Perform gene set analysis on the result of a pre-computed test statistic. Test whether statistics in a gene set are larger/smaller than statistics not in the set.
Usage
zenithPR_gsa(
statistics,
ids,
geneSets,
use.ranks = FALSE,
n_genes_min = 10,
progressbar = TRUE,
inter.gene.cor = 0.01
)
Arguments
- statistics
pre-computed test statistics
- ids
name of gene for each entry in
statistics
- geneSets
GeneSetCollection
- use.ranks
do a rank-based test
TRUE
or a parametric testFALSE
? default: FALSE- n_genes_min
minumum number of genes in a geneset
- progressbar
if TRUE, show progress bar
- inter.gene.cor
correlation of test statistics with in gene set
Value
NGenes
: number of genes in this setCorrelation
: mean correlation between expression of genes in this setdelta
: difference in mean t-statistic for genes in this set compared to genes not in this setse
: standard error ofdelta
p.less
: p-value for hypothesis test ofH0: delta < 0
p.greater
: p-value for hypothesis test ofH0: delta > 0
PValue
: p-value for hypothesis testH0: delta != 0
Direction
: direction of effect based on sign(delta)FDR
: false discovery rate based on Benjamini-Hochberg method inp.adjust
Details
This is the same as zenith_gsa()
, but uses pre-computed test statistics. Note that zenithPR_gsa()
may give slightly different results for small samples sizes, if zenithPR_gsa()
is fed t-statistics instead of z-statistics.