Integrating single-cell and single-nucleus datasets improves bulk RNA-seq deconvolution.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Bulk RNA sequencing (RNA-seq) deconvolution commonly relies on single-cell RNA sequencing (scRNA-seq) references.
• Single-nucleus RNA sequencing (snRNA-seq) detects cell types missed by scRNA-seq but captures nuclear, not cytoplasmic, transcripts, potentially impacting deconvolution accuracy.

Purpose of the Study:

• To benchmark integration strategies for using snRNA-seq data as references for bulk RNA-seq deconvolution.
• To evaluate methods for improving deconvolution accuracy when incorporating snRNA-seq data.

Main Methods:

• Benchmarking integration strategies across four tissues.
• Comparing principal component (PC)-based latent shifts, conditional and non-conditional scVI (single cell variational inference), and cross-modality differentially expressed gene (DEG) filtering.
• Assessing cell-fraction estimation robustness in real adipose bulk samples.

Main Results:

• All tested integration strategies improved deconvolution accuracy over raw snRNA-seq data.
• Pruning cross-modality DEGs yielded the most substantial improvements, often matching or surpassing scRNA-seq-only references.
• Conditional scVI demonstrated comparable performance and utility when matched cell types were unavailable.

Conclusions:

• Prioritize scRNA-seq references for bulk RNA-seq deconvolution when available.
• Append snRNA-seq data after removing cross-modality DEGs for enhanced accuracy.
• Conditional scVI serves as a practical alternative when DEG information is limited.