NeuroEmbed
收藏资源简介:
NeuroEmbed是一个用于神经退行性疾病研究的语义精准嵌入空间工程方法。该方法包括从公共数据库中提取数据集、使用生物医学本体和嵌入空间聚类进行元数据标准化和增强、基于随机组合的标准化元数据维度生成问答数据集、以及针对特定领域的嵌入器进行微调等四个阶段。通过NeuroEmbed方法,我们对2,801个数据集和150,924个样本进行了语义索引,并将GEO数据库中超过1,700个异构组织标签标准化为326个与本体一致的概念,同时通过引入新的本体一致术语丰富了注释。该数据集的创建旨在解决神经退行性疾病研究中元数据标准化和语义检索的问题,为自动化生物信息学流程构建提供支持。
NeuroEmbed is a semantically precise embedding space engineering approach for neurodegenerative disease research. It comprises four stages: extracting datasets from public databases, standardizing and enriching metadata using biomedical ontologies and embedding space clustering, generating question-answering datasets based on randomly combined standardized metadata dimensions, and fine-tuning domain-specific embedders. Using the NeuroEmbed approach, we performed semantic indexing on 2,801 datasets and 150,924 samples, standardized over 1,700 heterogeneous tissue labels from the GEO database into 326 ontology-consistent concepts, and enriched annotations by introducing novel ontology-consistent terminology. This dataset was developed to address the challenges of metadata standardization and semantic retrieval in neurodegenerative disease research, providing support for the construction of automated bioinformatics workflows.
NeuroEmbed数据集概述
基本信息
- 数据集名称:NeuroEmbed
- 托管平台:GitHub
- 托管地址:https://github.com/JoseAdrian3/NeuroEmbed
数据集描述
(注:根据提供的README内容,该数据集未包含具体描述信息)




