Refine clusters — refine_cl • scrattch.hicat

Refine clusters

refine_cl(
  cl,
  co.ratio = NULL,
  cl.mat = NULL,
  confusion.th = 0.6,
  min.cells = 4,
  niter = 50,
  tol.th = 0.02,
  verbose = 0
)

Arguments

cl: Cluster membership cluster
co.ratio: cell-cell co-clustering matrix.
cl.mat: cell-cluster matrix collected from bootstrapping iterations. Either co.ratio or cl.mat should not be NULL.
confusion.th: Clusters with average confusion score greater than this threshold will be removed. Cells in this cluster will be re-distributed to other most likely clusters.
min.cells: Clusters with fewer than this many cells will be removed. Cells in this cluster will be re-distributed to other most likely clusters.
niter: maxmimal mumber of refinement iterations.
tol.th: If improvement is smaller than this threshold, terminate refinement step.
verbose: If true, print out step-by-step improvement.

Zizhen Yao