Generate scores for cell types for each level

Generate scores for cell types for each level

Usage

hitype_assign(
  clusters,
  scores,
  gs = NULL,
  fallback = "Unknown",
  threshold = NULL,
  top = 10,
  mode = c("cluster", "cell")
)

Arguments

clusters

A named vector of original cluster assignments (names - cell names, values - cluster assignments)

scores

A list of matrices of cell type scores for each level

gs

The gene sets prepared by gs_prepare The cell_names is actually used. One could also pass gs$cell_names directly.

fallback

A fallback cell type if no cell type is assigned

threshold

Confidence threshold as top1/top2 score ratio. NULL (default) means no confidence filtering. A number marks the rank-1 cell type of a cluster as <UNKNOWN> when its score is less than threshold times the second-best score.

top

The number of top cell types to assign for each cluster in the result.

mode

"cluster" (default) aggregates scores per cluster then assigns the top cell type. "cell" assigns each cell individually then reports majority vote per cluster.

Value

A dataframe with columns: Level, Cluster, CellType, Score, and Margin. For each level and cluster, the top cell types are returned. You can use summary.hitype_result to print the combination of cell types for each cluster.