Cell clusters are filtered on the basis of size, their corresponding expression programs are subsequently filtered on the basis of number of genes with sufficiently high fold-change values and sufficiently small p-values, and lastly filtered on the basis of sufficiently low jaccard similarity between any two pairs of cell clusters, whereby the cell cluster with a larger number of significant genes is kept.
programs(
m,
groups = NULL,
nsig1 = 50,
nsig2 = 10,
jaccard = 0.7,
p = 0.01,
lfc = log2(2),
pmethod = "BH"
)expression matrix of genes by cells. not row-centered
cell clusters provided and will not be computed from data.
minimum number of genes with p-value 'p' to keep a cluster. Default: 50
number of genes with p-value 'p'/10. Default: 10
allowed jaccard similarity between clusters of cells. Default: 0.7
DEA adjusted p value for genes. Default: 0.01
DEA log2-FC value for genes. Default: log2(2)
adjust method. Default: 'BH'
The cell clusters, their gene expression profiles (all genes log2-foldchanges), their top DE genes (n = nsig1) and their p-values are all returned.