Alignment of single-cell RNA-seq samples without overcorrection using kernel density matching.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) offers cell type-specific gene expression insights.
• Batch effects in scRNA-seq data hinder integration and cross-experiment comparisons.
• Existing alignment methods may overcorrect or fail with partially overlapping cell types.

Purpose of the Study:

• To introduce Dmatch, a novel method for aligning multiple scRNA-seq experiments.
• To enable robust integration of scRNA-seq datasets despite batch variations and cell type heterogeneity.
• To enhance downstream biological analyses, including differential gene expression and eQTL mapping.

Main Methods:

• Leverages an external human primary cell expression atlas.
• Employs kernel density matching for data alignment.
• Tested via simulations and application to clinical autoimmune disease patient samples.

Main Results:

• Dmatch effectively reduces sample-specific clustering and avoids overcorrection compared to other methods.
• Enabled cell type-specific differential gene expression analysis across biopsy sites and disease conditions.
• Identified shared pro-inflammatory monocytes in Rheumatoid Arthritis patients and increased eQTL mapping from scRNA-seq data.

Conclusions:

• Dmatch is a fast, scalable, and effective tool for aligning scRNA-seq data.
• Improves the utility of scRNA-seq for biological and medical applications.
• Facilitates deeper insights from integrated scRNA-seq datasets, including clinical applications.