getApproxComMembership: Get clusters by community detection on approximate nearest...
In JEFworks/MUDAN: Multi-sample Unified Discriminant ANalysis

Description Usage Arguments Value Examples

Group cells into clusters based on graph-based community detection on approximate nearest neighbors for random subset of cells For when getComMembership takes too long due to there being too many cells

1
2
3

getApproxComMembership(mat, k, nsubsample = ncol(mat) * 0.5,
  method = igraph::cluster_walktrap, seed = 0, vote = FALSE,
  verbose = TRUE)

`mat`	Matrix of cells as columns. Features as rows (such as PCs).
`k`	K-nearest neighbor parameter.
`nsubsample`	Number of cells in subset (default: ncol(mat)*0.5)
`method`	Community detection method from igraph. (default: igraph::cluster_walktrap)
`seed`	Random seed for reproducibility
`vote`	Use neighbor voting system to annotate rest of cells not in subset. If false, will use machine-learning model. (default: FALSE)
`verbose`	Verbosity (default: TRUE)

Vector of community annotations

## Not run: 
data(pbmcA)
cd <- pbmcA
mat <- cleanCounts(cd)
mat <- normalizeVariance(mat)
pcs <- getPcs(mat)
com <- getApproxComMembership(pcs, k=30, getApproxComMembership=1000)

## End(Not run)