NPInter v5.0: ncRNA interaction database in a new era.

Noncoding RNAs (ncRNAs) play key regulatory roles in biological processes by interacting with other biomolecules. With the development of high-throughput sequencing and experimental technologies, extensive ncRNA interactions have been accumulated. Therefore, we updated the NPInter database to a fifth version to document these interactions. ncRNA interaction entries were doubled from 1 100 618 to 2 596 695 by manual literature mining and high-throughput data processing. We integrated global RNA-DNA interactions from iMARGI, ChAR-seq and GRID-seq, greatly expanding the number of RNA-DNA interactions (from 888 915 to 8 329 382). In addition, we collected different types of RNA interaction between SARS-CoV-2 virus and its host from recently published studies. Long noncoding RNA (lncRNA) expression specificity in different cell types from tumor single cell RNA-seq (scRNA-seq) data were also integrated to provide a cell-type level view of interactions. A new module named RBP was built to display the interactions of RNA-binding proteins with annotations of localization, binding domains and functions. In conclusion, NPInter v5.0 (http://bigdata.ibp.ac.cn/npinter5/) provides informative and valuable ncRNA interactions for biological researchers.

Characterizing mobile element insertions in 5675 genomes.

Mobile element insertions (MEIs) are a major class of structural variants (SVs) and have been linked to many human genetic disorders, including hemophilia, neurofibromatosis, and various cancers. However, human MEI resources from large-scale genome sequencing are still lacking compared to those for SNPs and SVs. Here, we report a comprehensive map of 36 699 non-reference MEIs constructed from 5675 genomes, comprising 2998 Chinese samples (∼26.2×, NyuWa) and 2677 samples from the 1000 Genomes Project (∼7.4×, 1KGP). We discovered that LINE-1 insertions were highly enriched in centromere regions, implying the role of chromosome context in retroelement insertion. After functional annotation, we estimated that MEIs are responsible for about 9.3% of all protein-truncating events per genome. Finally, we built a companion database named HMEID for public use. This resource represents the latest and largest genomewide study on MEIs and will have broad utility for exploration of human MEI findings.

Deeply Mining a Universe of Peptides Encoded by Long Noncoding RNAs.

Many small ORFs embedded in long noncoding RNA (lncRNA) transcripts have been shown to encode biologically functional polypeptides (small ORF-encoded polypeptides [SEPs]) in different organisms. Despite some novel SEPs have been found, the identification is still hampered by their poor predictability, diminutive size, and low relative abundance. Here, we take advantage of NONCODE, a repository containing the most complete collection and annotation of lncRNA transcripts from different species, to build a novel database that attempts to maximize a collection of SEPs from human and mouse lncRNA transcripts. In order to further improve SEP discovery, we implemented two effective and complementary polypeptide enrichment strategies using 30-kDa molecular weight cutoff filter and C8 solid-phase extraction column. These combined strategies enabled us to discover 353 SEPs from eight human cell lines and 409 SEPs from three mouse cell lines and eight mouse tissues. Importantly, 19 of them were then verified through in vitro expression, immunoblotting, parallel reaction monitoring, and synthetic peptides. Subsequent bioinformatics analysis revealed that some of the physical and chemical properties of these novel SEPs, including amino acid composition and codon usage, are different from those commonly found in canonical proteins. Intriguingly, nearly 65% of the identified SEPs were found to be initiated with non-AUG start codons. The 762 novel SEPs probably represent the largest number of SEPs detected by MS reported to date. These novel SEPs might not only provide new clues for the annotation of noncoding elements in the genome but also serve as a valuable resource for functional study.

NPInter v4.0: an integrated database of ncRNA interactions.

Noncoding RNAs (ncRNAs) play crucial regulatory roles in a variety of biological circuits. To document regulatory interactions between ncRNAs and biomolecules, we previously created the NPInter database (http://bigdata.ibp.ac.cn/npinter). Since the last version of NPInter was issued, a rapidly growing number of studies have reported novel interactions and accumulated numerous high-throughput interactome data. We have therefore updated NPInter to its fourth edition in which are integrated 600 000 new experimentally identified ncRNA interactions. ncRNA-DNA interactions derived from ChIRP-seq data and circular RNA interactions have been included in the database. Additionally, disease associations were annotated to the interacting molecules. The database website has also been redesigned with a more user-friendly interface and several additional functional modules. Overall, NPInter v4.0 now provides more comprehensive data and services for researchers working on ncRNAs and their interactions with other biomolecules.

Adaptive filter of frequency bands based coordinate attention network for EEG-based motor imagery classification.

Purpose: In the brain-computer interface (BCI), motor imagery (MI) could be defined as the Electroencephalogram (EEG) signals through imagined movements, and ultimately enabling individuals to control external devices. However, the low signal-to-noise ratio, multiple channels and non-linearity are the essential challenges of accurate MI classification. To tackle these issues, we investigate the role of adaptive frequency bands selection and spatial-temporal feature learning in decoding motor imagery. Methods: We propose an Adaptive Filter of Frequency Bands based Coordinate Attention Network (AFFB-CAN) to improve the performance of MI classification. Specifically, we design the AFFB to adaptively obtain the upper and lower limits of frequency bands in order to alleviate information loss caused by manual selection. Next, we propose the CAN-based network to emphasize the key brain regions and temporal segments. And we design a multi-scale module to enhance temporal context learning. Results: The conducted experiments on the BCI Competition IV-2a and 2b datasets reveal that our approach achieves an outstanding average accuracy, kappa values, and Macro F1-Score with 0.7825, 0.7104, and 0.7486 respectively. Similarly, for the BCI Competition IV-2b dataset, the average accuracy, kappa values, and F1-Score obtained are 0.8879, 0.7427, and 0.8734, respectively. Conclusion: The proposed AFFB-CAN method improves the performance of MI classification. In addition, our study confirms previous findings that motor imagery is mainly associated with µ and ß rhythms. Moreover, we also find that γ rhythms also play an important role in MI classification.

Cerebral asymmetry representation learning-based deep subdomain adaptation network for electroencephalogram-based emotion recognition.

Objective.Extracting discriminative spatial information from multiple electrodes is a crucial and challenging problem for electroencephalogram (EEG)-based emotion recognition. Additionally, the domain shift caused by the individual differences degrades the performance of cross-subject EEG classification.Approach.To deal with the above problems, we propose the cerebral asymmetry representation learning-based deep subdomain adaptation network (CARL-DSAN) to enhance cross-subject EEG-based emotion recognition. Specifically, the CARL module is inspired by the neuroscience findings that asymmetrical activations of the left and right brain hemispheres occur during cognitive and affective processes. In the CARL module, we introduce a novel two-step strategy for extracting discriminative features through intra-hemisphere spatial learning and asymmetry representation learning. Moreover, the transformer encoders within the CARL module can emphasize the contributive electrodes and electrode pairs. Subsequently, the DSAN module, known for its superior performance over global domain adaptation, is adopted to mitigate domain shift and further improve the cross-subject performance by aligning relevant subdomains that share the same class samples.Main Results.To validate the effectiveness of the CARL-DSAN, we conduct subject-independent experiments on the DEAP database, achieving accuracies of 68.67% and 67.11% for arousal and valence classification, respectively, and corresponding accuracies of 67.70% and 67.18% on the MAHNOB-HCI database.Significance.The results demonstrate that CARL-DSAN can achieve an outstanding cross-subject performance in both arousal and valence classification.

Joint low-rank tensor fusion and cross-modal attention for multimodal physiological signals based emotion recognition.

OBJECTIVE: Physiological signals based emotion recognition is a prominent research domain in the field of human-computer interaction. Previous studies predominantly focused on unimodal data, giving limited attention to the interplay among multiple modalities. Within the scope of multimodal emotion recognition, integrating the information from diverse modalities and leveraging the complementary information are the two essential issues to obtain the robust representations. APPROACH: Thus, we propose a intermediate fusion strategy for combining low-rank tensor fusion with the cross-modal attention to enhance the fusion of electroencephalogram (EEG), electrooculogram (EOG), electromyography (EMG), and galvanic skin response (GSR). Firstly, handcrafted features from distinct modalities are individually fed to corresponding feature extractors to obtain latent features. Subsequently, low-rank tensor is fused to integrate the information by the modality interaction representation. Finally, a cross-modal attention module is employed to explore the potential relationships between the distinct latent features and modality interaction representation, and recalibrate the weights of different modalities. And the resultant representation is adopted for emotion recognition. MAIN RESULTS: Furthermore, to validate the effectiveness of the proposed method, we execute subject-independent experiments within the DEAP dataset. The proposed method has achieved the accuracies of 73.82% and 74.55% for valence and arousal classification. SIGNIFICANCE: The results of extensive experiments verify the outstanding performance of the proposed method.

A learnable front-end based efficient channel attention network for heart sound classification.

Objective. To enhance the accuracy of heart sound classification, this study aims to overcome the limitations of common models which rely on handcrafted feature extraction. These traditional methods may distort or discard crucial pathological information within heart sounds due to their requirement of tedious parameter settings.Approach.We propose a learnable front-end based Efficient Channel Attention Network (ECA-Net) for heart sound classification. This novel approach optimizes the transformation of waveform-to-spectrogram, enabling adaptive feature extraction from heart sound signals without domain knowledge. The features are subsequently fed into an ECA-Net based convolutional recurrent neural network, which emphasizes informative features and suppresses irrelevant information. To address data imbalance, Focal loss is employed in our model.Main results.Using the well-known public PhysioNet challenge 2016 dataset, our method achieved a classification accuracy of 97.77%, outperforming the majority of previous studies and closely rivaling the best model with a difference of just 0.57%.Significance.The learnable front-end facilitates end-to-end training by replacing the conventional heart sound feature extraction module. This provides a novel and efficient approach for heart sound classification research and applications, enhancing the practical utility of end-to-end models in this field.

Long noncoding RNA MAGI2-AS3 regulates the H2O2 level and cell senescence via HSPA8.

The redox homeostasis system regulates many biological processes, intracellular antioxidant production and redox signaling. However, long noncoding RNAs (lncRNAs) involved in redox regulation have rarely been reported. Herein, we reported that downregulation of MAGI2-AS3 decreased the superoxide level in Human fibroblasts (Fbs), a replicative aging model, as detected by the fluorescent probes dihydroethidium (DHE) and MitoSOX™ Red. RNA pulldown combined with mass spectrometry showed that HSPA8 is a novel interacting protein of MAGI2-AS3, which was further confirmed by photoactivatable ribonucleoside-enhanced crosslinking and immunoprecipitation (PAR-CLIP). Downregulation of MAGI2-AS3 decreased the hydrogen peroxide (H2O2) content by stabilizing the HSPA8 protein level via inhibiting the protesome degradation of HSPA8. Further evidence showed that MAGI2-AS3 interacted with the C-terminal domain (CTD) of HSPA8. Downregulation of MAGI2-AS3 delayed cell senescence, while this antiaging effect was abolished by HSPA8 knockdown. The underlying molecular mechanism by which MAGI2-AS3 knockdown inhibited cell senescence was mediated via suppression of the ROS/MAP2K6/p38 signaling pathway. Taken together, these findings revealed that downregulation of lncRNA MAGI2-AS3 decreased the H2O2 content and delayed cell senescence by stabilizing the HSPA8 protein level, identifying a potential antiaging application.

Identification and characterization of long non-coding RNA Carip in modulating spatial learning and memory.

CaMKII has long been known to be a key effector for synaptic plasticity. Recent studies have shown that a variety of modulators interact with the subunits of CaMKII to regulate the long-term potentiation (LTP) of hippocampal neurons. However, whether long non-coding RNAs modulate the activity of CaMKII and affect synaptic plasticity is still elusive. Here, we identify a previously uncharacterized long non-coding RNA Carip that functions as a scaffold, specifically interacts with CaMKIIß, and regulates the phosphorylation of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-d-aspartate (NMDA) receptor subunits in the hippocampus. The absence of Carip causes dysfunction of synaptic transmission and attenuates LTP in hippocampal CA3-CA1 synapses, which further leads to impairment of spatial learning and memory. In summary, our findings demonstrate that Carip modulates long-term synaptic plasticity by changing AMPA receptor and NMDA receptor activities, thereby affecting spatial learning and memory in mice.

Identification of the transcriptional regulators by expression profiling infected with hepatitis B virus.

BACKGROUND: The molecular pathogenesis of infection by hepatitis B virus with human is extremely complex and heterogeneous. To date the molecular information is not clearly defined despite intensive research efforts. Thus, studies aimed at transcription and regulation during virus infection or combined researches of those already known to be beneficial are needed. AIMS: With the purpose of identifying the transcriptional regulators related to infection of hepatitis B virus in gene level, the gene expression profiles from some normal individuals and hepatitis B patients were analyzed in our study. METHODS: In this work, the differential expressed genes were selected primarily. The several genes among those were validated in an independent set by qRT-PCR. Then the differentially co-expression analysis was conducted to identify differentially co-expressed links and differential co-expressed genes. Next, the analysis of the regulatory impact factors was performed through mapping the links and regulatory data. In order to give a further insight to these regulators, the co-expression gene modules were identified using a threshold-based hierarchical clustering method. Incidentally, the construction of the regulatory network was generated using the computer software. RESULTS: A total of 137,284 differentially co-expressed links and 780 differential co-expressed genes were identified. These co-expressed genes were significantly enriched inflammatory response. The results of regulatory impact factors revealed several crucial regulators related to hepatocellular carcinoma and other high-rank regulators. Meanwhile, more than one hundred co-expression gene modules were identified using clustering method. CONCLUSIONS: In our study, some important transcriptional regulators were identified using a computational method, which may enhance the understanding of disease mechanisms and lead to an improved treatment of hepatitis B. However, further experimental studies are required to confirm these findings.

MitProNet: A knowledgebase and analysis platform of proteome, interactome and diseases for mammalian mitochondria.

Mitochondrion plays a central role in diverse biological processes in most eukaryotes, and its dysfunctions are critically involved in a large number of diseases and the aging process. A systematic identification of mitochondrial proteomes and characterization of functional linkages among mitochondrial proteins are fundamental in understanding the mechanisms underlying biological functions and human diseases associated with mitochondria. Here we present a database MitProNet which provides a comprehensive knowledgebase for mitochondrial proteome, interactome and human diseases. First an inventory of mammalian mitochondrial proteins was compiled by widely collecting proteomic datasets, and the proteins were classified by machine learning to achieve a high-confidence list of mitochondrial proteins. The current version of MitProNet covers 1124 high-confidence proteins, and the remainders were further classified as middle- or low-confidence. An organelle-specific network of functional linkages among mitochondrial proteins was then generated by integrating genomic features encoded by a wide range of datasets including genomic context, gene expression profiles, protein-protein interactions, functional similarity and metabolic pathways. The functional-linkage network should be a valuable resource for the study of biological functions of mitochondrial proteins and human mitochondrial diseases. Furthermore, we utilized the network to predict candidate genes for mitochondrial diseases using prioritization algorithms. All proteins, functional linkages and disease candidate genes in MitProNet were annotated according to the information collected from their original sources including GO, GEO, OMIM, KEGG, MIPS, HPRD and so on. MitProNet features a user-friendly graphic visualization interface to present functional analysis of linkage networks. As an up-to-date database and analysis platform, MitProNet should be particularly helpful in comprehensive studies of complicated biological mechanisms underlying mitochondrial functions and human mitochondrial diseases. MitProNet is freely accessible at http://bio.scu.edu.cn:8085/MitProNet.