gsnca_gsets: Run GSNCA over multiple gene sets

In hui-sheen/MetaGSCA: Meta-analysis of Gene Set differential Co-expression Analysis

Run GSNCA over multiple gene sets

Description

For each gene-set loop, genes are overlaid to expression matrix and too small gene sets are declined for GSNCA run.

Usage

gsnca_gsets(
  gsets,
  object,
  group,
  perm.list,
  cor.method = "pearson",
  max.skip = 50,
  min.sd = 0.001,
  minGsize = 3
)

Arguments

gsets	list for multiple gene sets.
object	gene expression matrix covering two groups. Row names are gene symbols.
group	original groupping of samples, vector of 1's and 2's.
perm.list	list of permutation specs. Each component gives permutated sample indices
cor.method	correlation method
max.skip	maximum number of repeated permutation/bootstrap times to avoid zero STD
min.sd	a valid data matrix per group must have at least this much per-feature STD
minGsize	considered gene set must have this minimum size after overlaying with gene expression matrix.

Details

Due to too small gene set size (with consideration of intersection with expression data), certain gene sets have NA as p and stat results.

Value

geneset-wise GSNCA results, each consisting of p-value and statistics out of GSNCA.

See Also

[gsnca_p()] for the GSNCA algorithm, which further calls on [gsnca_stat()] for coexpression distance statistics.

Examples

data(meta)
BRCA <- datasets[['BRCA']]
smpCode <- substr(colnames(BRCA),14,15)
grp1 <- which(smpCode=='01')
grp2 <- which(smpCode=='11')
object <- BRCA[1:min(66,nrow(BRCA)),c(grp1,grp2)]
group <- c(rep(1,length(grp1)),rep(2,length(grp1)))
perm.list <- vector('list',500)
for (i in seq_len(500)) {perm.list[[i]] <- sample(ncol(object))}
gsets <- split(rownames(object),rep(1:2,each=nrow(object)%/%2))
res <- gsnca_gsets(gsets,object,group,perm.list)

hui-sheen/MetaGSCA documentation built on April 9, 2022, 7:24 p.m.