Sesame(Sesamum indicum L.)is an ancient oilseed crop of the Pedaliaceae family with high oil content and potential health benefits.SHI RELATED SEQUENCE(SRS)proteins are the transcription factors(TFs)specific to plants...Sesame(Sesamum indicum L.)is an ancient oilseed crop of the Pedaliaceae family with high oil content and potential health benefits.SHI RELATED SEQUENCE(SRS)proteins are the transcription factors(TFs)specific to plants that contain RING-like zinc finger domain and are associated with the regulation of several physiological and biochemical processes.They also play vital roles in plant growth and development such as root formation,leaf development,floral development,hormone biosynthesis,signal transduction,and biotic and abiotic stress responses.Nevertheless,the SRS gene family was not reported in sesame yet.In this study,identification,molecular characterization,phylogenetic relationship,cis-acting regulatory elements,protein-protein interaction,syntenic relationship,duplication events and expression pattern of SRS genes were analyzed in S.indicum.We identified total six SiSRS genes on seven different linkage groups in the S.indicum genome by comparing with the other species,including the model plant Arabidopsis thaliana.The SiSRS genes showed variation in their structure like2–5 exons and 1–4 introns.Like other species,SiSRS proteins also contained‘RING-like zinc finger'and‘LRP1'domains.Then,the SiSRS genes were clustered into subclasses via phylogenetic analysis with proteins of S.indicum,A.thaliana,and some other plant species.The cis-acting regulatory elements analysis revealed that the promoter region of SiSRS4(SIN_1011561)showed the highest 13 and 16 elements for light-and phytohormone-responses whereas,SiSRS1(SIN_1015187)showed the highest 15 elements for stress-response.The ABREs,or ABA-responsive elements,were found in a maximum of 8 copies in the SiSRS3(SIN 1009100).Moreover,the available RNA-seq based expression of SiSRS genes revealed variation in expression patterns between stress-treated and non-treated samples,especially in drought and salinity conditions in.S.indicum.Two SiSRS genes like SiSRS1(SIN_1015187)and SiSRS5(SIN_1021065),also exhibited variable expression patterns between control vs PEG-treated sesame root samples and three SiSRS genes,including SiSRS1(SIN_1015187),SiSRS2(SIN_1003328)and SiSRS5(SIN_1021065)were responsive to salinity treatments.The present outcomes will encourage more research into the gene expression and functionality analysis of SiSRS genes in S.indicum and other related species.展开更多
目的虽然深度学习技术已大幅提高了图像超分辨率的性能,但是现有方法大多仅考虑了特定的整数比例因子,不能灵活地实现连续比例因子的超分辨率。现有方法通常为每个比例因子训练一次模型,导致耗费很长的训练时间和占用过多的模型存储空...目的虽然深度学习技术已大幅提高了图像超分辨率的性能,但是现有方法大多仅考虑了特定的整数比例因子,不能灵活地实现连续比例因子的超分辨率。现有方法通常为每个比例因子训练一次模型,导致耗费很长的训练时间和占用过多的模型存储空间。针对以上问题,本文提出了一种基于跨尺度耦合网络的连续比例因子超分辨率方法。方法提出一个用于替代传统上采样层的跨尺度耦合上采样模块,用于实现连续比例因子上采样。其次,提出一个跨尺度卷积层,可以在多个尺度上并行提取特征,通过动态地激活和聚合不同尺度的特征来挖掘跨尺度上下文信息,有效提升连续比例因子超分辨率任务的性能。结果在3个数据集上与最新的超分辨率方法进行比较,在连续比例因子任务中,相比于性能第2的对比算法Meta-SR(meta super-resolution),峰值信噪比提升达0.13 d B,而参数量减少了73%。在整数比例因子任务中,相比于参数量相近的轻量网络SRFBN(super-resolution feedback network),峰值信噪比提升达0.24 d B。同时,提出的算法能够生成视觉效果更加逼真、纹理更加清晰的结果。消融实验证明了所提算法中各个模块的有效性。结论本文提出的连续比例因子超分辨率模型,仅需要一次训练,就可以在任意比例因子上获得优秀的超分辨率结果。此外,跨尺度耦合上采样模块可以用于替代常用的亚像素层或反卷积层,在实现连续比例因子上采样的同时,保持模型性能。展开更多
文摘Sesame(Sesamum indicum L.)is an ancient oilseed crop of the Pedaliaceae family with high oil content and potential health benefits.SHI RELATED SEQUENCE(SRS)proteins are the transcription factors(TFs)specific to plants that contain RING-like zinc finger domain and are associated with the regulation of several physiological and biochemical processes.They also play vital roles in plant growth and development such as root formation,leaf development,floral development,hormone biosynthesis,signal transduction,and biotic and abiotic stress responses.Nevertheless,the SRS gene family was not reported in sesame yet.In this study,identification,molecular characterization,phylogenetic relationship,cis-acting regulatory elements,protein-protein interaction,syntenic relationship,duplication events and expression pattern of SRS genes were analyzed in S.indicum.We identified total six SiSRS genes on seven different linkage groups in the S.indicum genome by comparing with the other species,including the model plant Arabidopsis thaliana.The SiSRS genes showed variation in their structure like2–5 exons and 1–4 introns.Like other species,SiSRS proteins also contained‘RING-like zinc finger'and‘LRP1'domains.Then,the SiSRS genes were clustered into subclasses via phylogenetic analysis with proteins of S.indicum,A.thaliana,and some other plant species.The cis-acting regulatory elements analysis revealed that the promoter region of SiSRS4(SIN_1011561)showed the highest 13 and 16 elements for light-and phytohormone-responses whereas,SiSRS1(SIN_1015187)showed the highest 15 elements for stress-response.The ABREs,or ABA-responsive elements,were found in a maximum of 8 copies in the SiSRS3(SIN 1009100).Moreover,the available RNA-seq based expression of SiSRS genes revealed variation in expression patterns between stress-treated and non-treated samples,especially in drought and salinity conditions in.S.indicum.Two SiSRS genes like SiSRS1(SIN_1015187)and SiSRS5(SIN_1021065),also exhibited variable expression patterns between control vs PEG-treated sesame root samples and three SiSRS genes,including SiSRS1(SIN_1015187),SiSRS2(SIN_1003328)and SiSRS5(SIN_1021065)were responsive to salinity treatments.The present outcomes will encourage more research into the gene expression and functionality analysis of SiSRS genes in S.indicum and other related species.
文摘目的虽然深度学习技术已大幅提高了图像超分辨率的性能,但是现有方法大多仅考虑了特定的整数比例因子,不能灵活地实现连续比例因子的超分辨率。现有方法通常为每个比例因子训练一次模型,导致耗费很长的训练时间和占用过多的模型存储空间。针对以上问题,本文提出了一种基于跨尺度耦合网络的连续比例因子超分辨率方法。方法提出一个用于替代传统上采样层的跨尺度耦合上采样模块,用于实现连续比例因子上采样。其次,提出一个跨尺度卷积层,可以在多个尺度上并行提取特征,通过动态地激活和聚合不同尺度的特征来挖掘跨尺度上下文信息,有效提升连续比例因子超分辨率任务的性能。结果在3个数据集上与最新的超分辨率方法进行比较,在连续比例因子任务中,相比于性能第2的对比算法Meta-SR(meta super-resolution),峰值信噪比提升达0.13 d B,而参数量减少了73%。在整数比例因子任务中,相比于参数量相近的轻量网络SRFBN(super-resolution feedback network),峰值信噪比提升达0.24 d B。同时,提出的算法能够生成视觉效果更加逼真、纹理更加清晰的结果。消融实验证明了所提算法中各个模块的有效性。结论本文提出的连续比例因子超分辨率模型,仅需要一次训练,就可以在任意比例因子上获得优秀的超分辨率结果。此外,跨尺度耦合上采样模块可以用于替代常用的亚像素层或反卷积层,在实现连续比例因子上采样的同时,保持模型性能。
