Immune Annotation Branch

Immune-focused marker review on top of the general scRNA-seq workflow

What it does

This branch focuses on manual immune-cell annotation for a preprocessed scRNA-seq Seurat object. It reuses the general workflow structure but substitutes immune and blood-cell marker panels for the annotation step.

When to use it

Use this workflow when preprocessing and clustering are already complete and the main need is immune-specific cell type annotation. It is intended as a branch that replaces the generic annotation notebook for immune datasets rather than a standalone end-to-end pipeline.

Prerequisites

Steps

Load the processed Seurat object

The branch assumes a previously clustered Seurat object and jumps directly into marker review.

combined <- qs::qread(paste0("../scRNAseq_preprocess_main/combined.qsave"))
DefaultAssay(combined) <- "RNA"
Idents(combined) <- combined$seurat_clusters

Start from the cluster UMAP and normalize the RNA assay for marker review

The notebook first plots cluster identities on the UMAP, then normalizes the RNA assay specifically for visualization. This keeps the branch aligned with the general workflow while swapping in immune-specific annotation logic.

DimPlot(combined, reduction = "umap", label = TRUE, pt.size = 0.4, cols = cell_type_color)
combined <- NormalizeData(combined, verbose = FALSE)

Review immune and blood-cell marker panels

The notebook uses FeaturePlot() to inspect canonical markers for monocytes, dendritic cells, B cells, T-cell subsets, NK cells, megakaryocytes, and erythrocytes.

FeaturePlot(combined, features = c("CD14", "LYZ"), order = TRUE, min.cutoff = "q10", label = TRUE)
FeaturePlot(combined, features = c("FCGR3A", "MS4A7"), order = TRUE, min.cutoff = "q10", label = TRUE)
FeaturePlot(combined, features = c("GNLY", "NKG7", "PPBP", "HBB", "HBA2"), ncol = 2, label = TRUE)

The source table in the notebook makes the intended mapping explicit, covering CD14+ monocytes, FCGR3A+ monocytes, conventional and plasmacytoid dendritic cells, B cells, CD4/CD8 T cells, NK cells, megakaryocytes, and erythrocytes.

Recode cluster IDs into immune cell labels

After reviewing marker patterns, the notebook manually remaps cluster identities into a cell_type column. The code is explicit about the cluster-to-label mapping rather than inferring it implicitly.

tmp_ident <- as.factor(combined$seurat_clusters)
tmp_levels <- levels(tmp_ident)
tmp_levels[c(1)] <- "CD14+ monocytes"
tmp_levels[c(7)] <- "FCGR3A+ monocytes"
tmp_levels[c(10, 11)] <- "Conventional dendritic cells"
levels(tmp_ident) <- tmp_levels
combined <- AddMetaData(combined, tmp_ident, col.name = "cell_type")

Check the relabeled UMAP, sample splits, and composition summaries

The notebook then replots the object grouped by cell_type, splits the UMAP by orig.ident, and uses dittoBarPlot() to summarize cell-type composition by sample in both percent and count space.

Idents(combined) <- combined$cell_type
DimPlot(combined, reduction = "umap", split.by = "orig.ident", label = TRUE, ncol = 2)
dittoBarPlot(combined, "orig.ident", group.by = "cell_type", scale = "percent")
dittoBarPlot(combined, "orig.ident", group.by = "cell_type", scale = "count")

Save the updated Seurat object

After reviewing marker expression, the notebook relabels clusters as immune cell types and writes the updated metadata back into the Seurat object.

Gotchas / notes

  • The README explicitly says this folder is a branch of the complete workflow and should be paired with the corresponding preprocessing notebooks from scRNAseq_general_workflow.
  • The source material is narrowly focused on annotation, so there are no committed preprocessing figures or separate local images here.
  • The marker lists are hard-coded in the notebook and should be checked against the species and experiment.

📄 View source on GitHub