Plant DNA Methylation: An Epigenetic Mark in Development,Environmental Interactions,and Evolution

DNA methylation is an epigenetic modification of the genome involved in the regulation of gene expression and modulation of chromatin structure. Plant genomes are widely methylated, and the methylation generally occurs on the cytosine bases through the activity of specific enzymes called DNA methyltransferases. On the other hand, methylated DNA can also undergo demethylation through the action of demethylases. The methylation landscape is finely tuned and assumes a pivotal role in plant development and evolution. This review illustrates different molecular aspects of DNA methylation and some plant physiological processes influenced by this epigenetic modification in model species, crops, and ornamental plants such as orchids. In addition, this review aims to describe the relationship between the changes in plant DNA methylation levels and the response to biotic and abiotic stress. Finally, we discuss the possible evolutionary implications and biotechnological applications of DNA methylation.     

Reduced VDAC1, Maintained Mitochondrial Dynamics and Enhanced Mitochondrial Biogenesis in a Transgenic Tau Mouse Model of Alzheimer’s Disease

Alzheimer’s disease (AD) is one of the most common forms of neurodegeneration, defined by reduced cognitive function, which is caused by the gradual death of neurons in the brain. Recent studies have shown an age-dependent rise in the levels of voltage-dependent anion channel 1 (VDAC1) in AD. In addition, we discovered an aberrant interaction between VDAC1 and P-TAU in the brains of AD patients, which led to abnormalities in the structural and functional integrity of the mitochondria. The purpose of our study is to understand the protective effects of reduced VDAC1 against impaired mitochondrial dynamics and defective mitochondrial biogenesis in transgenic TAU mice. Recently, we crossed heterozygote VDAC1 knockout (VDAC1^+/−) mice with transgenic TAU mice to obtain double-mutant VDAC1^+/−/TAU mice. Our goal was to evaluate whether a partial decrease in VDAC1 lessens the amount of mitochondrial toxicity in transgenic Tau (P301L) mice. We found that mitochondrial fission proteins were significantly reduced, and mitochondrial fusion and biogenesis proteins were increased in double-mutant mice compared to TAU mice. On the basis of these discoveries, the current work may have significance for the development of reduced-VDAC1-based treatments for individuals suffering from AD as well as other tauopathies.     

双波段彩色融合图像色彩和谐性客观评价

针对现有的图像质量评价方法较少利用人眼视网膜和视觉皮层的颜色编码机制,并且未能充分考虑图像色彩信息对图像质量的影响,提出了一种基于多视觉特征的可见光(微光)与红外彩色融合图像色彩和谐性客观评价模型.该模型在图像质量评估中融入了更多的颜色信息,综合考虑多种人眼视觉特征包括视觉对立色彩特征、色彩信息波动特征和高级视觉内容特征,经过特征融合和支持向量回归训练,实现彩色融合图像的色彩和谐性客观评价.采用3种典型场景融合图像数据库进行实验比较与分析.实验结果表明,与现有的8种图像质量客观评价方法相比,所提出的方法与人眼主观感受更加一致,具有较高的预测准确度.     

跨尺度混合注意力的遥感图像超分辨率重建

为了解决现有遥感图像超分辨率重建模型对长期特征相似性和多尺度特征相关性关注不足的问题, 提出了一种基于跨尺度混合注意力机制的遥感图像超分辨率重建算法. 首先提出了一个全局层注意力机制(global layer attention, GLA), 利用层注意力机制加权融合不同层级的全局特征, 建模低分辨率与高分辨率图像特征间的长期依赖关系. 同时, 设计了跨尺度局部注意力机制(cross-scale local attention, CSLA), 在多尺度的低分辨率特征图中寻找与高分辨率图像匹配的局部信息补丁, 并融合不同尺度的补丁特征, 以优化模型对图像细节信息的恢复能力. 最后, 提出一种局部信息感知损失函数来指导图像的重建过程, 进一步提高了重建图像的视觉质量和细节保留能力. 在UC-Merced数据集上的实验结果表明, 本文方法在3种放大倍数下的平均PSNR/SSIM优于大多数主流方法, 并在视觉效果方面展现出更高的质量和更好的细节保留能力.     

改进YOLOv7的视频监控小目标检测

小目标检测作为目标检测中一项极具挑战性的项目, 广泛分布于日常生活中, 在视频监控场景中, 距离摄像头约20 m远处的行人人脸就可以被认为是小目标. 由于人脸可能相互遮挡并容易受到噪声和天气光照条件的影响, 现有的目标检测模型在这类小目标上的性能劣于中大型目标. 针对此类问题, 本文提出了改进后的YOLOv7模型, 添加了高分辨率检测头, 并基于GhostNetV2对骨干网络进行了改造; 同时基于BiFPN和SA注意力模块替换PANet结构, 增强多尺度特征融合能力; 结合Wasserstein距离改进了原来的CIoU损失函数, 降低了小目标对锚框位置偏移的敏感性. 本文在公开数据集VisDrone2019以及自制的视频监控数据集上进行了对比实验. 实验表明, 本文提出的改进方法mAP指标在VisDrone2019数据集上提高到了50.1%, 在自制视频监控数据集上高于现有方法1.6个百分点, 有效提高了小目标检测的能力, 并在GTX1080Ti上达到了较好的实时性.     

基于车载环境的交通目标跟踪

针对车载环境下小目标难以识别和相机动态移动造成的目标跟踪精度下降问题, 提出一种基于改进YOLOv5与ByteTrack的交通目标跟踪方法. 首先, 引入Transformer与加权特征金字塔(BiFPN)结构的思想重构YOLOv5检测网络, 有效捕获了特征的全局依赖关系, 缓解了深层卷积小目标信息丢失问题, 改善了车载环境下的目标检测性能. 此后, 以ByteTrack为基础提出了添加相机移动补偿的CMC-ByteTrack跟踪方法, 更精准地描述了视频前后帧的数据关联关系, 提高了相机大幅位移时的跟踪精度. 实验结果表明, 改进YOLOv5的平均检测精度(mAP)达到了82.2%, 相比原算法提高了3.9%, 与CMC-ByteTrack结合后的跟踪准确性(MOTA)相比改进前的跟踪方法提高了2.8%.     

跨模态融合的双注意力脑肿瘤分割算法

针对脑肿瘤多模态信息融合不充分以及肿瘤区域细节信息丢失等问题,提出了一种跨模态融合的双注意力脑肿瘤图像分割网络(CFDA-Net).在编码器-解码器的基础结构上,首先在编码器分支采用密集块与大内核注意力并行的新卷积块,可以使全局和局部信息有效融合且可以防止反向传播时梯度消失的问题;其次在编码器的第2、3和4层的左侧加入多模态深度融合模块,有效地利用不同模态间的互补信息;然后在解码器分支使用Shuffle Attention注意力将特征图分组处理后再聚合,其中分组的子特征一分为二地获取空间与通道的重要注意特征.最后使用二进制交叉熵(binary cross entropy, BCE)、Dice Loss与L2 Loss组成新的混合损失函数,缓解了脑肿瘤数据的类别不平衡问题,进一步提升分割性能.在BraTS2019脑肿瘤数据集上的实验结果表明,该模型在整体肿瘤区域、肿瘤核心区域和肿瘤增强区域的平均Dice系数值分别为0.887、0.892和0.815.与其他先进的分割方法 ADHDC-Net、SDS-MSA-Net等相比,该模型在肿瘤核心区域和增强区域具有更好的分割效果.     

Next-Generation Grade and Survival Expression Biomarkers of Human Gliomas Based on Algorithmically Reconstructed Molecular Pathways

In gliomas, expression of certain marker genes is strongly associated with survival and tumor type and often exceeds histological assessments. Using a human interactome model, we algorithmically reconstructed 7494 new-type molecular pathways that are centered each on an individual protein. Each single-gene expression and gene-centric pathway activation was tested as a survival and tumor grade biomarker in gliomas and their diagnostic subgroups (IDH mutant or wild type, IDH mutant with 1p/19q co-deletion, MGMT promoter methylated or unmethylated), including the three major molecular subtypes of glioblastoma (proneural, mesenchymal, classical). We used three datasets from The Cancer Genome Atlas and the Chinese Glioma Genome Atlas, which in total include 527 glioblastoma and 1097 low grade glioma profiles. We identified 2724 such gene and 2418 pathway survival biomarkers out of total 17,717 genes and 7494 pathways analyzed. We then assessed tumor grade and molecular subtype biomarkers and with the threshold of AUC > 0.7 identified 1322/982 gene biomarkers and 472/537 pathway biomarkers. This suggests roughly two times greater efficacy of the reconstructed pathway approach compared to gene biomarkers. Thus, we conclude that activation levels of algorithmically reconstructed gene-centric pathways are a potent class of new-generation diagnostic and prognostic biomarkers for gliomas.     

基于通道注意力机制和多尺度卷积的行人重识别

针对真实环境中由于复杂背景和物体遮挡、角度变换、行人姿态变化带来的行人重识别(person re-identification,person re-ID) 问题,设计了基于通道注意力(efficient channel attention,ECA) 机制和多尺度卷积(poly-scale convolution,PSConv) 的行人重识别模型。首先利用残差网络提取全局特征,在网络末端加入基于ECA机制及PSConv的特征融合模块,将全局特征和该模块提取的全局特征进行融合,之后将新的全局特征进行分割得到局部特征,最后将新的全局特征和分割得到的局部特征融合得到最终特征,并计算损失函数。模型在Market1501和DukeMTMC-reID 数据集上进行实验验证。在Market1501数据集中,Rank-1和平均精度均值分别达到94.3%和85.2%,在DukeMTMC-reID数据集中,上述两参数分别达到86.3%和75.4%。实验结果可知,该模型可应对实际环境中的复杂情况,增强行人特征的辨别力,有效提高行人重识别的准确率和精度。