the cluster core, ensuring that the cluster is stable. Notably, Ho32 can remain highly stable when bombarded with different ion source energies in HRESI-MS or immersed in an aqueous solution of different pH values for 24 h. The possible formation mechanism of Ho32 was proposed to be Ho(III), (L)- and H2O → Ho3(L)3/Ho3(L)4 → Ho4(L)4/Ho4(L)5 → Ho6(L)6/Ho6(L)7 → Ho16(L)19 → Ho28(L)15 → Ho32(L)24/Ho32(L)21/Ho32(L)23. To the best of our knowledge, this is the first study of the assembly mechanism of spherical high-nuclear lanthanide clusters. Spherical cluster Gd32, a form of highly aggregated Gd(III), exhibits a high longitudinal relaxation rate (1 T, r1 = 265.87 mM-1·s-1). More notably, compared with the clinically used commercial material Gd-DTPA, Gd32 has a clearer and higher-contrast T1-weighted MRI effect in mice bearing 4T1 tumors. This is the first time that high-nuclear lanthanide clusters with high water stability have been utilized for MRI. High-nuclear Gd clusters containing highly aggregated Gd(III) at the molecular level have higher imaging contrast than traditional Gd chelates; thus, using large doses of traditional gadolinium contrast agents can be avoided.

Development of [18F]Thiazolylacylaminopyridine-Based Glycogen synthase Kinase-3ß ligands for positron emission tomography imaging.

Glycogen synthase kinase-3ß (GSK-3ß) regulates numerous of CNS-specific signaling pathways, and is particularly implicated in various pathogenetic mechanisms of Alzheimer's disease (AD). A noninvasive method for detecting GSK-3ß in AD brains via positron emission tomography (PET) imaging could enhance the understanding of AD pathogenesis and aid in the development of AD therapeutic drugs. In this study, an array of fluorinated thiazolyl acylaminopyridines (FTAAP) targeting GSK-3ß were designed and synthesized. These compounds showed moderate to high affinities (IC50 = 6.0 - 426 nM) for GSK-3ß in vitro. A potential GSK-3ß tracer, [18F]8, was successfully radiolabeled. [18F]8 had unsatisfactory initial brain uptake despite its suitable lipophilicity, molecular size and good stability. Further structural refinement of the lead compound is needed to develop promising [18F]-labeled radiotracers for the detection of GSK-3ß in AD brains.

iEnhancer-DCSV: Predicting enhancers and their strength based on DenseNet and improved convolutional block attention module.

Enhancers play a crucial role in controlling gene transcription and expression. Therefore, bioinformatics puts many emphases on predicting enhancers and their strength. It is vital to create quick and accurate calculating techniques because conventional biomedical tests take too long time and are too expensive. This paper proposed a new predictor called iEnhancer-DCSV built on a modified densely connected convolutional network (DenseNet) and an improved convolutional block attention module (CBAM). Coding was performed using one-hot and nucleotide chemical property (NCP). DenseNet was used to extract advanced features from raw coding. The channel attention and spatial attention modules were used to evaluate the significance of the advanced features and then input into a fully connected neural network to yield the prediction probabilities. Finally, ensemble learning was employed on the final categorization findings via voting. According to the experimental results on the test set, the first layer of enhancer recognition achieved an accuracy of 78.95%, and the Matthews correlation coefficient value was 0.5809. The second layer of enhancer strength prediction achieved an accuracy of 80.70%, and the Matthews correlation coefficient value was 0.6609. The iEnhancer-DCSV method can be found at https://github.com/leirufeng/iEnhancer-DCSV. It is easy to obtain the desired results without using the complex mathematical formulas involved.

DeepDN_iGlu: prediction of lysine glutarylation sites based on attention residual learning method and DenseNet.

As a key issue in orchestrating various biological processes and functions, protein post-translational modification (PTM) occurs widely in the mechanism of protein's function of animals and plants. Glutarylation is a type of protein-translational modification that occurs at active ε-amino groups of specific lysine residues in proteins, which is associated with various human diseases, including diabetes, cancer, and glutaric aciduria type I. Therefore, the issue of prediction for glutarylation sites is particularly important. This study developed a brand-new deep learning-based prediction model for glutarylation sites named DeepDN_iGlu via adopting attention residual learning method and DenseNet. The focal loss function is utilized in this study in place of the traditional cross-entropy loss function to address the issue of a substantial imbalance in the number of positive and negative samples. It can be noted that DeepDN_iGlu based on the deep learning model offers a greater potential for the glutarylation site prediction after employing the straightforward one hot encoding method, with Sensitivity (Sn), Specificity (Sp), Accuracy (ACC), Mathews Correlation Coefficient (MCC), and Area Under Curve (AUC) of 89.29%, 61.97%, 65.15%, 0.33 and 0.80 accordingly on the independent test set. To the best of the authors' knowledge, this is the first time that DenseNet has been used for the prediction of glutarylation sites. DeepDN_iGlu has been deployed as a web server (https://bioinfo.wugenqiang.top/~smw/DeepDN_iGlu/) that is available to make glutarylation site prediction data more accessible.

Magnetic and Luminescent Dual Responses of Photochromic Hexaazamacrocyclic Lanthanide Complexes.

Herein, hexaazamacrocyclic ligand LN6 was employed to construct a series of photochromic rare-earth complexes, [Ln(LN6)(NO3)2](BPh4) [1-Ln, Ln = Dy, Tb, Eu, Gd, Y; LN6 = (3E,5E,10E,12E)-3,6,10,13-tetraaza-1,8(2,6)-dipyridinacyclotetradecaphane-3,5,10,12-tetraene]. The behavior of photogenerated radicals of hexaazamacrocyclic ligands was revealed for the first time. Upon 365 nm light irradiation, complexes 1-Ln exhibit photochromic behavior induced by photogenerated radicals according to EPR and UV-vis analyses. Static and dynamic magnetic studies of 1-Dy and irradiated product 1-Dy* indicate weak ferromagnetic interactions among DyIII ions and photogenerated LN6 radicals, as well as slow magnetization relaxation behavior under a 2 kOe applied field. Further fitting analyses show that the magnetization relaxation in 1-Dy* is markedly different from 1-Dy. Time-dependent fluorescence measurements reveal the characteristic luminescence quenching dynamics of lanthanide in the photochromic process. Especially for irradiated product 1-Eu*, the luminescence is almost completely quenched within 5 min with a quenching efficiency of 98.4%. The results reported here provide a prospect for the design of radical-induced photochromic lanthanide single-molecule magnets and will promote the further development of multiresponsive photomagnetic materials.

A new class of CYP1B1 inhibitors derived from bentranil.

Cytochrome P450 1B1 (CYP1B1) is highly expressed in a variety of tumors and implicated to drug resistance. More and more researches have suggested that CYP1B1 is a new target for cancer prevention and therapy. Various CYP1B1 inhibitors with a rigid polycyclic skeleton have been developed, such as flavonoids, trans-stilbenes, and quinazolines. To obtain a new class of CYP1B1 inhibitors, we designed and synthesized a series of bentranil analogues, moreover, IC50 determinations were performed for CYP1B1 inhibition of five of these compounds and found that 6o and 6q were the best inhibitors, with IC50 values in the nM range. The selectivity index (SI) of CYP1B1 over CYP1A1 and CYP1A2 was 30-fold higher than that of α-naphthoflavone (ANF). The molecular docking results showed that compound 6q fitted better into the CYP1B1 binding site than other compounds, which was consistent with our experimental results. On the basis of 6o and 6q, it is expected to develop CYP1B1 inhibitors with stronger affinity, higher selectivity and better solubility.

DLC-ac4C: A Prediction Model for N4-acetylcytidine Sites in Human mRNA Based on DenseNet and Bidirectional LSTM Methods.

Introduction: N4 acetylcytidine (ac4C) is a highly conserved nucleoside modification that is essential for the regulation of immune functions in organisms. Currently, the identification of ac4C is primarily achieved using biological methods, which can be time-consuming and labor-intensive. In contrast, accurate identification of ac4C by computational methods has become a more effective method for classification and prediction. Aim: To the best of our knowledge, although there are several computational methods for ac4C locus prediction, the performance of the models they constructed is poor, and the network structure they used is relatively simple and suffers from the disadvantage of network degradation. This study aims to improve these limitations by proposing a predictive model based on integrated deep learning to better help identify ac4C sites. Methods: In this study, we propose a new integrated deep learning prediction framework, DLC-ac4C. First, we encode RNA sequences based on three feature encoding schemes, namely C2 encoding, nucleotide chemical property (NCP) encoding, and nucleotide density (ND) encoding. Second, one-dimensional convolutional layers and densely connected convolutional networks (DenseNet) are used to learn local features, and bi-directional long short-term memory networks (Bi-LSTM) are used to learn global features. Third, a channel attention mechanism is introduced to determine the importance of sequence characteristics. Finally, a homomorphic integration strategy is used to limit the generalization error of the model, which further improves the performance of the model. Results: The DLC-ac4C model performed well in terms of sensitivity (Sn), specificity (Sp), accuracy (Acc), Mathews correlation coefficient (MCC), and area under the curve (AUC) for the independent test data with 86.23%, 79.71%, 82.97%, 66.08%, and 90.42%, respectively, which was significantly better than the prediction accuracy of the existing methods. Conclusion: Our model not only combines DenseNet and Bi-LSTM, but also uses the channel attention mechanism to better capture hidden information features from a sequence perspective, and can identify ac4C sites more effectively.

im5C-DSCGA: A Proposed Hybrid Framework Based on Improved DenseNet and Attention Mechanisms for Identifying 5-methylcytosine Sites in Human RNA.

BACKGROUND: 5-methylcytosine (m5C) is a key post-transcriptional modification that plays a critical role in RNA metabolism. Owing to the large increase in identified m5C modification sites in organisms, their epigenetic roles are becoming increasingly unknown. Therefore, it is crucial to precisely identify m5C modification sites to gain more insight into cellular processes and other mechanisms related to biological functions. Although researchers have proposed some traditional computational methods and machine learning algorithms, some limitations still remain. In this study, we propose a more powerful and reliable deep-learning model, im5C-DSCGA, to identify novel RNA m5C modification sites in humans. METHODS: Our proposed im5C-DSCGA model uses three feature encoding methods initially-one-hot, nucleotide chemical property (NCP), and nucleotide density (ND)-to extract the original features in RNA sequences and ensure splicing; next, the original features are fed into the improved densely connected convolutional network (DenseNet) and Convolutional Block Attention Module (CBAM) mechanisms to extract the advanced local features; then, the bidirectional gated recurrent unit (BGRU) method is used to capture the long-term dependencies from advanced local features and extract global features using Self-Attention; Finally, ensemble learning is used and full connectivity is used to classify and predict the m5C site. RESULTS: Unsurprisingly, the deep-learning-based im5C-DSCGA model performed well in terms of sensitivity (Sn), specificity (SP), accuracy (Acc), Matthew's correlation coefficient (MCC), and area under the curve (AUC), generating values of 81.0%, 90.8%, 85.9%, 72.1%, and 92.6%, respectively, in the independent test dataset following the use of three feature encoding methods. CONCLUSIONS: We critically evaluated the performance of im5C-DSCGA using five-fold cross-validation and independent testing and compared it to existing methods. The MCC metric reached 72.1% when using the independent test, which is 3.0% higher than the current state-of-the-art prediction method Deepm5C model. The results show that the im5C-DSCGA model achieves more accurate and stable performances and is an effective tool for predicting m5C modification sites. To the authors' knowledge, this is the first time that the improved DenseNet, BGRU, CBAM Attention mechanism, and Self-Attention mechanism have been combined to predict novel m5C sites in human RNA.