limma/voom Differential Expression Analysis — h_diff_expr

This helper functions performs the differential expression analysis with the voom method from the limma package (via limma::voom(), limma::lmFit() and limma::eBayes()) for given counts in a AnyHermesData object and a corresponding design matrix.

Usage

h_diff_expr_voom(object, design, ...)

Arguments

object: (AnyHermesData)
input.
design: (matrix)
design matrix.
...: additional arguments internally passed to limma::eBayes() (robust, trend, proportion, winsor.tail.p, stdev.coef.lim).

Value

A data frame with columns log2_fc (estimated log2 fold change), stat (moderated t-statistic), p_val (raw p-value), adj_p_pval (Benjamini-Hochberg adjusted p-value).

References

limma_packagehermes

voom_methodhermes

Examples

object <- hermes_data

# Create the design matrix corresponding to the factor of interest.
design <- model.matrix(~SEX, colData(object))

# Then perform the differential expression analysis.
result <- h_diff_expr_voom(object, design)
head(result)
#>                  log2_fc      stat        p_val adj_p_val
#> GeneID:8000   -2.3200712 -4.239244 0.0004209192 0.9118348
#> GeneID:51227  -1.0467295 -4.032550 0.0006788432 0.9118348
#> GeneID:344558  1.6896266  3.824046 0.0010993677 0.9118348
#> GeneID:51575  -0.7760844 -3.746293 0.0013155805 0.9118348
#> GeneID:151242 -3.4358998 -3.570103 0.0019741997 0.9118348
#> GeneID:8904   -0.6100354 -3.542864 0.0021017199 0.9118348

# Sometimes we might want to specify method details.
result2 <- h_diff_expr_voom(object, design, trend = TRUE, robust = TRUE)
head(result2)
#>                  log2_fc      stat        p_val adj_p_val
#> GeneID:8000   -2.3200712 -3.848050 0.0008857274 0.8995289
#> GeneID:51227  -1.0467295 -3.724654 0.0011933484 0.8995289
#> GeneID:344558  1.6896266  3.684191 0.0013155791 0.8995289
#> GeneID:51575  -0.7760844 -3.498308 0.0020551283 0.8995289
#> GeneID:151242 -3.4358998 -3.441687 0.0023525055 0.8995289
#> GeneID:123036  2.9002547  3.431263 0.0024329189 0.8995289