Integrative analysis of single-cell gene expression: A comprehensive database approach

The exponential growth of single-cell datasets provides unprecedented opportunities to advance our understanding of complex biological systems. However, effectively locating and integrating related studies for meaningful insights remains challenging. Traditional databases primarily index basic metadata, which necessitates time-consuming downloading and re-filtering based on gene expression and cell type or tissue composition, followed by computationally intensive aggregation. This process often results in excessively large datasets that are difficult to analyze effectively, further complicated by batch effects. To address these issues, we have developed a computational approach to efficiently extract and index both expression data and annotations. Our comprehensive database incorporates detailed author annotations and gene expression profiles, enabling refined searches and integrated analyses to uncover common biological patterns while accounting for the repeatability of patterns across multiple studies and mitigating batch effects. This approach significantly reduces computational demands and enhances the accessibility and utility of single-cell transcriptomics data for the broader research community. In the first version, we release a human database comprising 244 datasets from 236 cell types, 35 tissues, and 31 conditions.