AssignStableCluster: Assign stable cluster

In crazyhottommy/scclusteval: Evaluate the single cell clustering

Description

Assign stable cluster

Usage

AssignStableCluster(
  idents1,
  idents2,
  method = "jaccard_median",
  jaccard_cutoff = 0.6,
  percent_cutoff = 0.6
)

Arguments

idents1	A list of cluster identity copied from the orginal data sets. idents1 is a list of the cluster identity from the subsampled data sets before reclustering.
idents2	A list of cluster identity from the subsampled data sets. idents2 is a list of the cluster identity from the subsampled data sets after reclustering. The order of identities in idents1 and idents2 should correspond to each other.
method	what way to summarize the jaccard index across all simulations. to determine a cluster is stable or not. options are "Jaccard_mean", "Jaccard_median" and "Jaccard_percent"
jaccard_cutoff	Cutoff of the jaccard index to determin a cluster is stable or not. it is the mean or median cutoff when the method is "jaccard_mean" or "jaccard_median" and it is the cutoff for every subsampling when the method is "jaccard_percent"
percent_cutoff	The percentage of jaccard index greater than jaccard_cutoff. Used when method is "jaccard_percent". specify 0.6 when you mean 60%.

Value

A list containing the raw data for jaccard index for all simulations, TRUE or FALSE of stable cluster for each cluster and a number of stable clusters. A cluster is deemed as stable if the median (or mean) jaccard index is > cutoff. in addtion, a stable_index is calculated, which is the pecentage of jaccard index > cutoff for all the subsampling. e.g. for 100 times subsampling, 0.8 means 80% of the time, the jaccard index is > cutoff. Sometimes, we see bimodal distrbution of the 100 jaccard index, the percentage is a better measurement than the mean or median of the 100 jaccard index.

Examples

data(idents)

AssignStableCluster(idents, idents)

crazyhottommy/scclusteval documentation built on Aug. 5, 2021, 3:20 p.m.