drop unused levels in barplot

[cstrlln]

wanted to know what would be the simplest way of dropping unused levels in barplots, specially when splitting by different variables this gets tricky. Thank you!

[dtm2451]

Hmmmm that's a great question. I'm not sure if there is a good way... I think you've just highlighted a hole in the function's functionality that I'd wanna fix.

Perhaps I'd be able to transfer over the groupings.drop.unused functionality that I specifically added to dittoDotPlot() and dittoPlotVarsAcrossGroups() easily... I'll look into it and follow up here.

In the meantime, either 1) getting rid of retain.factor.levels but that punts your ordering =/, OR 2) making a copy of your metadata data.frame and subsetting externally to where you can manually adjust the factor levels while keeping retain.factor.levels = TRUE. Not great either way. For "2)" though, I'd recommend dittoViz's barPlot function cuz then you don't need to copy and subset the whole single-cell object in achieving this purpose, just the metadata. And then maybe something like this for trimming the levels before plotting:

trim_levels <- function(factor_data) {
    levs_all <- levels(factor_data)
    levs_used <- unique(as.character(factor_data))
    factor(
        factor_data,
        levels = levs_all[levs_all %in% levs_used]
    )
}

df_subset$<column> <- trim_levels(df_subset$<column>)