Performs iterative reclustering on each major cluster found by FindClusters in a Seurat or SCE object.
It refines the clusters using the FindSubCluster function for better resolution and fine-tuned annotation.
The new clustering results are stored in a metadata column called annotation_recluster.
Suitable for improving cluster precision and granularity after initial clustering.
Usage
DO.FullRecluster(
sce_object,
over_clustering = "seurat_clusters",
res = 0.5,
algorithm = 4,
graph.name = "RNA_snn"
)Arguments
- sce_object
The seurat or SCE object
- over_clustering
Column in metadata in object with clustering assignments for cells, default seurat_clusters
- res
Resolution for the new clusters, default 0.5
- algorithm
Set one of the available algorithms found in FindSubCLuster function, default = 4: leiden
- graph.name
A builded neirest neighbor graph
Examples
sce_data <- readRDS(system.file("extdata", "sce_data.rds", package = "DOtools"))
sce_data <- DO.FullRecluster(
sce_object = sce_data
)
#> Computing nearest neighbor graph
#> Computing SNN
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> 2 singletons identified. 3 final clusters.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> 1 singletons identified. 3 final clusters.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> 1 singletons identified. 2 final clusters.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> 1 singletons identified. 3 final clusters.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#> Warning: `random.seed` must be greater than 0 for leiden clustering, resetting `random.seed` to 1.
#>