scMRdb links cell-specific gene activity to evidence on complex traits

scMRdb combines 5206 genome-wide association study datasets with 585 single-cell gene-regulation datasets to investigate potential causal links between gene activity and human traits and diseases. Described in a 2026 paper, the database uses a uniform analysis pipeline to sort results into six confidence tiers. The association studies cover European and East Asian ancestry; the single-cell datasets span 11 tissues and groups of cells studied together, including 110 cell types.
Genetic analyses that measure gene activity across whole tissues can miss effects confined to particular cell types. Averaging activity across different cell populations can dilute or obscure those signals. Evidence from single-cell studies has also been scattered across studies using inconsistent methods, preventing standardized comparisons between cell types.
scMRdb applies the same analytical process to these datasets. Its gene-regulation data concern expression quantitative trait loci, or eQTLs: genetic variants associated with differences in gene activity. Single-cell datasets let researchers examine those associations within particular cell types. The genome-wide association studies provide summary statistics, the numerical results describing associations between genetic variation and human traits and diseases.
The database uses Mendelian randomization, a method that uses genetic variants to investigate whether differences in gene activity might contribute to disease. The variants serve as tools for assessing that relationship while reducing the influence of confounding, where other factors distort an apparent connection. The method also aims to reduce reverse causation: disease changing gene activity rather than gene activity contributing to disease. The causal links inferred through this approach remain provisional.
Every gene-trait pair undergoes Mendelian randomization followed by Steiger directionality testing, which assesses the proposed direction of the relationship. Significant associations then undergo colocalization analysis. This checks whether the genetic signals for gene activity and the trait point to a shared underlying variant. Results are grouped into six confidence tiers, with information about drug targets included alongside them.
The database is available at https://ngdc.cncb.ac.cn/scmrdb.
The consistent workflow allows researchers to compare evidence across cell types and identify candidate causal genes whose signals may be obscured in whole-tissue analyses. The authors expect this evidence to help researchers form hypotheses about how complex diseases arise and prioritize potential therapeutic targets.
Written by the Genomes desk from the primary source cited and linked above and checked against it. Research use only; not medical advice. Corrections: [email protected].