Inferring Relevant Cell Types for Complex Traits by Using Single-Cell Gene Expression.

Area of Science:

• Genetics
• Neuroscience
• Computational Biology

Background:

• Previous methods for identifying trait-relevant cell types using genome-wide association study (GWAS) signals are limited by cell resolution.
• Characterizing novel cell populations using single-cell RNA sequencing (RNA-seq) has advanced, but linking these to specific phenotypes remains challenging.

Purpose of the Study:

• To develop a computational model, RolyPoly, that integrates GWAS summary statistics and gene expression data to prioritize trait-relevant cell types and genes.
• To enhance the cell-type resolution in genetic studies of complex traits.

Main Methods:

• RolyPoly is a regression-based polygenic model utilizing GWAS summary statistics and gene expression data (bulk or single-cell RNA-seq).
• Model accuracy was assessed via simulation and validation against known tissue-trait associations.
• Gene prioritization was achieved by computing a trait-relevance score for each gene based on cell-type-specific expression.

Main Results:

• RolyPoly identified significant associations between microglia and late-onset Alzheimer disease.
• Schizophrenia associations were found with oligodendrocytes and replicating fetal cortical cells.
• Genes highly ranked by RolyPoly scores were enriched among differentially expressed genes in the Alzheimer disease prefrontal cortex.

Conclusions:

• RolyPoly provides a powerful framework for dissecting the genetic architecture of complex traits by linking common variants to specific cell types.
• The model improves cell-type resolution in genetic analyses, enabling discovery of novel trait-cell type associations.
• RolyPoly facilitates the identification of key genes within relevant cell types contributing to disease pathogenesis.