Evaluating gene set rankings for the number of (significant) sets

These functions evaluate gene set rankings obtained from applying enrichment methods to multiple datasets. This allows to assess resulting rankings for granularity (how many gene sets have a unique p-value?) and statistical significance (how many gene sets have a p-value below a significance threshold?).

Usage

evalNrSigSets(ea.ranks, alpha = 0.05, padj = "none", perc = TRUE)

evalNrSets(ea.ranks, uniq.pval = TRUE, perc = TRUE)

Arguments

ea.ranks

Enrichment analysis rankings. A list with an entry for each enrichment method applied. Each entry is a list that stores for each dataset analyzed the resulting gene set ranking as obtained from applying the respective method to the respective dataset.

alpha

Statistical significance level. Defaults to 0.05.

padj

Character. Method for adjusting p-values to multiple testing. For available methods see the man page of the stats function p.adjust. Defaults to "none".

perc

Logical. Should the percentage or absolute number of gene sets be returned? Percentage is typically more useful for comparison between rankings with a potentially different total number of gene sets. Defaults to TRUE.

uniq.pval

Logical. Should the number of gene sets with a unique p-value or the total number of gene sets per ranking be returned? Defaults to TRUE.

Value

A list of numeric vectors storing for each method the number of (significant) gene sets for each dataset analyzed. If each element of the resulting list is of equal length (corresponds to successful application of each enrichment method to each dataset), the list is automatically simplified to a numeric matrix (rows = datasets, columns = methods).

See also

runEA to apply enrichment methods to multiple datasets; readResults to read saved rankings as an input for the eval-functions.

Author

Ludwig Geistlinger <Ludwig.Geistlinger@sph.cuny.edu>

Examples

    # simulated setup:
    # 2 methods & 2 datasets
    methods <- paste0("m", 1:2)
    data.ids <- paste0("d", 1:2)

    # simulate gene set rankings
    getRankingForDataset <- function(d)
    {
         r <- EnrichmentBrowser::makeExampleData("ea.res") 
         EnrichmentBrowser::gsRanking(r, signif.only=FALSE)
    }
  
    getRankingsForMethod <- function(m)
    {
         rs <- lapply(data.ids, getRankingForDataset)
         names(rs) <- data.ids
         rs
    }
        
    ea.ranks <- lapply(methods, getRankingsForMethod) 
    names(ea.ranks) <- methods 

    # evaluate
    evalNrSets(ea.ranks)
#>     m1  m2
#> d1  90  90
#> d2 100 100
    evalNrSigSets(ea.ranks)
#>    m1 m2
#> d1 50 50
#> d2 40 50