Normalise test perturbation scores by permutation results

normalise_by_permu(
  permutedScore,
  testScore,
  pAdj_method = "fdr",
  sortBy = c("gs_name", "sample", "pvalue")
)

Arguments

permutedScore: A list. Output of generate_permuted_scores
testScore: A data.frame. Output of `pathwayPertScore``
pAdj_method: Method for adjusting p-values for multiple comparisons. See ?p.adjust for methods available. Default to FDR.
sortBy: Sort the output by p-value, gene-set name or sample names.

Value

A data.frame

Details

Normalise the test perturbation scores generated by weight_ss_fc() through the permuted perturbation scores derived from the generate_permuted_scores() function. The mean absolute deviation(MAD) and median of perturbation scores for each pathway are firstly derived from the permuted perturbation scores. The test perturbation scores are then converted to robust z-scores using MADs and medians calculated.

Examples

if (FALSE) {
load(system.file("extdata", "gsTopology.rda", package = "sSNAPPY"))
data(metadata_example)
data(logCPM_example)
metadata_example <- dplyr::mutate(metadata_example, treatment = factor(
   treatment, levels = c("Vehicle", "E2+R5020", "R5020")))
ls <- weight_ss_fc(logCPM_example, metadata = metadata_example,
groupBy = "patient", sampleColumn = "sample", treatColumn = "treatment")

# compute raw gene-wise perturbation scores
genePertScore <- raw_gene_pert(ls$logFC, gsTopology)

# sum gene-wise perturbation scores to derive the pathway-level
# single-sample perturbation scores
pathwayPertScore <- pathway_pert(genePertScore)

# simulate the null distribution of scores through sample permutation
permutedScore <- generate_permuted_scores(logCPM_example, numOfTreat = 3,
 NB = 5, gsTopology = gsTopology, weight = ls$weight)

# normlise the test perturbation scores using the permutation results
normalisedScores <- normalise_by_permu(permutedScore, pathwayPertScore,
sortBy = "pvalue")
 }