Search

Abstract Background Neurocognitive dysfunction is observationally associated with the risk of psychiatric disorders. Blood metabolites, which are readily accessible, may become highly promising biomarkers for brain disorders. However, the causal role of blood metabolites in neurocognitive function, and the biological pathways underlying their association with psychiatric disorders remain unclear. Methods To explore their putative causalities, we conducted bidirectional two-sample Mendelian randomization (MR) using genetic variants associated with 317 human blood metabolites (n max = 215,551), g-Factor (an integrated index of multiple neurocognitive tests with n max = 332,050), and 10 different psychiatric disorders (n = 9,725 to 807,553) from the large-scale genome-wide association studies of European ancestry. Mediation analysis was used to assess the potential causal pathway among the candidate metabolite, neurocognitive trait and corresponding psychiatric disorder. Results MR evidence indicated that genetically predicted acetylornithine was positively associated with g-Factor (0.035 standard deviation units increase in g-Factor per one standard deviation increase in acetylornithine level; 95% confidence interval, 0.021 to 0.049; P = 1.15 × 10−6). Genetically predicted butyrylcarnitine was negatively associated with g-Factor (0.028 standard deviation units decrease in g-Factor per one standard deviation increase in genetically proxied butyrylcarnitine; 95% confidence interval, −0.041 to −0.015; P = 1.31 × 10−5). There was no evidence of associations between genetically proxied g-Factor and metabolites. Furthermore, the mediation analysis via two-step MR revealed that the causal pathway from acetylornithine to bipolar disorder was partly mediated by g-Factor, with a mediated proportion of 37.1%. Besides, g-Factor mediated the causal pathway from butyrylcarnitine to schizophrenia, with a mediated proportion of 37.5%. Other neurocognitive traits from different sources provided consistent findings. Conclusion Our results provide genetic evidence that acetylornithine protects against bipolar disorder through neurocognitive abilities, while butyrylcarnitine has an adverse effect on schizophrenia through neurocognition. These findings may provide insight into interventions at the metabolic level for risk of neurocognitive and related disorders.

Abstract Deep vein thrombosis (DVT) is a leading cause of morbidity and mortality after trauma. Here, we integrate plasma metabolomics and proteomics to evaluate the metabolic alterations and their function in up to 680 individuals with and without DVT after trauma (pt-DVT). We identify 28 metabolites and 2 clinical parameter clusters associated with pt-DVT. Then, we develop a panel of 9 metabolites (hexadecanedioic acid, pyruvic acid, L-Carnitine, serotonin, PE(P-18:1(11Z)/18:2(9Z,12Z)), 3-Hydroxycapric acid, 5,6-DHET, 3-Methoxybenzenepropanoic acid and pentanenitrile) that can predict pt-DVT with high performance, which can be verified in an independent cohort. Furthermore, the integration analysis of metabolomics and proteomics data indicates that the upregulation of glycolysis/gluconeogenesis-TCA cycle may promote thrombosis by regulating ROS levels in red blood cells, suggesting that interfering with this process might be potential therapeutic strategies for pt-DVT. Together, our study comprehensively delineates the metabolic and hematological dysregulations for pt-DVT, and provides potential biomarkers for early detection.

Abstract Background Decision fatigue is a new concept in the field of psychology and refers to a state of fatigue alongside impaired cognitive processing and emotional regulation ability. Previous studies have confirmed that nurses are prone to decision fatigue, and nurses who experience decision fatigue may implement nursing measures that are inconsistent with clinical evidence, thus affecting patients’ benefits. COVID-19, as a large-scale global public health emergency, increased the workload and burden of nurses and aggravated decision fatigue. However, the factors leading to decision fatigue among nurses have not yet been identified. Methods This study is guided by interpretative phenomenology. During the epidemic period of COVID-19: From November 2022 to February 2023, a one-to-one, semi-structured in-depth interview was conducted among nurses with decision fatigue experience who were participating in front-line work in Jilin Province using homogenous sampling. The interview recordings and related data were transcribed into text within 24 h, and data analysis was assisted by NVivo 12.0 software. Results After a total of 14 front-line nurses were analyzed in this study, The thematic level reaches saturation, the findings present a persuasive and coherent narrative, and the study is terminated, and finally extracted and formed three core themes: “Cognition, influence and attitude of decision fatigue”, “Approaching factors of decision fatigue” and “Avoidant factors of decision fatigue”. Conclusion This study confirmed that decision fatigue was widespread in the work of front-line nurses, affecting the physical and psychological health of nurses, the quality of nursing work, the degree of benefit of patients and the clinical outcome. However, nursing staff do not know enough about decision fatigue, so the popularization and research of decision fatigue should be strengthened. Improve the attention of medical institutions, nursing managers and nursing staff.Some suggestions are put forward for the intervention of decision fatigue through personnel, task, tool and technology, organization and environment.

Abstract The human brain has been implicated in the pathogenesis of several complex diseases. Taking advantage of single-cell techniques, genome-wide association studies (GWAS) have taken it a step further and revealed brain cell-type-specific functions for disease loci. However, genetic causal associations inferred by Mendelian randomization (MR) studies usually include all instrumental variables from GWAS, which hampers the understanding of cell-specific causality. Here, we developed an analytical framework, Cell-Stratified MR (csMR), to investigate cell-stratified causality through colocalizing GWAS signals with single-cell eQTL from different brain cells. By applying to obesity-related traits, our results demonstrate the cell-type-specific effects of GWAS variants on gene expression, and indicate the benefits of csMR to identify cell-type-specific causal effect that is often hidden from bulk analyses. We also found csMR valuable to reveal distinct causal pathways between different obesity indicators. These findings suggest the value of our approach to prioritize target cells for extending genetic causation studies.

Abstract Ovarian cancer is a major gynecological cancer that has poor prognosis associated mainly to its late diagnosis. Cisplatin is an FDA approved ovarian cancer therapy and even though the therapy is initially promising, the patients mostly progress to resistance against cisplatin. The underlying mechanisms are complex and not very clearly understood. Using two different paired cell lines representing cisplatin-sensitive and the cisplatin-resistant ovarian cancer cells, the ES2 and the A2780 parental and cisplatin-resistant cells, we show an elevated proto-oncogene c-Myb in resistant cells. We further show down-regulated lncRNA NKILA in resistant cells with its de-repression in resistant cells when c-Myb is silenced. NKILA negatively correlates with cancer cell and invasion but has no effect on cellular proliferation or cell cycle. C-Myb activates NF-κB signaling which is inhibited by NKILA. The cisplatin resistant cells are also marked by upregulated stem cell markers, particularly LIN28A and OCT4, and downregulated LIN28A-targeted let-7 family miRNAs. Whereas LIN28A and downregulated let-7s individually de-repress c-Myb-mediated cisplatin resistance, the ectopic expression of let-7s attenuates LIN28A effects, thus underlying a c-Myb-NKILA-LIN28A-let-7 axis in cisplatin resistance of ovarian cancer cells that needs to be further explored for therapeutic intervention.

Abstract The synovium is an important component of any synovial joint and is the major target tissue of inflammatory arthritis. However, the multi-omics landscape of synovium required for functional inference is absent from large-scale resources. Here we integrate genomics with transcriptomics and chromatin accessibility features of human synovium in up to 245 arthritic patients, to characterize the landscape of genetic regulation on gene expression and the regulatory mechanisms mediating arthritic diseases predisposition. We identify 4765 independent primary and 616 secondary cis-expression quantitative trait loci (cis-eQTLs) in the synovium and find that the eQTLs with multiple independent signals have stronger effects and heritability than single independent eQTLs. Integration of genome-wide association studies (GWASs) and eQTLs identifies 84 arthritis related genes, revealing 38 novel genes which have not been reported by previous studies using eQTL data from the GTEx project or immune cells. We further develop a method called eQTac to identify variants that could affect gene expression by affecting chromatin accessibility and identify 1517 regions with potential regulatory function of chromatin accessibility. Altogether, our study provides a comprehensive synovium multi-omics resource for arthritic diseases and gains new insights into the regulation of gene expression.

Abstract Background Stroke is a major cause of mortality and long-term disability worldwide. Whether the associations between brain imaging-derived phenotypes (IDPs) and stroke are causal is uncertain. Methods We performed two-sample bidirectional Mendelian randomization (MR) analyses to explore the causal associations between IDPs and stroke. Summary data of 587 brain IDPs (up to 33,224 individuals) from the UK Biobank and five stroke types (sample size range from 301,663 to 446,696, case number range from 5,386 to 40,585) from the MEGASTROKE consortium were used. Results Forward MR indicated 14 IDPs belong to projection fibers or association fibers were associated with stroke. For example, higher genetically determined mean diffusivity (MD) in the right external capsule was causally associated with an increased risk of small vessel stroke (IVW OR = 2.76, 95% CI 2.07 to 3.68, P = 5.87 × 10−12). Reverse MR indicated that genetically determined higher risk of any ischemic stroke was associated with increased isotropic or free water volume fraction (ISOVF) in body of corpus callosum (IVW β = 0.23, 95% CI 0.14 to 0.33, P = 3.22 × 10−7). This IDP is a commissural fiber and it is not included in the IDPs identified by forward MR. Conclusions We identified 14 IDPs with statistically significant evidence of causal effects on stroke or stroke subtypes. We also identified potential causal effects of stroke on one IDP of commissural fiber. These findings might guide further work toward identifying preventative strategies at the brain imaging levels.

Abstract Human mesenchymal stem cells (hMSCs) can be differentiated into osteoblasts and adipocytes. During these processes, super enhancers (SEs) play important roles. Here, we performed comprehensive characterization of the SEs changes associated with adipogenic and osteogenic differentiation of hMSCs, and revealed that SEs changed more dramatically compared with typical enhancers. We identified a set of lineage-selective SEs, whose target genes were enriched with cell type-specific functions. Functional experiments in lineage-selective SEs demonstrated their specific roles in directed differentiation of hMSCs. We also found that some key transcription factors regulated by lineage-selective SEs could form core regulatory circuitry (CRC) to regulate each other’s expression and control the hMSCs fate determination. In addition, we found that GWAS SNPs of osteoporosis and obesity were significantly enriched in osteoblasts-selective SEs or adipocytes-selective SEs, respectively. Taken together, our studies unveiled important roles of lineage-selective SEs in hMSCs differentiation into osteoblasts and adipocytes.

Abstract Background Clustering the metagenomic contigs into potential genomes is a key step to investigate the functional roles of microbial populations. Existing algorithms have achieved considerable success with simulated or real sequencing datasets. However, accurately classifying contigs from complex metagenomes is still a challenge. Results We introduced a novel clustering algorithm, MetaDecoder, which can classify metagenomic contigs based on the frequencies of k-mers and coverages. MetaDecoder was built as a two-layer model with the first layer being a GPU-based modified Dirichlet process Gaussian mixture model (DPGMM), which controls the weight of each DPGMM cluster to avoid over-segmentation by dynamically dissolving contigs in small clusters and reassigning them to the remaining clusters. The second layer comprises a semi-supervised k-mer frequency probabilistic model and a modified Gaussian mixture model for modeling the coverage based on single copy marker genes. Benchmarks on simulated and real-world datasets demonstrated that MetaDecoder can be served as a promising approach for effectively clustering metagenomic contigs. Conclusions In conclusion, we developed the GPU-based MetaDecoder for effectively clustering metagenomic contigs and reconstructing microbial communities from microbial data. Applying MetaDecoder on both simulated and real-world datasets demonstrated that it could generate more complete clusters with lower contamination. Using MetaDecoder, we identified novel high-quality genomes and expanded the existing catalog of bacterial genomes. Video Abstract

Summary: Background: Accumulative evidences have shown that dysregulation of biological pathways contributed to the initiation and progression of malignant tumours. Several methods for pathway activity measurement have been proposed, but they are restricted to making comparisons between groups or sensitive to experimental batch effects. Methods: We introduced a novel method for individualized pathway activity measurement (IPAM) that is based on the ranking of gene expression levels in individual sample. Taking advantage of IPAM, we calculated the pathway activity of 318 pathways from KEGG database in the 10528 tumour/normal samples of 33 cancer types from TCGA to identify characteristic dysregulated pathways among different cancer types. Findings: IPAM precisely quantified the level of activity of each pathway in pan-cancer analysis and exhibited better performance in cancer classification and prognosis prediction over five widely used tools. The average ROC-AUC of cancer diagnostic model using tumour-educated platelets (TEPs) reached 92.84%, suggesting the potential of our algorithm in early diagnosis of cancer. We identified several pathways significantly deregulated and associated with patient survival in a large fraction of cancer types, such as tyrosine metabolism, fatty acid degradation, cell cycle, p53 signalling pathway and DNA replication. We also confirmed the dominant role of metabolic pathways in cancer pathway dysregulation and identified the driving factors of specific pathway dysregulation, such as PPARA for branched-chain amino acid metabolism and NR1I2, NR1I3 for fatty acid metabolism. Interpretation: Our study will provide novel clues for understanding the pathological mechanisms of cancer, ultimately paving the way for personalized medicine of cancer. Funding: A full list of funding can be found in the Acknowledgements section.

Abstract Background Childhood obesity is reported to be associated with the risk of many diseases in adulthood. However, observational studies cannot fully account for confounding factors. We aimed to systematically assess the causal associations between childhood body mass index (BMI) and various adult traits/diseases using two-sample Mendelian randomization (MR). Methods After data filtering, 263 adult traits genetically correlated with childhood BMI (P 0.05). For dietary habits, different from conventional understanding, we found that higher childhood BMI was associated with low calorie density food intake. However, this association might be specific to the UK Biobank population. Conclusions In summary, we provided a phenome-wide view of the effects of childhood BMI on adult traits. Multivariable MR analysis suggested that the associations between childhood BMI and increased risks of diseases in adulthood are likely attributed to individuals remaining obese in later life. Therefore, ensuring that childhood obesity does not persist into later life might be useful for reducing the detrimental effects of childhood obesity on adult diseases.

Detecting SNPs associated with drug efficacy or toxicity is helpful to facilitate personalized medicine. Previous studies usually find SNPs associated with clinical outcome only in patients received a specific treatment. However, without information from patients without drug treatment, it is possible that the detected SNPs are associated with patients’ clinical outcome even without drug treatment. Here we aimed to detect drug response SNPs based on data from patients with and without drug treatment through combing the cox proportional-hazards model and pairwise Kaplan-Meier survival analysis. A pipeline named Detection of Drug Response SNPs (DDRS) was built and applied to TCGA breast cancer data including 363 patients with doxorubicin treatment and 321 patients without any drug treatment. We identified 548 doxorubicin associated SNPs. Drug response score derived from these SNPs were associated with drug-resistant level (indicated by IC50) of breast cancer cell lines. Enrichment analyses showed that these SNPs were enriched in active epigenetic regulation markers (e.g., H3K27ac). Compared with random genes, the cis-eQTL genes of these SNPs had a shorter protein–protein interaction distance to doxorubicin associated genes. In addition, linear discriminant analysis showed that the eQTL gene expression levels could be used to predict clinical outcome for patients with doxorubicin treatment (AUC = 0.738). Specifically, we identified rs2817101 as a drug response SNP for doxorubicin treatment. Higher expression level of its cis-eQTL gene GSTA1 is associated with poorer survival. This approach can also be applied to identify new drug associated SNPs in other cancers.

Although genome-wide association studies (GWASs) have successfully identified thousands of risk variants for human complex diseases, understanding the biological function and molecular mechanisms of the associated SNPs involved in complex diseases is challenging. Here we developed a framework named integrative multi-omics network-based approach (IMNA), aiming to identify potential key genes in regulatory networks by integrating molecular interactions across multiple biological scales, including GWAS signals, gene expression-based signatures, chromatin interactions and protein interactions from the network topology. We applied this approach to breast cancer, and prioritized key genes involved in regulatory networks. We also developed an abnormal gene expression score (AGES) signature based on the gene expression deviation of the top 20 rank-ordered genes in breast cancer. The AGES values are associated with genetic variants, tumor properties and patient survival outcomes. Among the top 20 genes, RNASEH2A was identified as a new candidate gene for breast cancer. Thus, our integrative network-based approach provides a genetic-driven framework to unveil tissue-specific interactions from multiple biological scales and reveal potential key regulatory genes for breast cancer. This approach can also be applied in other complex diseases such as ovarian cancer to unravel underlying mechanisms and help for developing therapeutic targets.

More than 90% of autoimmune-associated variants are located in noncoding regions, leading to challenges in deciphering the underlying causal roles of functional variants and genes and biological mechanisms. Therefore, to reduce the gap between traditional genetic findings and mechanistic understanding of disease etiologies and clinical drug development, it is important to translate systematically the regulatory mechanisms underlying noncoding variants. Here, we prioritized functional noncoding SNPs with regulatory gene targets associated with 19 autoimmune diseases by incorporating hundreds of immune cell–specific multiomics data. The prioritized SNPs are associated with transcription factor (TF) binding, histone modification, or chromatin accessibility, indicating their allele-specific regulatory roles. Their target genes are significantly enriched in immunologically related pathways and other known immunologically related functions. We found that 90.1% of target genes are regulated by distal SNPs involving several TFs (e.g., the DNA-binding protein CCCTC-binding factor [CTCF]), suggesting the importance of long-range chromatin interaction in autoimmune diseases. Moreover, we predicted potential drug targets for autoimmune diseases, including 2 genes (NFKB1 and SH2B3) with known drug indications on other diseases, highlighting their potential drug repurposing opportunities. Taken together, these findings may provide useful information for future experimental follow-up and drug applications on autoimmune diseases.

Inflammation plays an important role in nerve defects caused by intracerebral hemorrhage. Repairing brain damage by inhibiting the macrophage-inducible C-type lectin/spleen tyrosine kinase (Mincle/Syk) signaling pathway is a potential new target for treating cerebral hemorrhage. In this study, we aimed to determine whether acupuncture through Baihui (DU20) to Qubin (GB7) is an effective treatment for intracerebral hemorrhage through the Mincle/Syk signaling pathway. An intracerebral hemorrhage rat model was established by autologous blood infusion into the caudate nucleus. Acupuncture through Baihui to Qubin was performed for 30 minutes, once every 12 hours, for a total of three times. Piceatannol (34.62 mg/kg), a Syk inhibitor, was intraperitoneally injected as a control. Modified neurological severity score was used to assess neurological function. Brain water content was measured. Immunohistochemistry and western blot assay were used to detect immunoreactivity and protein expression levels of Mincle, Syk, and CARD9. Real-time polymerase chain reaction was used to determine interleukin-1β mRNA levels. Hematoxylin-eosin staining was performed to observe histopathological changes. Our results showed that acupuncture through Baihui to Qubin remarkably improved neurological function and brain water content, and inhibited immunoreactivity and expression of Mincle, Syk, CARD9, and interkeukin-1β. Moreover, this effect was similar to piceatannol. These findings suggest that acupuncture through Baihui to Qubin can improve neurological impairment after cerebral hemorrhage by inhibiting the Mincle/Syk signaling pathway.

The removal of boron (B) from water by co-precipitation with hydroxyapatite (HAP) has been extensively studied due to its low cost, ease of use and high efficiency. However, there is no explicit mechanism to express how resolved B was trapped by HAP. Thus, in this work, the process of removing B from water was studied using a low-cost calcium (Ca) precipitation agent derived from used waste oyster shells. The results showed that the removal rate of B in the simulated wastewater by calcined oyster shell (COS) in the presence of phosphorus (P) is up to more than 90%, as opposed to virtually no removal without phosphate. For B removal, the treated water needs to be an alkaline solution with a high pH above 12, where B is removed as [CaB(OH)4]+ but is not molecular. Finally, the synergistic mechanism of co-precipitation between HAP and dissolved B, occlusion co-precipitation, was explained in detail. The proposed method discovered the relationship between Ca, P and B, and was aimed at removing B without secondary pollution through co-precipitation.

Genome-wide association studies (GWASs) have successfully identified thousands of susceptibility loci for human complex diseases. However, missing heritability is still a challenging problem. Considering most GWAS loci are located in regulatory elements, we recently developed a pipeline named functional disease-associated single-nucleotide polymorphisms (SNPs) prediction (FDSP), to predict novel susceptibility loci for complex diseases based on the interpretation of regulatory features and published GWAS results with machine learning. When applied to type 2 diabetes and hypertension, the predicted susceptibility loci by FDSP were proved to be capable of explaining additional heritability. In addition, potential target genes of the predicted positive SNPs were significantly enriched in disease-related pathways. Our results suggested that taking regulatory features into consideration might be a useful way to address the missing heritability problem. We hope FDSP could offer help for the identification of novel susceptibility loci for complex diseases.

Magnolia sinostellata Chiu & Chen is a rare and endangered species endemic to subtropical China. Here we assembled and annotated the complete chloroplast (cp) genome of M. sinostellata. The chloroplast genome of M. sinostellata is 160,076 bp in length and encodes 79 protein-coding genes, 30 transfer RNA (tRNA) genes and four ribosomal RNA (rRNA) genes. The maximum-likelihood (ML) phylogenetic analysis result reveals that M. sinostellata is most closely related to M. biondii.

DDR2 gene, playing an essential role in regulating osteoblast differentiation and chondrocyte maturation, may influence bone mineral density (BMD) and osteoporosis, but the genetic variations actually leading to the association remain to be elucidated. Therefore, the aim of this study was to investigate whether the genetic variants in DDR2 are associated with BMD and fracture risk. This study was performed in three samples from two ethnicities, including 1,300 Chinese Han subjects, 700 Chinese Han subjects (350 with osteoporotic hip fractures and 350 healthy controls) and 2,286 US white subjects. Twenty-eight SNPs in DDR2 were genotyped and tested for associations with hip BMD and fractures. We identified 3 SNPs in DDR2 significantly associated with hip BMD in the Chinese population after multiple testing adjustments, which were rs7521233 (P = 1.06×10-4, β: -0.018 for allele C), rs7553831 (P = 1.30×10-4, β: -0.018 for allele T), and rs6697469 (P = 1.59×10-3, β: -0.015 for allele C), separately. These three SNPs were in high linkage disequilibrium. Haplotype analyses detected two significantly associated haplotypes, including one haplotype in block 2 (P = 9.54×10-4, β: -0.016) where these three SNPs located. SNP rs6697469 was also associated with hip fractures (P = 0.043, OR: 1.42) in the Chinese population. The effect on fracture risk was consistent with its association with lower BMD. However, in the white population, we didn't observe significant associations with hip BMD. eQTL analyses revealed that SNPs associated with BMD also affected DDR2 mRNA expression levels in Chinese. Our findings, together with the prior biological evidence, suggest that DDR2 could be a new candidate for osteoporosis in Chinese population. Our results also reveal an ethnic difference, which highlights the need for further genetic studies in each ethnic group.

Osteoporosis is a major public health problem. It is mainly characterized by low bone mineral density (BMD) and/or low-trauma osteoporotic fractures (OF), both of which have strong genetic determination. The specific genes influencing these phenotypic traits, however, are largely unknown. Using the Affymetrix 500K array set, we performed a case-control genome-wide association study (GWAS) in 700 elderly Chinese Han subjects (350 with hip OF and 350 healthy matched controls). A follow-up replication study was conducted to validate our major GWAS findings in an independent Chinese sample containing 390 cases with hip OF and 516 controls. We found that a SNP, rs13182402 within the ALDH7A1 gene on chromosome 5q31, was strongly associated with OF with evidence combined GWAS and replication studies (P = 2.08x10(-9), odds ratio = 2.25). In order to explore the target risk factors and potential mechanism underlying hip OF risk, we further examined this candidate SNP's relevance to hip BMD both in Chinese and Caucasian populations involving 9,962 additional subjects. This SNP was confirmed as consistently associated with hip BMD even across ethnic boundaries, in both Chinese and Caucasians (combined P = 6.39x10(-6)), further attesting to its potential effect on osteoporosis. ALDH7A1 degrades and detoxifies acetaldehyde, which inhibits osteoblast proliferation and results in decreased bone formation. Our findings may provide new insights into the pathogenesis of osteoporosis.

Video captioning aims to automatically describe a video clip with informative sentences. At present, deep learning-based models have become the mainstream for this task and achieved competitive results on public datasets. Usually, these methods leverage different types of features to generate sentences, e.g., semantic information, 2D or 3D features.However, some methods only treat semantic information as a complement of visual representations and cannot fully exploit it; some of them ignore the relationship between different types of features. In addition, most of them select multiple frames of a video with an equally spaced sampling scheme, resulting in much redundant information. To address these issues, we present a novel video-captioning framework, Semantic Enhanced video captioning with Multi-feature Fusion, SEMF for short. It optimizes the use of different types of features from three aspects. First, a semantic encoder is designed to enhance meaningful semantic features through a semantic dictionary to boost performance. Second, a discrete selection module pays attention to important features and obtains different contexts at different steps to reduce feature redundancy. Finally, a multi-feature fusionmodule uses a novel relation-aware attentionmechanism to separate the common and complementary components of different features to provide more effective visual features for the next step. Moreover, the entire framework can be trained in an end-to-endmanner. Extensive experiments are conducted on Microsoft Research Video Description Corpus (MSVD) and MSR-Video to Text (MSR-VTT) datasets. The results demonstrate that SEMF is able to achieve state-of-the-art results. [ABSTRACT FROM AUTHOR]