getPvals: Estimate the empirical p-values

In DavisLaboratory/singscore: Rank-based single-sample gene set scoring method

Estimate the empirical p-values

Description

With null distributions estimated using the generateNull() function, p-values are estimated using a one-tailed test. A minimum p-value of 1/B can be achieved with B permutations.

Usage

getPvals(permuteResult, scoredf, subSamples = NULL)

Arguments

permuteResult	A matrix, null distributions for each sample generated using the generateNull() function
scoredf	A dataframe, the scored results of samples under test generated using the simpleScore() function
subSamples	A vector of sample labels/indices that will be used to subset the score matrix. All samples will be scored if not provided

Value

Estimated p-values for enrichment of the signature in each sample. A p-value of 1/B indicates that the estimated p-value is less than or equal to 1/B.

Examples

ranked <- rankGenes(toy_expr_se)
scoredf <- simpleScore(ranked, upSet = toy_gs_up, downSet = toy_gs_dn)
# find out what backends can be registered on your machine
BiocParallel::registered()
# the first one is the default backend, and it can be changed explicitly.
# See vignette for more details
permuteResult = generateNull(upSet = toy_gs_up, downSet = toy_gs_dn, ranked,
B =10, seed = 1, useBPPARAM = NULL)

# call the permutation function to generate the empirical scores
# for B times.
pvals <- getPvals(permuteResult,scoredf)

DavisLaboratory/singscore documentation built on Nov. 21, 2024, 8:17 a.m.