scrna-seurat-general

name	scrna-seurat-general
description	Baseline Seurat v5 scRNA-seq pipeline — raw 10x matrices through QC, clustering, marker-based cell-type annotation, and per-cell-type differential expression. Reach for this to preprocess 10x gene-expression data, cluster and annotate cells, or run a two-condition DEG comparison from the lab's default settings. For batch integration across many samples use scrna-scvi-integration; for label transfer from a reference use scrna-label-transfer instead.
license	lab-internal
requirements	["r-4.3","seurat-5","bioc-3.18"]
metadata	{"display-name":"scRNA-seq General Workflow (Seurat)","source_workflow":"scRNAseq_general_workflow"}

scRNA-seq General Workflow (Seurat) — QC → cluster → annotate → DEG

The lab's baseline Seurat v5 pipeline for two-condition single-cell RNA-seq. It reads raw 10x matrices, applies dataset-specific QC thresholds, clusters, annotates cell types from markers, and runs cell-type-specific differential expression. Input is one or more 10x raw_feature_bc_matrix directories; output is an annotated Seurat object plus per-cell-type DEG tables. Use it as the default starting point when you have standard 10x gene-expression matrices and want the lab's committed QC/clustering/ annotation conventions rather than a bespoke pipeline.

When to use it

Use this when you have standard 10x gene-expression matrices and want a general-purpose preprocessing-through-DEG path with the lab's default thresholds. For integrating many batches into one corrected latent space use scrna-scvi-integration; for transferring annotations from an existing reference onto a query use scrna-label-transfer; for CNV inference use scrna-infercnv. This skill assumes a small number of samples compared directly, not a large atlas.

Inputs

• Object / format: one or more 10x raw_feature_bc_matrix/ directories (barcodes, features, matrix). Raw integer counts — not filtered, not normalized.
• Required upstream: none; this is the entry point. _common/functions.R must be sourceable for shared helpers.
• Example data: scRNAseq_general_workflow/data/ctrl_raw_feature_bc_matrix/ and .../stim_raw_feature_bc_matrix/ (committed control + stimulation matrices).

How to run

suppressPackageStartupMessages({
  library(Seurat)      # v5
  library(purrr)
})

# 1. Load raw 10x matrices and merge (counts stay raw)
data_dirs <- c("data/ctrl_raw_feature_bc_matrix", "data/stim_raw_feature_bc_matrix")
data_list <- lapply(data_dirs, function(d) {
  tmp <- Read10X(d, unique.features = TRUE, strip.suffix = FALSE)
  CreateSeuratObject(tmp, assay = "RNA", min.cells = 3, min.features = 200,
                     project = basename(d))
})
combined <- purrr::reduce(data_list, function(x, y) merge(x, y, merge.data = TRUE))
combined <- PercentageFeatureSet(combined, "^MT-", col.name = "percent.mito")
combined <- PercentageFeatureSet(combined, "^RP[SL]", col.name = "percent.ribo")

# 2. QC filter — thresholds are dataset-specific; inspect before committing them
combined_qc <- subset(
  combined,
  subset = percent.mito < 5 & percent.ribo < 50 & nFeature_RNA > 200
)

# 3. Normalize → HVG → scale → PCA → cluster → UMAP
combined_qc <- NormalizeData(combined_qc)
combined_qc <- FindVariableFeatures(combined_qc, nfeatures = 2000)
combined_qc <- ScaleData(combined_qc)
combined_qc <- RunPCA(combined_qc, npcs = 30)
combined_qc <- FindNeighbors(combined_qc, dims = 1:30)
combined_qc <- FindClusters(combined_qc, resolution = 0.5)
combined_qc <- RunUMAP(combined_qc, dims = 1:30)

# 4. Markers + manual annotation (annotation is a human decision from these markers)
combined_qc <- JoinLayers(combined_qc)   # Seurat v5: needed before marker/DE calls
markers <- FindAllMarkers(combined_qc, only.pos = TRUE, min.pct = 0.25)
# ... assign combined_qc$cell_type from markers ...

# 5. Per-cell-type DEG between conditions
# ... see 3_deg.rmd ...

Output

Key / object	What it holds
combined_qc$seurat_clusters	cluster id per cell
combined_qc$cell_type	assigned cell-type label per cell
markers	per-cluster positive markers
<deg output files>	TODO — per-cell-type DEG tables (name them after a run)

Running on OSC

module load R/4.3.0
# sbatch template — TODO

Gotchas

Gotcha	What happens / fix
filled after first real run

Troubleshooting

Symptom	Fix
filled after first real run

---

Next: annotate downstream with scrna-module (module enrichment) or hand the DEG tables to enrichment-pathway.