Blood samples have traditionally been used as the main source of DNA for genetic analysis. How-ever, this source can be difficult in terms of collection, transportation, and long-term storage. In this study, we invest...Blood samples have traditionally been used as the main source of DNA for genetic analysis. How-ever, this source can be difficult in terms of collection, transportation, and long-term storage. In this study, we investigated whether human nail clippings could be used as a source of DNA for SNP genotyping, null-allele detection, and whole-genome amplification. From extracted nail DNA, we achieved amplicons up to a length of ~400 bp and >96% concordance for SNP genotyping and 100% concordance for null-allele detection compared to DNA derived from matched blood sam-ples. For whole-genome amplification, OmniPlex performed better than Multiple Displacement Amplification with a success rate of 89.3% and 76.8% for SNP genotyping and null-allele detection, respectively. Concordance was ~98% for both methods. When combined with OmniPlex whole-genome amplification, human nail clippings could potentially be used as an alternative to whole blood as a less invasive and more convenient source of DNA for genotyping studies.展开更多
图像描述生成算法是计算机视觉中的关键环节,旨在从给定的输入图像中预测相关文本信息,以实现对图像内容的准确理解与表达。提出一种借鉴平均教师算法的模型,并采用独特的双分支网络架构。为提升模型准确性与稳定性,在每个分支中引入位...图像描述生成算法是计算机视觉中的关键环节,旨在从给定的输入图像中预测相关文本信息,以实现对图像内容的准确理解与表达。提出一种借鉴平均教师算法的模型,并采用独特的双分支网络架构。为提升模型准确性与稳定性,在每个分支中引入位置前馈块。在图像特征提取方面,运用对比语言图像预训练(CLIP)方法,以获取图像的多层次特征,从而更好地捕捉图像的语义信息。在描述生成阶段,通过映射网络将图像特征转化为文本信息,进而利用GPT-2技术来提升预测的准确度与语义的连贯性。为验证模型性能,在Microsoft common objects in context(MSCOCO)和Flickr30k等图像描述数据集上进行充分的训练与测试。测试结果显示所提模型在两个数据集上均表现出色,证实其在图像描述生成任务中的高效性与实用性。研究为图像描述生成领域提供了新的思路与方法,具有深远的理论与实践意义。展开更多
文摘Blood samples have traditionally been used as the main source of DNA for genetic analysis. How-ever, this source can be difficult in terms of collection, transportation, and long-term storage. In this study, we investigated whether human nail clippings could be used as a source of DNA for SNP genotyping, null-allele detection, and whole-genome amplification. From extracted nail DNA, we achieved amplicons up to a length of ~400 bp and >96% concordance for SNP genotyping and 100% concordance for null-allele detection compared to DNA derived from matched blood sam-ples. For whole-genome amplification, OmniPlex performed better than Multiple Displacement Amplification with a success rate of 89.3% and 76.8% for SNP genotyping and null-allele detection, respectively. Concordance was ~98% for both methods. When combined with OmniPlex whole-genome amplification, human nail clippings could potentially be used as an alternative to whole blood as a less invasive and more convenient source of DNA for genotyping studies.
文摘图像描述生成算法是计算机视觉中的关键环节,旨在从给定的输入图像中预测相关文本信息,以实现对图像内容的准确理解与表达。提出一种借鉴平均教师算法的模型,并采用独特的双分支网络架构。为提升模型准确性与稳定性,在每个分支中引入位置前馈块。在图像特征提取方面,运用对比语言图像预训练(CLIP)方法,以获取图像的多层次特征,从而更好地捕捉图像的语义信息。在描述生成阶段,通过映射网络将图像特征转化为文本信息,进而利用GPT-2技术来提升预测的准确度与语义的连贯性。为验证模型性能,在Microsoft common objects in context(MSCOCO)和Flickr30k等图像描述数据集上进行充分的训练与测试。测试结果显示所提模型在两个数据集上均表现出色,证实其在图像描述生成任务中的高效性与实用性。研究为图像描述生成领域提供了新的思路与方法,具有深远的理论与实践意义。
