Mouse Brain snRNA-seq Analysis

A Seurat v5 workflow for mouse-brain single-nucleus RNA sequencing. The project includes per-sample quality control and annotation, two-sample CCA integration, cell-type-specific differential expression, GO enrichment, genes-of-interest analyses, and a transcriptional assessment of blood-brain barrier (BBB)-related programs.

Important limitation: the project contains one 10x sample per condition (n = 1). All between-sample results are descriptive and hypothesis-generating. They do not support group-level statistical inference or establish treatment safety.

1. Study design

Sample	Biological condition	Comparison role
Sample1	Untreated AD model mouse	Reference
Sample2	AD model mouse after repeated ALNP treatment	Treated sample

The primary comparison is:

Sample2 (AD + ALNP) - Sample1 (untreated AD)

2. Workflow

10x filtered_feature_bc_matrix.h5
        |
        +-- Per-sample analysis (Sample1 / Sample2)
        |   QC -> per-library scDblFinder -> LogNormalize -> HVG -> ScaleData -> PCA
        |   -> clustering -> t-SNE -> markers -> cell-type/region annotation
        |
        +-- Two-sample integration (Integration)
            merge -> Seurat v5 CCAIntegration -> joint clustering/t-SNE
            -> cell-type DEG -> GO -> microglia -> genes of interest
            -> expression/QC export -> BBB transcriptional assessment

In the main integration workflow, scDblFinder is run independently on each physical 10x library before merging. Using seed 20260728 and an expected 10x multiplet rate of approximately 1% per 1,000 recovered nuclei, 287/4,309 nuclei in Sample1 and 283/4,305 nuclei in Sample2 were flagged and excluded; 4,022 predicted singlets from each library entered the Seurat integration workflow.

3. Repository layout

snRNAseq_mouseBrain/
├── README.md
├── .gitignore
├── Sample1/
│   └── snRNAseq_mouseBrain_sample1_tsne.R
├── Sample2/
│   └── snRNAseq_mouseBrain_sample 2_tsne.R

└── Integration/
    ├── snRNAseq_mouseBrain_integration.R
    ├── scDblFinder_doublet_detection.R
    ├── snRNAseq_mouseBrain_integration_downstream.R
    ├── expression_qc_dotplot.R
    ├── gene_of_interest_per_sample.R
    ├── gene_of_interest_per_celltype.R
    └── snRNAseq_mouseBrain_BBB_integrity.R

Raw inputs and all generated analysis outputs are local artifacts and are excluded from Git. The repository contains source code, repository configuration, and this README only.

4. Main scripts

Script	Purpose	Main input	Main output
Sample1/snRNAseq_mouseBrain_sample1_tsne.R	Sample1 QC, clustering, annotation, and expression export	Sample1 10x H5	Sample1 result directories
Sample2/snRNAseq_mouseBrain_sample 2_tsne.R	Equivalent Sample2 workflow	Sample2 10x H5	Sample2 result directories
Integration/snRNAseq_mouseBrain_integration.R	CCA integration and principal analyses from the two H5 files	Two 10x H5 files	Integrated objects, t-SNE, DEG, GO, and microglia results
Integration/scDblFinder_doublet_detection.R	Per-library doublet calling used by the main integration script	Individual 10x count matrices	Singlet barcodes, doublet calls, and per-library counts
Integration/snRNAseq_mouseBrain_integration_downstream.R	Resume downstream analyses from an integrated object	mousebrain_integrated.rds	DEG, GO, microglia, and composition
Integration/expression_qc_dotplot.R	Expression matrices, QC summaries, and dot plots	Final integrated object	CSV/XLSX tables and figures
Integration/gene_of_interest_per_sample.R	Sample-level genes-of-interest comparison	Final integrated object	Wilcoxon results and SVG figures
Integration/gene_of_interest_per_celltype.R	Cell-type-specific genes-of-interest comparison	Final integrated object	Cell-type statistics and SVG figures
Integration/snRNAseq_mouseBrain_BBB_integrity.R	BBB-related composition, module, and gene assessment	Final integrated object	Integration/BBB/
Integration/BBB/export_genes_logexpr_per_cell.R	Export per-cell source data for BBB violin plots	Final integrated object	Metadata-bearing CSV

5. Software environment

The project has been run with:

• R 4.3.3
• Seurat 5.2.1
• scDblFinder 1.16.0

Core R packages:

Seurat
dplyr
tidyr
tibble
ggplot2
patchwork
svglite
Matrix
openxlsx
scDblFinder
SingleCellExperiment
scater
scuttle
BiocSingular

GO enrichment additionally requires:

clusterProfiler
org.Mm.eg.db

The repository does not currently contain an renv.lock file, so the complete package environment is not locked.