Integrated single-nucleus RNA-seq dataset of mouse brain from a multi-factorial benchmarking study

Single-nucleus RNA sequencing (snRNA-seq) is pivotal for dissecting cellular heterogeneity in the brain, yet the individual and interactive effects of technical variables on cell composition and gene expression remain poorly quantified. Here we present the first multi-factorial benchmarking study of snRNA-seq technical confounders in the adult mouse brain, by performing a systematic evaluation of tissue harvest methods (post-mortem vs. in vivo) and nuclear fixation (methanol-fixed vs. fresh), using the identical sequencing platforms (SeekOne Digital Droplet). We find that standard post-mortem harvest induces a rapid, selective loss of metabolically active glutamatergic neuron subpopulations, which obscures transcripts related to advanced cognitive functions while preserving overall cellular proportions. While methanol fixation minimally alters cell-type composition, it introduces cell-type-specific gene expression shifts that preferentially stabilize structural and core functional transcripts. We validated these technical artifacts to provide a systematic map of confounders, identifying glutamatergic neurons as uniquely vulnerable to workflow-induced bias. Our study offers a robust experimental and analytical framework to separate technical noise from true biological signal. These findings establish that precise reporting and control of workflow variables are prerequisites for building reproducible, integrative brain cell atlases and ensuring accurate biological discovery in single-cell transcriptomics.