A major bottleneck in training end-to-end task-oriented dialog system is the lack of data. To utilize limited training data more efficiently, we propose Modular Supervision Network (MOSS), an encoder-decoder training framework that could incorporate supervision from various intermediate dialog system modules including natural language understanding, dialog state tracking, dialog policy learning and natural language generation. With only 60% of the training data, MOSS-all (i.e., MOSS with supervision from all four dialog modules) outperforms state-of-the-art models on CamRest676. Moreover, introducing modular supervision has even bigger benefits when the dialog task has a more complex dialog state and action space. With only 40% of the training data, MOSS-all outperforms the state-of-the-art model on a complex laptop network trouble shooting dataset, LaptopNetwork, that we introduced. LaptopNetwork consists of conversations between real customers and customer service agents in Chinese. Moreover, MOSS framework can accommodate dialogs that have supervision from different dialog modules at both framework level and model level. Therefore, MOSS is extremely flexible to update in real-world deployment.
Objective. This study is aimed at understanding the molecular mechanisms and exploring potential therapeutic targets for atrial fibrillation (AF) by multiomics analysis. Methods. Transcriptomics and methylation data of AF patients were retrieved from the Gene Expression Omnibus (GEO). Differentially expressed genes (DEGs) and differentially methylated sites between AF and normal samples were screened. Then, highly expressed and hypomethylated and lowly expressed and hypermethylated genes were identified for AF. Weighted gene coexpression network analysis (WGCNA) was presented to construct AF-related coexpression networks. 52 AF blood samples were used for whole exome sequence. The mutation was visualized by the maftools package in R. Key genes were validated in AF using independent datasets. Results. DEGs were identified between AF and controls, which were enriched in neutrophil activation and regulation of actin cytoskeleton. RHOA, CCR2, CASP8, and SYNPO2L exhibited abnormal expression and methylation, which have been confirmed to be related to AF. PCDHA family genes had high methylation and low expression in AF. We constructed two AF-related coexpression modules. Single-nucleotide polymorphism (SNP) was the most common mutation type in AF, especially T > C . MUC4 was the most frequent mutation gene, followed by PHLDA1, AHNAK2, and MAML3. There was no statistical difference in expression of AHNAK2 and MAML3, for AF. PHLDA1 and MUC4 were confirmed to be abnormally expressed in AF. Conclusion. Our findings identified DEGs related to DNA methylation and mutation for AF, which may offer possible therapeutic targets and a new insight into the pathogenesis of AF from a multiomics perspective.
Objective. Multiple genes have been identified to cause dilated cardiomyopathy (DCM). Nevertheless, there is still a lack of comprehensive elucidation of the molecular characteristics for DCM. Herein, we aimed to uncover putative molecular features for DCM by multiomics analysis. Methods. Differentially expressed genes (DEGs) were obtained from different RNA sequencing (RNA-seq) datasets of left ventricle samples from healthy donors and DCM patients. Furthermore, protein-protein interaction (PPI) analysis was then presented. Differentially methylated genes (DMGs) were identified between DCM and control samples. Following integration of DEGs and DMGs, differentially expressed and methylated genes were acquired and their biological functions were analyzed by the clusterProfiler package. Whole exome sequencing of blood samples from 69 DCM patients was constructed in our cohort, which was analyzed the maftools package. The expression of key mutated genes was verified by three independent datasets. Results. 1407 common DEGs were identified for DCM after integration of the two RNA-seq datasets. A PPI network was constructed, composed of 171 up- and 136 downregulated genes. Four hub genes were identified for DCM, including C3 ( degree = 24 ), GNB3 ( degree = 23 ), QSOX1 ( degree = 21 ), and APOB ( degree = 17 ). Moreover, 285 hyper- and 321 hypomethylated genes were screened for DCM. After integration, 20 differentially expressed and methylated genes were identified, which were associated with cell differentiation and protein digestion and absorption. Among single-nucleotide variant (SNV), C>T was the most frequent mutation classification for DCM. MUC4 was the most frequent mutation gene which occupied 71% across 69 samples, followed by PHLDA1, AHNAK2, and MAML3. These mutated genes were confirmed to be differentially expressed between DCM and control samples. Conclusion. Our findings comprehensively analyzed molecular characteristics from the transcriptome, epigenome, and genome perspectives for DCM, which could provide practical implications for DCM.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.