Resolve the scale of a fold-change column, cross-checking a user declaration and a column-naming convention against the observed data — resolve_foldchange_scale • shinyngs

Mirrors the log2 guessing done for assay matrices (cond_log2_transform_matrix) but for differential statistics fold changes, and errors rather than silently proceeding when the available signals disagree - see https://github.com/pinin4fjords/shinyngs/issues/125.

Usage

resolve_foldchange_scale(values, fc_column = NULL, declared_scale = "auto")

Arguments

values: Numeric vector of fold change values.
fc_column: Name of the column the values came from, used for the log2FoldChange-style naming heuristic. Can be NULL if unknown.
declared_scale: One of "auto" (default), "log2" or "linear".

Value

Either "log2" or "linear".

Examples

resolve_foldchange_scale(c(-2.1, 0.3, 1.8), fc_column = "log2FoldChange")
#> [1] "log2"
resolve_foldchange_scale(c(-8, 45, -120), fc_column = "FoldChange")
#> [1] "linear"