Compute PCA tables from transformed expression

Selects the ntop genes with the highest variance after transformation and runs stats::prcomp() with samples as observations, centering on, and no scaling.

Usage

compute_pca(
  study,
  transform = c("vst", "log2-tpm"),
  ntop = 500L,
  components = 10L
)

Arguments

study: A rsemflow_study or path.
transform: vst or log2-tpm.
ntop: Number of most variable genes to use.
components: Maximum number of PCs to return.

Value

A list containing scores, loadings, and variance data frames, plus the transform and the number of genes used (ntop).

Examples

# \donttest{
ex <- system.file("extdata", "example", package = "rsemflow")
study <- read_rsem_study(file.path(ex, "data"), file.path(ex, "metadata.tsv"))
p <- compute_pca(study, ntop = 100, components = 3)
#> Warning: replacing previous import ‘S4Arrays::makeNindexFromArrayViewport’ by ‘DelayedArray::makeNindexFromArrayViewport’ when loading ‘SummarizedExperiment’
p$variance
#>   component   eigenvalue variance_fraction cumulative_variance
#> 1       PC1 1.949986e+02      6.454382e-01           0.6454382
#> 2       PC2 4.978460e+01      1.647852e-01           0.8102234
#> 3       PC3 2.768292e+01      9.162944e-02           0.9018528
#> 4       PC4 1.652027e+01      5.468149e-02           0.9565343
#> 5       PC5 1.313179e+01      4.346572e-02           1.0000000
#> 6       PC6 1.740680e-29      5.761586e-32           1.0000000
# }