Cell Cycle Scoring

Purpose

Cell cycle scoring estimates whether each cell is likely in S phase, G2M phase, or neither.

This is useful because cell cycle can be a strong source of variation in scRNA-seq data. Sometimes it is a biological signal; sometimes it is a confounder.

Typical position:

NormalizeData() -> FindVariableFeatures() -> CellCycleScoring() -> ScaleData()

Gene Sets

Create S phase and G2M phase gene vectors:

s_genes <- Seurat::cc.genes$s.genes
g2m_genes <- Seurat::cc.genes$g2m.genes

If gene-name casing does not match the object, use CaseMatch():

s_genes <- Seurat::CaseMatch(
  search = s_genes,
  match = rownames(seu)
)

g2m_genes <- Seurat::CaseMatch(
  search = g2m_genes,
  match = rownames(seu)
)

Score Cell Cycle

Run cell cycle scoring:

seu <- Seurat::CellCycleScoring(
  object = seu,             # Seurat object after normalization
  s.features = s_genes,     # S phase genes
  g2m.features = g2m_genes, # G2M phase genes
  set.ident = FALSE         # keep current identities unchanged
)

This adds metadata columns:

• S.Score
• G2M.Score
• Phase

Check Scores

Inspect metadata:

metadata <- seu[[]]

head(metadata[, c("S.Score", "G2M.Score", "Phase")])
table(seu$Phase)

Visualize score distributions:

Seurat::VlnPlot(
  object = seu,
  features = c("S.Score", "G2M.Score"),
  group.by = "Phase",
  pt.size = 0.1,
  ncol = 2
)

If clustering or embeddings already exist, check whether cell cycle explains structure:

Seurat::DimPlot(
  object = seu,
  group.by = "Phase"
)

Regress Or Not

Cell cycle scores can be regressed during scaling:

seu <- Seurat::ScaleData(
  object = seu,
  vars.to.regress = c("S.Score", "G2M.Score")
)

But do not regress cell cycle blindly.

Consider regression when:

• clusters are mainly separated by cell cycle phase
• cell cycle is not part of the biological question
• proliferating cells obscure the main cell-type structure

Avoid or be cautious when:

• studying proliferation
• studying tumor, development, regeneration, stem cells, or immune activation
• cell cycle state is itself biologically meaningful

Note

Cell cycle scoring is a diagnostic step. The important decision is not only how to calculate S.Score and G2M.Score, but whether those scores should be treated as unwanted variation.