Batch effects in single-cell RNA-sequencing data are corrected by matching mutual nearest neighbors

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Large-scale single-cell RNA sequencing (scRNA-seq) data present challenges due to batch effects from varying experimental conditions.
• Existing methods often fail by assuming identical cell population compositions across batches, limiting data integration.

Purpose of the Study:

• To introduce a novel batch correction strategy for scRNA-seq data.
• To overcome limitations of current methods by not requiring predefined or equal cell population compositions.

Main Methods:

• A strategy based on detecting mutual nearest neighbors (MNNs) in high-dimensional gene expression space.
• The MNN approach requires only a shared subset of cell populations between batches for effective correction.

Main Results:

• Demonstrated superior performance compared to existing methods using simulated and real scRNA-seq datasets.
• Successfully scaled the MNN batch-effect correction method to large numbers of cells from droplet-based scRNA-seq data.

Conclusions:

• The MNN-based strategy offers a robust and scalable solution for batch effect correction in scRNA-seq.
• This method enhances the integration and interpretation of diverse scRNA-seq datasets, even with varying cell compositions.