UMAP

Purpose

UMAP is a nonlinear dimensionality reduction method used to visualize cells in two dimensions.

In Seurat, UMAP is usually run with RunUMAP().

UMAP is a visualization method. It does not create clusters by itself. Cluster labels usually come from FindClusters().

Where It Fits

Typical position:

RunPCA() -> FindNeighbors() -> FindClusters() -> RunUMAP()

After Harmony integration:

RunPCA() -> RunHarmony() -> FindNeighbors() -> FindClusters() -> RunUMAP()

Run UMAP after choosing the reduction and dimensions used for downstream analysis.

Choose Reduction And Dims

Use the same reduction and dimensions used for graph construction.

Common choices:

Workflow	reduction	dims
Classic PCA	"pca"	selected PCs, often 1:30
SCT PCA	"pca"	selected PCs, often 1:40
CCA/SCT integration	"pca"	PCs computed from integrated assay
Harmony integration	"harmony"	selected Harmony dimensions

Run UMAP

For classic, SCT, or integrated assay PCA:

dims <- 1:30

seu <- Seurat::RunUMAP(
  object = seu,
  reduction = "pca",
  dims = dims,
  n.neighbors = 30,
  min.dist = 0.3,
  dim.embed = 2,
  seed.use = 42,
  verbose = TRUE
)

For Harmony:

dims <- 1:30

seu <- Seurat::RunUMAP(
  object = seu,
  reduction = "harmony",
  dims = dims,
  n.neighbors = 30,
  min.dist = 0.3,
  dim.embed = 2,
  seed.use = 42,
  verbose = TRUE
)

RunUMAP() stores the result as a reduction, usually named umap.

dim.embed = 2 sets the output embedding dimension. The default 2D UMAP layout is usually used for visualization.

Parameters

Common UMAP parameters:

Parameter	Meaning
n.neighbors	size of local neighborhood used by UMAP
min.dist	how tightly points can be packed in the embedding
dim.embed	number of output dimensions
seed.use	random seed for reproducible layout

Smaller min.dist can produce tighter clusters. Larger min.dist can produce a more spread-out layout.

Check UMAP

Check reductions:

Seurat::Reductions(seu)

Extract UMAP coordinates:

umap_embeddings <- Seurat::Embeddings(
  object = seu,
  reduction = "umap"
)

head(umap_embeddings)

Plot UMAP

Plot clusters on UMAP:

Seurat::DimPlot(
  object = seu,
  reduction = "umap",
  group.by = "seurat_clusters",
  label = TRUE, # show cluster labels
  repel = TRUE # avoid label overlap
)

Plot metadata on UMAP:

Seurat::DimPlot(
  object = seu,
  reduction = "umap",
  group.by = "sample"
)

Plot gene expression on UMAP:

Seurat::FeaturePlot(
  object = seu,
  features = "MS4A1",
  reduction = "umap"
)

Note

UMAP is useful for visualization, but distances between far-apart groups should not be overinterpreted.

Changing reduction, dims, n.neighbors, min.dist, or seed.use can change the UMAP layout.